I've never even met you, but I wouldn't have my career if not for you. So I thought this would be a good time to say thanks.
It was the Macintosh computer you championed that first drew me into developing software. That business didn't make me rich, but it eventually got me hired by Apple. Unfortunately, you left a year before I got to Apple, but the company's goals were still the things you preached -- do something insanely great, change the world.
I spent ten years at the company you co-founded, and it was both a great education and a fun ride. Unfortunately, in a case of spectacularly poor judgment, I quit in early 1997, after the NeXT acquisition but before you took back control of the company. I didn't believe you'd take over, and I lost faith in the previous management. My only contact with you was a single meeting that you and I both attended. I was there as an observer, so I sat in the back and said nothing. My only impression of you was, "wow, he really doesn't wear socks."
Perhaps it's just as well that I quit.
Although I was no longer with Apple, you still played a huge role in my career. For a time in the late 1990s, it looked like Silicon Valley was becoming a backwater in technology. Software was dominated by Microsoft after its demolition of Netscape, AOL on the east coast was the online leader, and Dell in Texas plus the Asian companies were the leaders in PC hardware. The Valley's leadership role was saved, I believe, by Yahoo and Google in the web world, and by Apple's resurrection in computer systems.
Other people are doing a great job of recapping all of Apple's product successes since your return, so I won't bother repeating them here. But I want to talk about two other accomplishments that stand out to me. The first is how you've reset the way the tech industry looks at consumer products. Even a few years ago, most people still said that Microsoft's business model -- in which the hardware was designed separately from the software -- was the only viable way to make computing devices. Today, everyone talks about codeveloping hardware and software, and it's because of you.
The other accomplishment that stands out to me is your creation of an organization at Apple that could turn out hit after hit, reliably and with great quality. Most people don't appreciate how hard that is, mostly because Apple makes it look so easy.
It's because of the organization you built that I'm confident Apple will continue to do well, even as you reduce your role. I hope your health will improve, and it would be great to see you back as CEO some day. But that's speculation for another time.
Right now, I just wanted to say thanks, Steve. It was insanely great, and you did indeed change the world.
Wednesday, 24 August 2011
Sunday, 21 August 2011
I'm Speaking at Mobile 2.0
FYI, I'll be speaking on a panel at the Mobile 2.0 conference September 1, 2011 in San Francisco (link). The panel is about native apps vs. web, and should be a lot of fun, especially since Marc Davis is also on the panel. He's a great thinker and speaker.
The Mobile 2.0 folks have offered a discount to Mobile Opportunity readers. If you register using the code "TwentyFive" you'll get a 25% discount.
The Mobile 2.0 folks have offered a discount to Mobile Opportunity readers. If you register using the code "TwentyFive" you'll get a 25% discount.
Friday, 19 August 2011
The Part of Palm that Smartphone Companies Should be Bidding For
Anytime a CEO gets ousted in disgrace, his or her pet projects are vulnerable to a quick trip to the gallows if they falter. Mark Hurd was the CEO who bought Palm, so it was at risk from the moment Hurd left. HP's mobile device performance had not been good this year -- the Veer smartphone launched and vanished on the same day, and the TouchPad turned out to be a sales disaster. If Leo Apotheker had chosen to invest further in the business, it would have turned into his responsibility. It's far easier to just walk away.
You can make an argument that HP should have given the business more time, and it's a shame that we'll never get to see the Pre 3. But Palm's sales have been troubled for years, and I think its fundamental mistake was that it tried to be too much like Apple. From the start, Pre was aimed at the same users and the same usages as the iPhone (even down to a failed effort to tie the phone directly to iTunes). HP proved that most people don't want to buy an incremental improvement to the iPhone that can't run iOS apps.
Then just for kicks, HP went and proved the same point again with the TouchPad.
The lesson to other mobile companies, I think, is that unless you're a low-cost Asian vendor, you need to differentiate from Apple, not draft behind it.
I'd love to see Web OS live on, but the hardware debacle makes that less likely. As I mentioned the other day, licensees choose an OS because they think it'll generate a lot of unit sales for them. Since Web OS couldn't do that for HP, who else would want to license it?
If you believe that every smartphone company needs to own its own OS, we ought to see a mad bidding war between LG, HTC, Sony Ericsson, Dell, and maybe Samsung to buy Web OS. (The loser could get RIM as a consolation prize.) Maybe a buyout will still happen, but I think HP has probably been quietly shopping Web OS for a while, and if there were interest it would have tried to close a deal before today's announcement.
(By the way, HTC, if you do buy Web OS, you should insist that HP give you the Palm brand name as well. It's still far better known than the HTC brand in the US. The same logic applies for LG.)
But I'm not persuaded that buying an OS is the right way to go for any smartphone company. Turning yourself into a second-class imitation of Apple isn't a winning strategy, especially if your company doesn't know how to manage an operating system. (Case in point, look what it did to HP.) You can create great mobile systems without controlling the OS; all you need is a great system development team and the freedom to put a software layer on top of whatever OS you use.
That means the real crown jewel in the Web OS business unit is the system development people -- the product managers and engineers -- that HP just threw in the garbage. In my opinion, that's the part of Palm that smartphone companies should be fighting for.
You can make an argument that HP should have given the business more time, and it's a shame that we'll never get to see the Pre 3. But Palm's sales have been troubled for years, and I think its fundamental mistake was that it tried to be too much like Apple. From the start, Pre was aimed at the same users and the same usages as the iPhone (even down to a failed effort to tie the phone directly to iTunes). HP proved that most people don't want to buy an incremental improvement to the iPhone that can't run iOS apps.
Then just for kicks, HP went and proved the same point again with the TouchPad.
The lesson to other mobile companies, I think, is that unless you're a low-cost Asian vendor, you need to differentiate from Apple, not draft behind it.
I'd love to see Web OS live on, but the hardware debacle makes that less likely. As I mentioned the other day, licensees choose an OS because they think it'll generate a lot of unit sales for them. Since Web OS couldn't do that for HP, who else would want to license it?
If you believe that every smartphone company needs to own its own OS, we ought to see a mad bidding war between LG, HTC, Sony Ericsson, Dell, and maybe Samsung to buy Web OS. (The loser could get RIM as a consolation prize.) Maybe a buyout will still happen, but I think HP has probably been quietly shopping Web OS for a while, and if there were interest it would have tried to close a deal before today's announcement.
(By the way, HTC, if you do buy Web OS, you should insist that HP give you the Palm brand name as well. It's still far better known than the HTC brand in the US. The same logic applies for LG.)
But I'm not persuaded that buying an OS is the right way to go for any smartphone company. Turning yourself into a second-class imitation of Apple isn't a winning strategy, especially if your company doesn't know how to manage an operating system. (Case in point, look what it did to HP.) You can create great mobile systems without controlling the OS; all you need is a great system development team and the freedom to put a software layer on top of whatever OS you use.
That means the real crown jewel in the Web OS business unit is the system development people -- the product managers and engineers -- that HP just threw in the garbage. In my opinion, that's the part of Palm that smartphone companies should be fighting for.
Tuesday, 16 August 2011
Google and Motorola: What the #@!*%?
It's two days later and I'm still confused. When I saw the headline yesterday, my jaw literally dropped. "Google bought who? That's got to be a misprint. They must have bought a mobile operator, like Sprint or something. But Motorola? Really?"
Usually when a big tech merger happens you can see the logic behind it. Even if you don't agree with the logic, you understand why they made the deal. But in this case the more I think about it the more confused I get.
Did Google buy Motorola for the patents? If so, why isn't it spinning out the hardware business? Or did Google buy Motorola because it wants to be in the hardware business? If so, does it understand what a world of other problems that will create for Android and the rest of Google? Seriously, if Google tries to integrate Motorola into its business we could end up citing this as the deal that permanently broke Google.
Why roll the dice like that? Maybe I'm missing something, maybe Google has a screw loose, maybe both of the above. Or maybe I'm wrong to look for airtight logic. Companies sometimes make decisions on impulse, especially when they are under stress, and it's a sure thing that Google is under stress these days on IP issues.
So I have a lot more questions than answers. My questions are about Google's intent, its next steps, and how other companies will react...
Why did Google do it, really? The conventional answer is that Google wanted Motorola Mobility for its patents. That's what Google itself implied, and Marguerite Reardon over at CNET agreed (link). That might well be the explanation. Om Malik had a really intriguing take: Google bought Motorola as a defensive move to prevent Microsoft from getting the Motorola patents (link). And Richard Windsor of Nomura, who I respect deeply, said in an e-mail that this is all about the patents. He predicts that Google's new patent portfolio will create a balance of power enabling Google to quickly force a settlement to the patent lawsuits against its licensees.
But if you wanted only the patents, I think you'd buy Motorola, keep the patents and then spin out the hardware company to avoid antagonizing your licensees. Google says it intends to keep Motorola and run it.
Besides, as Andrew Sorkin pointed out in the New York Times, Google could have bought a different but also important mobile patent portfolio from InterDigital for about $10 billion less than Motorola (link). Maybe there's some magic patent at Motorola that Google feels is worth $10 billion more, or maybe there are some terms in Motorola's patent cross-license agreements that Google desperately needs. But again, if that's the case, why not keep the patents and resell the hardware business?
Unless Google is lying about keeping Motorola intact, I think Google intends to be in the mobile hardware business. Which raises the next question...
Does Google know how to run a hardware business? No, of course not. The processes, disciplines, and skills are utterly different. The same business practices that made Google good in software will be a liability in hardware. Google's engineers-first, research driven product management philosophy is effective in the development of web software, because you can run experiments and revise your web app every day in response to user feedback. But in hardware, you have to make feature decisions 18 months before you ship, and you have to live with those decisions for another 18 months while your product sells through. You can't afford to wait for science. Instead, you need dictatorial product managers who operate on artistry and intuition. All of those concepts (dictatorship, artistry, intuition) are anathema to Google's culture. Either Google's worldview will dominate and ruin Motorola, or worse yet the Motorola worldview will infect Google. Google with Motorola inside it is like a python that swallowed a minivan.
To put it another way, I think Google has about as much chance of successfully managing a device business as Nokia had of running an OS business.
But the real question is, does Google realize that it doesn't know how to make hardware? I doubt it. Speaking as someone who worked at PalmSource for its whole independent history, an OS company always believes that it could do a better job of making hardware than its licensees. It's incredibly frustrating to have a vision for what people should do with your software, and then see them screw it up over and over. The temptation is to build some hardware yourself, just to show those idiots how to do it right.
I think maybe Google just gave in to that temptation.
But if Google really wants to sell hardware, that raises questions for the other Android licensees...
How will Google really manage Motorola? Google says it's going to treat Motorola as an independent company without any special access to the Android team. But what's the point in that? Motorola hasn't exactly been dominating the mobile device world lately, so I find it very, very hard to believe that Google would buy it and leave it intact. Wouldn't you want to have Motorola create special products that take advantage of the latest Android features? Kind of like a flagship operation? Then when you announce a new initiative at Google IO, you can have some nice new Motorola hardware ready to ship with it on day one. Of course, the other Android licensees will be allowed to participate too. They're welcome to run flat out to keep up with every Google software initiative, disregarding expense and business risk, just like Google's Motorola subsidiary will.
Which makes you wonder...
How will the Android licensees react? I think we can safely disregard the positive quotes from the other Android licensees. What would you do if your company depended utterly on Android, and Google called you up twelve hours before the announcement and asked for a quote? Would you risk Google's anger by refusing to give a nice quote? Of course not.
But would you honestly be happy? Of course not. In the last year, you gained share at the expense of Motorola. Now instead of being a weak and failing vendor you can snack on, Motorola has infinite financial resources and cannot physically go broke. Sure, I am happy to compete with that.
The other issue is the one everyone else has already pointed out -- even though Google says there will be a firewall between Motorola and Android, you suspect it'll be semi-permeable, meaning you'll always be at a bit of a disadvantage.
So what do you do? A lot of people are predicting that Android could be in danger of losing licensees. For example, Horace Dediu at Asymco drew a parallel to the Symbian consortium, whose members were uncomfortable because Nokia held the largest share of the ownership (link). But when Symbian was launched, those companies were happy to sign up, despite the asymmetric ownership, because they thought Symbian was going to dominate the mobile OS market, and they were scared of Microsoft. They dropped out only after it was proven conclusively that only Nokia was capable of making a Symbian phone that sold well in Europe.
I can tell you from personal experience at Palm that licensees don't care about governance issues when they think your OS will help them sell a lot of units. It's only after growth slows down that they get twitchy. As long as Android continues to grow explosively, the licensees will be right there with it because they're terrified not to be.
Google probably knows the licensees can't go anywhere. In fact, it has a history of treating them very roughly in private (check out the nasty tone in the private memos between Google and Samsung exposed by the Skyhook lawsuit here). So in some ways the Motorola deal is just more of the same.
But there is still a risk to Google. Android licensees will probably be more willing to talk to Microsoft now, and they might do a few more Windows Phone products, if only to get leverage against Google. So Google has just thrown a lifeline to Windows Phone, which otherwise might have been headed for extinction if the first round of Nokia products failed.
This might also be an opportunity for other mobile platforms. If there were any...
Is there a third path? The Android licensees are probably pretty wary of both Google and Microsoft at this point, and may be wishing forlornly that there was a third alternative for mobile operating systems.
Unfortunately, I don't think there is. The handset vendors' embrace of "royalty-free" Android strangled the other Linux mobile platforms. TrollTech was bought by Nokia and then killed, while Access's evolution of Palm OS died for lack of customers.
There's speculation that HP might broadly license Web OS (link). But HP has its own hardware conflict of interest (a much stronger one than either Google or Microsoft). Far more importantly, keep in mind that mobile phone companies license an OS because they believe it's going to sell millions of units for them. If HP, with all of its resources and channel presence and strong brand, can't sell significant numbers of Web OS phones, why would HTC or Samsung believe they could do it?
[Edit: In the original version of this post, I failed to mention MeeGo. A couple of people have told me that was unfair, and I think they are right. Based on past experience, I have a lot of skepticism about OS consortia, especially ones involving Intel. But if MeeGo's ever going to get serious consideration from hardware companies, now is the time, and I should have acknowledged that.]
Hint to Android licensees: If you build up HTML 5 as a platform, you won't have to depend on anyone else's platform. But in the meantime, your realistic choices are Android and Microsoft.
Speaking of Microsoft...
What will Microsoft do now? Steve Ballmer faces a very interesting decision. Windows Phone just got a boost because it's now seen as a more vendor-neutral platform than Android. The door is probably open for Microsoft to build deeper relationships with Android licensees. If Microsoft sill believes in its licensing model, it will focus on walking through that door.
But as others have pointed out, Microsoft's position is now a bit lonely in some ways. The other major smartphone platforms (iOS and Android) now have captive hardware arms. Even RIM has both hardware and OS, although it's been a while since RIM was held up as a model for others to emulate. Will Microsoft feel exposed without its own hardware business? And if it does feel exposed, will it buy Nokia?
I'd be very surprised if it did. Buying Nokia would decisively end the Windows Mobile licensing business. You'd be betting Microsoft's mobile future even more completely on the ability of Nokia to execute in hardware. Besides, why buy the cow when you're already milking it?
I'd also like to think that Microsoft learned from the Zune debacle that it's not great at creating mobile hardware.
And then there's the fruit company...
What will Apple do? Apple's history since Steve returned is that it doesn't react to competitors; it forces competitors to react to it. Apple is brilliant at setting the terms of the competition so other companies are forced to compete on Apple's turf. Everyone else is focused on building licensed commodity hardware, so Apple creates integrated systems. Everyone else has optimized their supply chains to sell through third party retailers, so Apple creates its own stores. Everyone else stopped making touchscreen smartphones, so what does Apple make?
You get the picture. So I don't expect Apple to make any changes in response to the Motorola deal, but I would be shocked if Apple didn't have plans for changing the terms of the competition again now that Google is trying to build more integrated hardware and software. There are all sorts of game-changing moves Apple could make -- do a much larger push in web services, create an iPhone Nano (fewer features and lower price), even create its own search engine or social network (potentially valuable just to make Google crazy).
What's next?
To sum it all up, it's impossible to predict what will happen. Hopefully the new balance of power in patents will make the big lawsuits go away, although I doubt we'd see a resolution before the deal closes, and that could take many months. If Google bought Motorola for the patents, it'll either sell the company or let it gracefully rot, and we'll go back to business as usual.
On the other hand, if Google tries to integrate Motorola into its business, that's a noble mission, and I hope they'll succeed because the mobile industry needs more competition to Apple in systems design. I dearly hope Google will take the challenge seriously and recognize that it'll need to make fundamental changes to its culture. But those changes would be daunting even for a company experienced in mergers, and Google's never done a deal this big before. I think the most likely outcome of the Google-Motorola merger is some flavor of train wreck.
I hope I'm wrong.
Usually when a big tech merger happens you can see the logic behind it. Even if you don't agree with the logic, you understand why they made the deal. But in this case the more I think about it the more confused I get.
Did Google buy Motorola for the patents? If so, why isn't it spinning out the hardware business? Or did Google buy Motorola because it wants to be in the hardware business? If so, does it understand what a world of other problems that will create for Android and the rest of Google? Seriously, if Google tries to integrate Motorola into its business we could end up citing this as the deal that permanently broke Google.
Why roll the dice like that? Maybe I'm missing something, maybe Google has a screw loose, maybe both of the above. Or maybe I'm wrong to look for airtight logic. Companies sometimes make decisions on impulse, especially when they are under stress, and it's a sure thing that Google is under stress these days on IP issues.
So I have a lot more questions than answers. My questions are about Google's intent, its next steps, and how other companies will react...
Why did Google do it, really? The conventional answer is that Google wanted Motorola Mobility for its patents. That's what Google itself implied, and Marguerite Reardon over at CNET agreed (link). That might well be the explanation. Om Malik had a really intriguing take: Google bought Motorola as a defensive move to prevent Microsoft from getting the Motorola patents (link). And Richard Windsor of Nomura, who I respect deeply, said in an e-mail that this is all about the patents. He predicts that Google's new patent portfolio will create a balance of power enabling Google to quickly force a settlement to the patent lawsuits against its licensees.
But if you wanted only the patents, I think you'd buy Motorola, keep the patents and then spin out the hardware company to avoid antagonizing your licensees. Google says it intends to keep Motorola and run it.
Besides, as Andrew Sorkin pointed out in the New York Times, Google could have bought a different but also important mobile patent portfolio from InterDigital for about $10 billion less than Motorola (link). Maybe there's some magic patent at Motorola that Google feels is worth $10 billion more, or maybe there are some terms in Motorola's patent cross-license agreements that Google desperately needs. But again, if that's the case, why not keep the patents and resell the hardware business?
Unless Google is lying about keeping Motorola intact, I think Google intends to be in the mobile hardware business. Which raises the next question...
Does Google know how to run a hardware business? No, of course not. The processes, disciplines, and skills are utterly different. The same business practices that made Google good in software will be a liability in hardware. Google's engineers-first, research driven product management philosophy is effective in the development of web software, because you can run experiments and revise your web app every day in response to user feedback. But in hardware, you have to make feature decisions 18 months before you ship, and you have to live with those decisions for another 18 months while your product sells through. You can't afford to wait for science. Instead, you need dictatorial product managers who operate on artistry and intuition. All of those concepts (dictatorship, artistry, intuition) are anathema to Google's culture. Either Google's worldview will dominate and ruin Motorola, or worse yet the Motorola worldview will infect Google. Google with Motorola inside it is like a python that swallowed a minivan.
To put it another way, I think Google has about as much chance of successfully managing a device business as Nokia had of running an OS business.
But the real question is, does Google realize that it doesn't know how to make hardware? I doubt it. Speaking as someone who worked at PalmSource for its whole independent history, an OS company always believes that it could do a better job of making hardware than its licensees. It's incredibly frustrating to have a vision for what people should do with your software, and then see them screw it up over and over. The temptation is to build some hardware yourself, just to show those idiots how to do it right.
I think maybe Google just gave in to that temptation.
But if Google really wants to sell hardware, that raises questions for the other Android licensees...
How will Google really manage Motorola? Google says it's going to treat Motorola as an independent company without any special access to the Android team. But what's the point in that? Motorola hasn't exactly been dominating the mobile device world lately, so I find it very, very hard to believe that Google would buy it and leave it intact. Wouldn't you want to have Motorola create special products that take advantage of the latest Android features? Kind of like a flagship operation? Then when you announce a new initiative at Google IO, you can have some nice new Motorola hardware ready to ship with it on day one. Of course, the other Android licensees will be allowed to participate too. They're welcome to run flat out to keep up with every Google software initiative, disregarding expense and business risk, just like Google's Motorola subsidiary will.
Which makes you wonder...
How will the Android licensees react? I think we can safely disregard the positive quotes from the other Android licensees. What would you do if your company depended utterly on Android, and Google called you up twelve hours before the announcement and asked for a quote? Would you risk Google's anger by refusing to give a nice quote? Of course not.
But would you honestly be happy? Of course not. In the last year, you gained share at the expense of Motorola. Now instead of being a weak and failing vendor you can snack on, Motorola has infinite financial resources and cannot physically go broke. Sure, I am happy to compete with that.
The other issue is the one everyone else has already pointed out -- even though Google says there will be a firewall between Motorola and Android, you suspect it'll be semi-permeable, meaning you'll always be at a bit of a disadvantage.
So what do you do? A lot of people are predicting that Android could be in danger of losing licensees. For example, Horace Dediu at Asymco drew a parallel to the Symbian consortium, whose members were uncomfortable because Nokia held the largest share of the ownership (link). But when Symbian was launched, those companies were happy to sign up, despite the asymmetric ownership, because they thought Symbian was going to dominate the mobile OS market, and they were scared of Microsoft. They dropped out only after it was proven conclusively that only Nokia was capable of making a Symbian phone that sold well in Europe.
I can tell you from personal experience at Palm that licensees don't care about governance issues when they think your OS will help them sell a lot of units. It's only after growth slows down that they get twitchy. As long as Android continues to grow explosively, the licensees will be right there with it because they're terrified not to be.
Google probably knows the licensees can't go anywhere. In fact, it has a history of treating them very roughly in private (check out the nasty tone in the private memos between Google and Samsung exposed by the Skyhook lawsuit here). So in some ways the Motorola deal is just more of the same.
But there is still a risk to Google. Android licensees will probably be more willing to talk to Microsoft now, and they might do a few more Windows Phone products, if only to get leverage against Google. So Google has just thrown a lifeline to Windows Phone, which otherwise might have been headed for extinction if the first round of Nokia products failed.
This might also be an opportunity for other mobile platforms. If there were any...
Is there a third path? The Android licensees are probably pretty wary of both Google and Microsoft at this point, and may be wishing forlornly that there was a third alternative for mobile operating systems.
Unfortunately, I don't think there is. The handset vendors' embrace of "royalty-free" Android strangled the other Linux mobile platforms. TrollTech was bought by Nokia and then killed, while Access's evolution of Palm OS died for lack of customers.
There's speculation that HP might broadly license Web OS (link). But HP has its own hardware conflict of interest (a much stronger one than either Google or Microsoft). Far more importantly, keep in mind that mobile phone companies license an OS because they believe it's going to sell millions of units for them. If HP, with all of its resources and channel presence and strong brand, can't sell significant numbers of Web OS phones, why would HTC or Samsung believe they could do it?
[Edit: In the original version of this post, I failed to mention MeeGo. A couple of people have told me that was unfair, and I think they are right. Based on past experience, I have a lot of skepticism about OS consortia, especially ones involving Intel. But if MeeGo's ever going to get serious consideration from hardware companies, now is the time, and I should have acknowledged that.]
Hint to Android licensees: If you build up HTML 5 as a platform, you won't have to depend on anyone else's platform. But in the meantime, your realistic choices are Android and Microsoft.
Speaking of Microsoft...
What will Microsoft do now? Steve Ballmer faces a very interesting decision. Windows Phone just got a boost because it's now seen as a more vendor-neutral platform than Android. The door is probably open for Microsoft to build deeper relationships with Android licensees. If Microsoft sill believes in its licensing model, it will focus on walking through that door.
But as others have pointed out, Microsoft's position is now a bit lonely in some ways. The other major smartphone platforms (iOS and Android) now have captive hardware arms. Even RIM has both hardware and OS, although it's been a while since RIM was held up as a model for others to emulate. Will Microsoft feel exposed without its own hardware business? And if it does feel exposed, will it buy Nokia?
I'd be very surprised if it did. Buying Nokia would decisively end the Windows Mobile licensing business. You'd be betting Microsoft's mobile future even more completely on the ability of Nokia to execute in hardware. Besides, why buy the cow when you're already milking it?
I'd also like to think that Microsoft learned from the Zune debacle that it's not great at creating mobile hardware.
And then there's the fruit company...
What will Apple do? Apple's history since Steve returned is that it doesn't react to competitors; it forces competitors to react to it. Apple is brilliant at setting the terms of the competition so other companies are forced to compete on Apple's turf. Everyone else is focused on building licensed commodity hardware, so Apple creates integrated systems. Everyone else has optimized their supply chains to sell through third party retailers, so Apple creates its own stores. Everyone else stopped making touchscreen smartphones, so what does Apple make?
You get the picture. So I don't expect Apple to make any changes in response to the Motorola deal, but I would be shocked if Apple didn't have plans for changing the terms of the competition again now that Google is trying to build more integrated hardware and software. There are all sorts of game-changing moves Apple could make -- do a much larger push in web services, create an iPhone Nano (fewer features and lower price), even create its own search engine or social network (potentially valuable just to make Google crazy).
What's next?
To sum it all up, it's impossible to predict what will happen. Hopefully the new balance of power in patents will make the big lawsuits go away, although I doubt we'd see a resolution before the deal closes, and that could take many months. If Google bought Motorola for the patents, it'll either sell the company or let it gracefully rot, and we'll go back to business as usual.
On the other hand, if Google tries to integrate Motorola into its business, that's a noble mission, and I hope they'll succeed because the mobile industry needs more competition to Apple in systems design. I dearly hope Google will take the challenge seriously and recognize that it'll need to make fundamental changes to its culture. But those changes would be daunting even for a company experienced in mergers, and Google's never done a deal this big before. I think the most likely outcome of the Google-Motorola merger is some flavor of train wreck.
I hope I'm wrong.
Wednesday, 10 August 2011
The Case for Software Patents
It's become popular lately to call for the elimination of software patents. Tim Lee at Forbes sounded the call last month (link), and this week Mark Cuban joined the chorus (link):
I have a ton of respect for many of the people arguing against software patents, but I disagree strongly with their arguments. I think software patents play an important role in encouraging innovation, especially by small companies. The loss of them would make it harder for small companies to survive, and would discourage fundamental innovation in software.
The online debate about software patents is very contentious, and much of it focuses on philosophical issues like the nature of software and whether that's inherently patentable. The debate also often gets mixed with the contention that all software should be free. I'm not going to get into either topic; the arguments are arcane, sometimes quasi-religious in their fervor, and besides they've already been debated to death online.
What I want to focus on is the broader issue of the economic role of patents and how that applies to software. The patent system is designed to encourage innovation by giving a creator a temporary monopoly on the use of an invention. Does that mechanism work in software? What are the problems? And what's the best way to fix them?
When you take that perspective, I think it's clear that there are some genuine problems with software patents. (Actually, there are problems with patents in general, and software is just the most prominent example.) But I think there are better ways than a ban to solve those problems. To me, banning software patents to solve patent problems would be like banning automobiles to stop car theft. The cure is far worse than the disease.
The value of software patents
Let me start with a personal example. As I've mentioned before, I'm working on a startup. When we brief people on what we're doing, one of the first questions we get is, "how will you prevent [Google / Apple / Microsoft / insert hot web company here] from copying you?"
A big part of the answer is, "we've filed for a patent."
A patent isn't magic protection, of course. It might not be granted, and even if it's granted, patents are difficult to enforce against a really big company. But it reassures investors, and more importantly if a Facebook or Google wanted to copy our work, the patent makes it safer and quicker for them to buy our company rather than just ripping us off. So it helps to protect the value of our company.
Without the patent, I think it could be open season on us the moment we announce our product.
Advocates of eliminating software patents say there are other ways to protect software companies. The first is that software can be copyrighted. That's technically true, but the only thing copyright protects you from is word for word theft of your source code; it does not protect inventions. For most software innovations, copyright is no protection at all.
The second argument is that small companies should move quickly, so the big companies can't catch up with them. The idea is that if you move fast enough, you won't need patents. I think there's a consumer web app bias in that advice -- it works best for small apps that can be adopted quickly, or that have a strong social effect (so the user base is part of your competitive protection). It doesn't work well for software tools that have a slower adoption curve. The more complex and powerful the software, the slower the adoption cycle. This is especially true for enterprise tools. Without patents, those companies are exquisitely vulnerable to being ripped off soon after they launch, when they're just starting to gain word of mouth.
So for companies creating new categories of software, and especially for enterprise software tools, I think patents remain the best (and really only) protection from theft.
But what about the damage being caused by abuse of the patent system? If we keep software patents, are we then endorsing those abuses?
I don't think so. In the articles I've seen, there are two primary arguments for eliminating software patents: Trolls and patent warfare. They need to be discussed separately.
Discouraging patent trolls. The troll problem is something we all know about: patent licensing companies buy large collections of patents and then extort fees from companies that had no idea they were violating the patents. That's the core of the problem Mark Cuban was talking about above, and it is outrageous. These surprise lawsuits can have a devastating effect on smaller companies that can't afford to hire a lawyer to defend themselves. Even if you're in the right, it can be so expensive to defend yourself that you just have to give up and pay the license fee. That definitely has a chilling effect on innovation, it is contrary to the intent of the patent system, and therefore it needs to be restrained.
But I think the answer in that case is not to eliminate software patents; it's to restrict the right of "non-practicing entities" (patent trolls) to sue for patent infringement. That would still have a financial effect on small companies -- in the case of my startup, it would make it harder for us to sell our patent if we wanted to. But patent law exists to protect the process of innovation, not to protect inventors for their own sake. If you can't put your patent to good use, you aren't contributing to the public good and you shouldn't get the same level of protection as a company that has built a business around a patent.
Patent warfare is a very different issue. Several large tech companies are using patent lawsuits to slow down competitors and pull revenue out of them. Eliminating software patents would not stop these wars; they're also based on hardware patents, antitrust law, and any other field of law that the companies can apply. It's like one of those cartoon fights in a kitchen where a character opens up a drawer and throws everything inside it.
Yeah, like that.
The underlying problem here isn't about patents; it's about the use (and abuse) of the legal system as a competitive tool. I've had more involvement in tech industry legal wars than I want to think about: I gave depositions in the Apple-Microsoft IP wars, and I testified in Washington in the Microsoft antitrust lawsuit. The overall experience left me plenty cynical about the legal system, but it also persuaded me that big tech companies are perfectly capable of taking care of themselves in court. They do not need our help. You should think of lawsuits as just another way that tech companies express love for one-another.
I'm somewhat sympathetic to the Android vendors being sued by Apple, but a lot of it is their own fault. HTC in particular has no one to blame but itself for its situation, in my opinion. HTC was one of the first companies in mobile computing, creating PDAs for Compaq and early smartphones for Orange and O2. I've got to believe that if HTC had been thinking clearly about patents, there are a lot of fundamental mobile inventions it could have patented. Then it would have had a big enough patent portfolio to force a cross-licensing deal with Apple.
The same thing goes for Google. When it decided to enter the mobile OS business, it should have expected that it would end up at war with Apple and Microsoft (heck, anyone could have predicted that). Google should have bought up a big mobile patent portfolio (like maybe Palm's) back when they were inexpensive.
Are we obligated to change the patent laws just because Google and HTC were careless? No. Is it in the public interest for us to intervene anyway? I doubt it. Here's how the mobile patent wars will play out: The big boys will do a whole bunch more legal maneuvering, they'll scream bloody murder, and in the end one of them will write a check to the other. Then they'll all go back to work.
My advice: If it bothers you, stop reading the news stories about it. Or sit back and enjoy it as theatre. It's hardly an important enough issue to justify stripping the patent protection from every small software company in the US.
A world without software patents
If you want to understand the importance of software patents, go back and talk to the first people who patented software. That's what I did. Two years ago, I corresponded with Martin Goetz, holder of the first software patent (link).
Goetz was a manager at Applied Data Research, one of the first independent app companies in the 1960s. ADR made applications for mainframes, and IBM copied and gave away a version of ADR's Autoflow application (the first commercially marketed third party software app). ADR might have been wiped out, but it had patented Autoflow, and it was able to successfully sue IBM. That lawsuit, plus a related one by the US government, laid the foundations of the independent software industry by forcing IBM to stop giving away free apps for its mainframes.
The lawsuits involved a lot of legal issues, including antitrust, so you can't say that software patents alone led to the birth of the software industry. But I think it's clear that patents helped codify the value of software independent from hardware. If that value hadn't been recognized, the antitrust suit would have been meaningless because there would have been no damages.
So antitrust and patent law have worked together to help protect software innovation. Antitrust helped to restrain big companies from giving away free competitors to an app (although that protection has eroded lately), while patents restrained big companies from copying apps directly. It's like a ladder. If you pull out either leg, I worry that it won't stand.
In his online memoirs (link), Goetz makes the case that application innovation was slow and unresponsive to users in the decade before software patents, and accelerated dramatically in the decade after. I agree. That is exactly the sort of innovation that patent law was meant to encourage, and so I view software patents as a success.
Without software patents, I think it would be far too easy to go back to the bad old days when the big computing companies walked all over small software companies, the software industry consisted of only consultants and custom developers, and software innovation moved at a much slower pace.
_____
More reading: Software entrepreneur and investor Paul Graham wrote a nuanced and detailed take on the subject here. Some of his conclusions differ a bit from mine, but the essay is well worth reading.
"Because of software and process patents any company could be sued for almost anything. It is impossible to know what the next patent to be issued will be and whether or not your company will be at complete risk. It is impossible to go through the entire catalog of patents issued over the last 10, 15, 20 years and determine which will be used to initiate a suit against your company."
I have a ton of respect for many of the people arguing against software patents, but I disagree strongly with their arguments. I think software patents play an important role in encouraging innovation, especially by small companies. The loss of them would make it harder for small companies to survive, and would discourage fundamental innovation in software.
The online debate about software patents is very contentious, and much of it focuses on philosophical issues like the nature of software and whether that's inherently patentable. The debate also often gets mixed with the contention that all software should be free. I'm not going to get into either topic; the arguments are arcane, sometimes quasi-religious in their fervor, and besides they've already been debated to death online.
What I want to focus on is the broader issue of the economic role of patents and how that applies to software. The patent system is designed to encourage innovation by giving a creator a temporary monopoly on the use of an invention. Does that mechanism work in software? What are the problems? And what's the best way to fix them?
When you take that perspective, I think it's clear that there are some genuine problems with software patents. (Actually, there are problems with patents in general, and software is just the most prominent example.) But I think there are better ways than a ban to solve those problems. To me, banning software patents to solve patent problems would be like banning automobiles to stop car theft. The cure is far worse than the disease.
The value of software patents
Let me start with a personal example. As I've mentioned before, I'm working on a startup. When we brief people on what we're doing, one of the first questions we get is, "how will you prevent [Google / Apple / Microsoft / insert hot web company here] from copying you?"
A big part of the answer is, "we've filed for a patent."
A patent isn't magic protection, of course. It might not be granted, and even if it's granted, patents are difficult to enforce against a really big company. But it reassures investors, and more importantly if a Facebook or Google wanted to copy our work, the patent makes it safer and quicker for them to buy our company rather than just ripping us off. So it helps to protect the value of our company.
Without the patent, I think it could be open season on us the moment we announce our product.
Advocates of eliminating software patents say there are other ways to protect software companies. The first is that software can be copyrighted. That's technically true, but the only thing copyright protects you from is word for word theft of your source code; it does not protect inventions. For most software innovations, copyright is no protection at all.
The second argument is that small companies should move quickly, so the big companies can't catch up with them. The idea is that if you move fast enough, you won't need patents. I think there's a consumer web app bias in that advice -- it works best for small apps that can be adopted quickly, or that have a strong social effect (so the user base is part of your competitive protection). It doesn't work well for software tools that have a slower adoption curve. The more complex and powerful the software, the slower the adoption cycle. This is especially true for enterprise tools. Without patents, those companies are exquisitely vulnerable to being ripped off soon after they launch, when they're just starting to gain word of mouth.
So for companies creating new categories of software, and especially for enterprise software tools, I think patents remain the best (and really only) protection from theft.
But what about the damage being caused by abuse of the patent system? If we keep software patents, are we then endorsing those abuses?
I don't think so. In the articles I've seen, there are two primary arguments for eliminating software patents: Trolls and patent warfare. They need to be discussed separately.
Discouraging patent trolls. The troll problem is something we all know about: patent licensing companies buy large collections of patents and then extort fees from companies that had no idea they were violating the patents. That's the core of the problem Mark Cuban was talking about above, and it is outrageous. These surprise lawsuits can have a devastating effect on smaller companies that can't afford to hire a lawyer to defend themselves. Even if you're in the right, it can be so expensive to defend yourself that you just have to give up and pay the license fee. That definitely has a chilling effect on innovation, it is contrary to the intent of the patent system, and therefore it needs to be restrained.
But I think the answer in that case is not to eliminate software patents; it's to restrict the right of "non-practicing entities" (patent trolls) to sue for patent infringement. That would still have a financial effect on small companies -- in the case of my startup, it would make it harder for us to sell our patent if we wanted to. But patent law exists to protect the process of innovation, not to protect inventors for their own sake. If you can't put your patent to good use, you aren't contributing to the public good and you shouldn't get the same level of protection as a company that has built a business around a patent.
Patent warfare is a very different issue. Several large tech companies are using patent lawsuits to slow down competitors and pull revenue out of them. Eliminating software patents would not stop these wars; they're also based on hardware patents, antitrust law, and any other field of law that the companies can apply. It's like one of those cartoon fights in a kitchen where a character opens up a drawer and throws everything inside it.
Yeah, like that.
The underlying problem here isn't about patents; it's about the use (and abuse) of the legal system as a competitive tool. I've had more involvement in tech industry legal wars than I want to think about: I gave depositions in the Apple-Microsoft IP wars, and I testified in Washington in the Microsoft antitrust lawsuit. The overall experience left me plenty cynical about the legal system, but it also persuaded me that big tech companies are perfectly capable of taking care of themselves in court. They do not need our help. You should think of lawsuits as just another way that tech companies express love for one-another.
I'm somewhat sympathetic to the Android vendors being sued by Apple, but a lot of it is their own fault. HTC in particular has no one to blame but itself for its situation, in my opinion. HTC was one of the first companies in mobile computing, creating PDAs for Compaq and early smartphones for Orange and O2. I've got to believe that if HTC had been thinking clearly about patents, there are a lot of fundamental mobile inventions it could have patented. Then it would have had a big enough patent portfolio to force a cross-licensing deal with Apple.
The same thing goes for Google. When it decided to enter the mobile OS business, it should have expected that it would end up at war with Apple and Microsoft (heck, anyone could have predicted that). Google should have bought up a big mobile patent portfolio (like maybe Palm's) back when they were inexpensive.
Are we obligated to change the patent laws just because Google and HTC were careless? No. Is it in the public interest for us to intervene anyway? I doubt it. Here's how the mobile patent wars will play out: The big boys will do a whole bunch more legal maneuvering, they'll scream bloody murder, and in the end one of them will write a check to the other. Then they'll all go back to work.
My advice: If it bothers you, stop reading the news stories about it. Or sit back and enjoy it as theatre. It's hardly an important enough issue to justify stripping the patent protection from every small software company in the US.
A world without software patents
If you want to understand the importance of software patents, go back and talk to the first people who patented software. That's what I did. Two years ago, I corresponded with Martin Goetz, holder of the first software patent (link).
Goetz was a manager at Applied Data Research, one of the first independent app companies in the 1960s. ADR made applications for mainframes, and IBM copied and gave away a version of ADR's Autoflow application (the first commercially marketed third party software app). ADR might have been wiped out, but it had patented Autoflow, and it was able to successfully sue IBM. That lawsuit, plus a related one by the US government, laid the foundations of the independent software industry by forcing IBM to stop giving away free apps for its mainframes.
The lawsuits involved a lot of legal issues, including antitrust, so you can't say that software patents alone led to the birth of the software industry. But I think it's clear that patents helped codify the value of software independent from hardware. If that value hadn't been recognized, the antitrust suit would have been meaningless because there would have been no damages.
So antitrust and patent law have worked together to help protect software innovation. Antitrust helped to restrain big companies from giving away free competitors to an app (although that protection has eroded lately), while patents restrained big companies from copying apps directly. It's like a ladder. If you pull out either leg, I worry that it won't stand.
In his online memoirs (link), Goetz makes the case that application innovation was slow and unresponsive to users in the decade before software patents, and accelerated dramatically in the decade after. I agree. That is exactly the sort of innovation that patent law was meant to encourage, and so I view software patents as a success.
Without software patents, I think it would be far too easy to go back to the bad old days when the big computing companies walked all over small software companies, the software industry consisted of only consultants and custom developers, and software innovation moved at a much slower pace.
_____
More reading: Software entrepreneur and investor Paul Graham wrote a nuanced and detailed take on the subject here. Some of his conclusions differ a bit from mine, but the essay is well worth reading.
Wednesday, 3 August 2011
Quick Takes: Logitech's Misstep, Nokia's New Names and Fonts
Short thoughts on recent tech news...
Logitech strays from the path
I haven't written much about Logitech here, but they've long been one of my favorite tech companies because they have a history of breaking all the rules of Silicon Valley:
--The Valley says you can't make money in commodity hardware, but Logitech makes good profits from the most commoditized bits of the computer industry.
--The Valley says you especially can't make money in low-end consumer hardware, but that's where Logitech thrives.
--The Valley says that if you want to make money in hardware, you need to be based in a low-cost part of the world like China. But Logitech is Swiss-owned and headquartered in Silicon Valley, two of the highest-cost places to do business in the world.
Logitech succeeded by picking well-established markets like keyboards and webcams where its skills in design, user experience, and added-value features let it charge a bit more than the commodity players. As the folks at Logitech will tell you, "we're chefs, not farmers." In other words, we don't create new markets, we come to existing markets and do an especially nice job of rearranging the ingredients.
But Logitech's financial performance hasn't been great for the last two years, and the company's most recent quarter was a loss. The disappointing earnings report stood out to me because Logitech put much of the blame on its Google TV product, which apparently had apocalyptic negative sales last quarter (more returns from retailers than shipments) (link). Apparently somebody at Google convinced Logitech to do some farming, and the company is paying dearly for it.
I'm usually an advocate of companies taking risks and pioneering new markets, and I think that's something Logitech has the skills to do if it's careful. But in this case, it chose a terrible target: the uncertainties of a new market, but with a user experience constrained by Google's software. So Logitech took on the risks of farming without the ability to fully apply its own strengths. Logitech was more or less a passenger on Google's boat.
The lesson, obviously: Be very careful when you step away from your proven strategy. Oh, and never trust Google to develop a new market for you.
Good luck naming your phones, Nokia (again)
Nokia announced that it has changed the naming system for its phones, going back from letters to numbers (link). The last time Nokia changed its naming, five years ago, I wrote a piece saying why technology naming schemes eventually break down after about six years. This morning I thought about updating the post, but actually I think it's still valid as-is, except that naming schemes now apparently last only five years. Here's a link to the article, so you can judge the rest of it for yourself (link).
Font games
Speaking of Nokia, there are few things more embarrassing for a company than announcing to the world that you have chosen a new font for your corporate communication. Making font and logo changes is often interpreted as rearranging deck chairs on the Titanic -- "we can't fix our fundamental problems but by God at least we'll fix the font!" That's unfair, of course; Apple changed its logo in the middle of its rebirth, ditching the six colors, and it turned out just fine. So I didn't blame Nokia when it announced on March 25 that it was moving to a new font (link).
But I'm surprised that more than four months later, Nokia's old font is still splattered all over its web pages. You can find the new one in some spots, but a lot of website navigation, headlines, and even new product announcements are in the old font. Here are some examples (click on the images for a larger view):
Nokia Conversations (official company blog), August 2, 2011. Note the use of the old font to announce the company's newest product. That means the use of the old font isn't limited to legacy bits of the website that haven't been converted yet.
Nokia Press website, August 2, 2011. New and old logos together on the same page. That's got to annoy the corporate branding folks.
Nokia UK home page, August 2, 2011. New font used for the headline, but the old font is used elsewhere.
Nokia US home page, August 2, 2011. Old font almost everywhere. Well, you kind of expect Nokia's US site to be behind the times.
Nokia Finland home page, August 2, 2011. The greatest use of the new font -- except when they mention Ovi. Freudian slip?
I feel childish and petty for picking on Nokia about a little detail like this. Fonts stand out to me because I used to run a font company, but I know most people don't even notice them. Nokia has much bigger problems to solve.
However, four months is more than enough time to switch over to a new font, especially in newly-created headlines. Since the criticisms about Nokia's smartphones often center on inattention to detail and slow execution, you'd think they would want to execute crisply wherever they can. In its March announcement, Nokia wrote that the new font is important because "the letters flow into each other somewhat, creating the impression of forward movement." Since the old font lingers, does that create the impression that Nokia's not moving forward?
Logitech strays from the path
I haven't written much about Logitech here, but they've long been one of my favorite tech companies because they have a history of breaking all the rules of Silicon Valley:
--The Valley says you can't make money in commodity hardware, but Logitech makes good profits from the most commoditized bits of the computer industry.
--The Valley says you especially can't make money in low-end consumer hardware, but that's where Logitech thrives.
--The Valley says that if you want to make money in hardware, you need to be based in a low-cost part of the world like China. But Logitech is Swiss-owned and headquartered in Silicon Valley, two of the highest-cost places to do business in the world.
Logitech succeeded by picking well-established markets like keyboards and webcams where its skills in design, user experience, and added-value features let it charge a bit more than the commodity players. As the folks at Logitech will tell you, "we're chefs, not farmers." In other words, we don't create new markets, we come to existing markets and do an especially nice job of rearranging the ingredients.
But Logitech's financial performance hasn't been great for the last two years, and the company's most recent quarter was a loss. The disappointing earnings report stood out to me because Logitech put much of the blame on its Google TV product, which apparently had apocalyptic negative sales last quarter (more returns from retailers than shipments) (link). Apparently somebody at Google convinced Logitech to do some farming, and the company is paying dearly for it.
I'm usually an advocate of companies taking risks and pioneering new markets, and I think that's something Logitech has the skills to do if it's careful. But in this case, it chose a terrible target: the uncertainties of a new market, but with a user experience constrained by Google's software. So Logitech took on the risks of farming without the ability to fully apply its own strengths. Logitech was more or less a passenger on Google's boat.
The lesson, obviously: Be very careful when you step away from your proven strategy. Oh, and never trust Google to develop a new market for you.
Good luck naming your phones, Nokia (again)
Nokia announced that it has changed the naming system for its phones, going back from letters to numbers (link). The last time Nokia changed its naming, five years ago, I wrote a piece saying why technology naming schemes eventually break down after about six years. This morning I thought about updating the post, but actually I think it's still valid as-is, except that naming schemes now apparently last only five years. Here's a link to the article, so you can judge the rest of it for yourself (link).
Font games
Speaking of Nokia, there are few things more embarrassing for a company than announcing to the world that you have chosen a new font for your corporate communication. Making font and logo changes is often interpreted as rearranging deck chairs on the Titanic -- "we can't fix our fundamental problems but by God at least we'll fix the font!" That's unfair, of course; Apple changed its logo in the middle of its rebirth, ditching the six colors, and it turned out just fine. So I didn't blame Nokia when it announced on March 25 that it was moving to a new font (link).
But I'm surprised that more than four months later, Nokia's old font is still splattered all over its web pages. You can find the new one in some spots, but a lot of website navigation, headlines, and even new product announcements are in the old font. Here are some examples (click on the images for a larger view):
Nokia Conversations (official company blog), August 2, 2011. Note the use of the old font to announce the company's newest product. That means the use of the old font isn't limited to legacy bits of the website that haven't been converted yet.
Nokia Press website, August 2, 2011. New and old logos together on the same page. That's got to annoy the corporate branding folks.
Nokia UK home page, August 2, 2011. New font used for the headline, but the old font is used elsewhere.
Nokia US home page, August 2, 2011. Old font almost everywhere. Well, you kind of expect Nokia's US site to be behind the times.
Nokia Finland home page, August 2, 2011. The greatest use of the new font -- except when they mention Ovi. Freudian slip?
I feel childish and petty for picking on Nokia about a little detail like this. Fonts stand out to me because I used to run a font company, but I know most people don't even notice them. Nokia has much bigger problems to solve.
However, four months is more than enough time to switch over to a new font, especially in newly-created headlines. Since the criticisms about Nokia's smartphones often center on inattention to detail and slow execution, you'd think they would want to execute crisply wherever they can. In its March announcement, Nokia wrote that the new font is important because "the letters flow into each other somewhat, creating the impression of forward movement." Since the old font lingers, does that create the impression that Nokia's not moving forward?
Thursday, 23 June 2011
How to Shape the Mobile Data Market
(Part 3 of "Who Will Pay for Mobile Data?")
There's a big nasty dilemma hidden at the heart of mobile computing: No one knows how we'll pay for all that mobile data we're supposed to use in the next few years. The question doesn't get much publicity, but it drives some of the most intense debates in mobile, including net neutrality and the wireless bandwidth "crisis."
This is the conclusion of a three-part series on the issue. In Part 1 (link), I talked about the tech industry's unlimited vision for the growth of mobile data, and why I think it won't come true because we'll run out of people willing to pay for data service
In Part 2 (link), I discussed the alternate scenario, in which everyone is willing to pay for mobile data and adoption of it continues to accelerate. In this case, the mobile operators will need to invest urgently in increased capacity, and even with that investment we'll eventually run out of wireless bandwidth.
The two scenarios leave mobile operators trapped between the need to expand their networks and the fear that they won't be able to pay for the expansion. So the operators are trying to get other parties to help pay for the network. I believe that's the real driver behind the net neutrality debate and the rhetoric about a wireless bandwidth "crisis." Ultimately, government regulators will decide who will pay and how the mobile data network is structured, which will have a huge effect on which companies win and what we can do with the network.
In this part I'll give my take on what we should do about the situation, and I'll talk about the opportunities all of this change creates for operators, handset companies, and developers.
The look of mobile data in the future
If you only took away two messages from the first two posts in this series, these are the ones I'd want you to remember:
1. The only thing we can predict for sure about the future of mobile data is that it's unpredictable. Maybe I'm right that it'll saturate soon; maybe Cisco's right that it'll go on growing explosively for years; maybe we'll average out to something in the middle. The variables in play are so numerous, and so complicated, that absolutely no one can predict for sure what will happen.
In that sort of uncertain situation, I think our top priority should be to keep the mobile market as flexible as possible, so it can respond quickly and efficiently to whatever the customers decide to do. That means we should ensure that market signals -- things like pricing and customer demand -- are as clear and unambiguous as possible, so we'll all know what the real level of demand is, and we can all respond to the same base of information. The word "transparency" gets overused these days, but goodness gracious we need as much transparency as possible in mobile data.
2. We should plan wired and wireless data together. We need to deal with the reality of the mobile network and market, not what we might want it to be. And the reality is that we're not creating a separate wireless data network, we're creating a single integrated wired and wireless network. A lot of the political rhetoric about mobile data talks about a completely cellular data future as some sort of public goal. It's more like a public fantasy. Every forecast I've seen from the wireless operators requires that they be able to offload a lot of traffic to the wired network. Forget about wireless replacing wired; what we need to do is make sure they both work together well, with each focusing on what they do best. That means wired is used whenever possible because in most cases it's cheaper and higher capacity, while wireless fills in the gaps.
We should set up a level playing field between wired and wireless so the market can sort out which traffic should go where. Artificial political goals for the penetration of wireless, or favoring one network technology over another, are incredibly dangerous because they may lock in a market structure that turns out to be unaffordable. In fact, because the market is so unpredictable, those sorts of goals are almost certain to be wrong.
So I get queasy when the US Federal Communications Commission, and even big companies like Google, argue that wireless data should have different regulations than wired data. I think that increases the risk that we'll accidently bias the overall network in the wrong direction.
What we should do
As I've said before, I am not a big fan of government regulation in business, because it's usually inefficient and slow. However, there are some situations in which you can't get the government out of the market, and I think cellular wireless is one of those cases because the public ultimately owns the airwaves in most countries.
So if we're going to have government regulation, let's do it right.
The grand bargain. The operators are asking for some mammoth benefits. In the US, some of the biggest operators want to merge. Okay, let's let them do it. I don't think TMobile US is large enough to be viable in the long term anyway, so we need to merge it with either AT&T or Sprint. If TMobile joins AT&T, which is the current proposal, the next merger in the US will be Verizon-Sprint; I think we have to accept that as well, for the same reason.
The operators in the US and Europe want more spectrum allocated to them. Again, I'd go ahead with it. In the US, the television networks aren't using the extra spectrum, so it ought to go somewhere useful.
But in return, we should demand serious changes in the cellular data market. I'm not talking about tweaks at the edges, I mean permanent changes in the rules of the game, designed to ensure lasting competition and a more flexible market that responds better to customer needs.
Here's what I propose:
Stop whining about the wireless "crisis"
The first step is to change our rhetoric. The bandwidth "crisis" is the tech industry's equivalent of the War on Terror: it's based on a genuine problem, it can never be completely solved, and it can be used to justify many actions that people might not otherwise consider.
The idea of a wireless crisis is an incredibly convenient tool for motivating government regulators. Elected officials assume they are responsible for solving a wireless spectrum crisis, since they allocate wireless spectrum. If it were called a "Verizon and AT&T don't want to pay for a bunch more cell towers crisis," I don't think President Obama would propose spending $50 billion on it.
This isn't just a US issue. Anything one government does in mobile data is played back in other countries as a justification for equivalent actions there. On a recent trip to Australia, I was surprised to hear a radio commentator complaining at length about the government's plan to supply broadband service to many Australians through landlines rather than wireless. You can make a good argument for using landlines, since (as we discussed in part 2) they can carry a lot more data than wireless. But the commentator was upset that Australia was failing to do "what Barack Obama is doing in the United States."
It's reasonable to ask what's so wrong with a little crisis hype and international competition. After all, governments move far too slowly in most cases, so if a bit of alarming rhetoric makes them respond faster, isn't that a good thing? The trouble is that we'll all have to live with the results after the "crisis" is "solved." In that world, no matter how much spectrum we allocate to wireless data, service will continue to have slowdowns, outages and service gaps, especially in the United States, because it's more profitable for the operators to run their networks right at the edge of overload (in this sense they have the same financial incentives as airlines).
We're lying when we tell people that the whole wireless data network could collapse. Although service problems are a certainty, there is virtually zero risk of a full network collapse, unless the operators cause it themselves by underpricing data plans and selling more smartphones than they can support. And we're misleading people when we say that prices will go up unless we allocate more spectrum. Prices will eventually go up no matter how much spectrum we allocate to data, because demand for cellular data is growing faster than supply.
By overstating the risks and talking about the "crisis" as a temporary, fixable thing, we create an unrealistic public expectation for the quality and price of cellular data in the future. That may well be advantageous for a couple of quarters or even a year, but in the long run it will erode public trust when we don't deliver the benefits we promised. The wireless operators, especially AT&T in the US, already have big image problems. Overpromising will make the problems worse. To the extent that government agencies, and mobile tech companies like Apple and Google, participate in the crisis rhetoric, they risk their credibility as well.
We need to ask ourselves as an industry if we want to have the same sort of public image in five years as the airlines have today. If not, we should be honest with people now. For example, I think there is a convincing, legitimate case for reallocating old TV spectrum for data services. Without it, mobile data prices will go up faster, and a lot of the features many of us want from mobile data may not be affordable. But we should also be honest with people that cellular bandwidth overload is a chronic disease rather than a crisis, the network is not going to collapse unless we're incompetent, cellular service will not be as fast or cheap per bit as a wired, and cellular data will generally be a supplement to our wired broadband, not a replacement.
Make the cellular data market transparent
The problem with the cellular data market as it's structured today is that it often hides from users the real cost of the network they use, so they can't make well informed choices, and it's hard for us to tell which buying patterns are genuine and which ones have been created artificially. For example, the cost of your smart phone is subsidized, so you don't realize what an expensive piece of hardware you're carrying in your pocket. You're told that you have unlimited data, but actually if you use it too much your operator will probably reduce your data speed without telling you.
By making cellular data seem cheaper than it is, we encourage people to use the network more, increasing the very overload that we're supposed to be fixing. Some of the proposals for the future of mobile data would further increase the overuse of cellular data by making it seem even cheaper to users.
The structure of the mobile market also limits competition among mobile operators (especially in the US), and reduces competition between mobile phone manufacturers.
I think this systematic distortion of the market must stop. If people could see the real cost of cellular data, they would make better-informed decisions about when and how to use it, and we wouldn't need secret back-end controls on traffic. Meanwhile, more competition in services and phones would mean faster innovation, more consumer choice, and more efficient prices.
Here are some specific steps I think we should take:
1. Ban covert traffic limits. Today some wireless operators (and some wired ones as well) are quietly reducing the quality of service they deliver to some users, without telling them. This is done through various techniques including "traffic shaping" (prioritizing or delaying certain types of data packets) and "throttling" (reducing the throughput of the network, or the speed of certain transactions). In effect it usually means reducing the connection speed of people or apps that use the network the most. For example, Dean Bubley recently wrote about an ISP who consistently reduced data throughput at particular times of the day (link).
There are some types of traffic management that make sense. E-mail spam can be reduced through throttling that limits the number of e-mails that can be sent by a single account per second. Throttling can also be used to limit malware attacks, by reducing the ability of a rogue app to flood the network with traffic. And I think it's fine to enforce the speed you paid for in your Internet connection. For instance, if you've paid for a 10 MBPS connection and the operator limits your throughput to 10 MBPS, I do not have a problem with that.
But in some cases the operators are limiting network performance to covertly restrict users, either by interfering with certain types of traffic, or by limiting the speeds of some users without telling them. For example, the current Verizon Wireless terms of service give them the right to reduce the throughput in your "unlimited" data plan if you're in the top 5% of data users (link). They can do this without notifying you.
This sort of hidden restriction is damaging to the market because people may sign up for a wireless plan believing they will get more service than they actually will. They can't make a fully informed decision between wired and wireless service because they don't know how much wireless data they're really going to get. This may misallocate resources and make the wireless network even more overloaded than it would be otherwise.
The answer to this is simple: Require operators to notify a customer when they have throttled or shaped his or her service (other than enforcing the promised speed of the connection). I am not against throttling in general, but it should not be done without notification. A text message would be fine. The Internet speedometer, which I discuss below, will also help with this problem.
2. Require a data gas gauge and speedometer in smartphones. Can you imagine buying a car that didn't have a gas gauge and speedometer? That's essentially what we do today with smartphones. For most smartphone users today, there is no easy way to tell how much data throughput you're getting from the network, and how close you are to any limits on your data usage. Some operators bundle apps to do this, some have more arcane ways to check, and some send you a text if you get close to the limit. But I think it's fair to say that most people are in the dark about their usage until they get their monthly bill, and if they do go over a limit they will have trouble figuring out why.
This is an easy problem to fix. We should require that every smartphone have an app, accessible at the same level as the Settings app, that tells the user how close he or she is to hitting any data caps in the service plan (for example, if you are a Verizon user, how close are you to getting throttled?). The app should also show how much data you're using at any particular time, so you can see how much throughput the network is really giving you.
We also should modify the signal strength bars to change color depending on how much data you're consuming at any moment. This would show you when you're using a website or app that uses huge chunks of data. When customers see that video or Flash makes their signal bars turn red, they'll be much more cautious about using those sites on the wireless network.
3. Decouple the phone purchase from the network. Currently in the US and much of Europe, if you sign a contract for a data plan, you get a discount of several hundred dollars on a new phone purchased at the same time. But you have to buy the phone through the mobile operator, giving them huge control over the selection and features of the phones they sell. Basically, users are not free to pick the phones they want; they have to take the phones their operator chooses to sell.
This operator lock-in is subject to all sorts of backroom manipulation. Weak phone vendors are forced to comply with a huge list of tests and requirements, while for stronger vendors the rules are often waived. I've also been told privately by some operators that they deliberately discriminate against some handset vendors because they just don't like them.
The handset vendors aren't completely clean either. A vendor with a hot handset may restrict its availability to a single operator in order to extract concessions from them. Can you say iPhone?
It's a wonder that some operator or handset company hasn't been sued already for restraint of trade. With the amount of operator shelf space shrinking in the US due to mergers, I think it's only a matter of time before there's a legal detonation.
In addition to the legal risk, these restrictions have the effect of restricting customer choice and competition, so they are bad for transparency. It's time to open up the handset market. To make that happen, subsidies should be separated from the purchase of a particular phone. When someone signs up for a plan, they should get a voucher for a discount on any phone. The voucher can be used at that time to buy a phone in the operator's store, or it can be used later to buy a phone in any other store.
This would encourage more selection and competition in mobile phones. It would create more direct competition between operator service plans. And it would put the wireless and wired networks on an even footing (can you imagine a wired data provider limiting the brands of PC that you can use with your cable data connection?).
In the US, I think we should consider one other step to open up the handset market. In most of Europe, and many other parts of the world, there is a vigorous retail market in mobile phones sold separately from an operator. Because everyone is on the same network standard, and because all the phones use SIM cards, it is easy to buy a new phone at retail and pop your card into it. You do lose the subsidy, but virtually all customers know they can at least switch phones if they really want to. This leads to a much larger selection of phones, and to higher competition between operators because it's easier to choose separately the phone and service plan you want.
The US market is much less open. Most mobile phones are sold only through operator stores, and it can be very hard to switch from one operator to another because they have different network technologies, and some of them don't even use SIM cards. Because it's so hard to switch phones, I think most US mobile users are barely even aware of what a SIM card is, and how to find it in their phone (most of them would probably confuse it with the SD card).
To open up the handset market, the US should require that all mobile phones use SIM cards, and that they be switchable between the major operator networks. That way someone could go into a consumer electronics store, buy the phone they want, and use it with any network. This will have to be phased in over time, but we're already moving toward it anyway. Verizon and AT&T are both moving to LTE, and there are very strong rumors that Sprint will do so as well. So some day we'll have one standard cellular technology base in the US. In the meantime, we'll have to buy dual-mode phones that use both LTE and either GSM or CDMA, depending on which operator you use. But the chipsets for smartphones are increasingly capable of handling several different networks, so they can switch between LTE, GSM and CDMA. I think it would be reasonable to require that future smartphones sold in the US be SIM-based and capable of operating on all three standards. I think the real question is how quickly we could phase in that requirement; if you have thoughts on that please post a comment.
4. Enable toll-free apps and websites. As I discussed in Part 1, we need the data equivalent of a toll-free phone call, in which a website or mobile app company would pay for the data traffic generated by a particular app or site. This requires changes to the operators' billing infrastructure, but I think it will be essential for enabling the growth of mobile data. It should be an extremely high priority for the operators, and it's in the interest of web and app companies to get together with the operators to define standards for these charges, so they'll be easy for developers to work with. I suspect there's an important role government regulators can play in helping to encourage these negotiations.
5. Do not allow the operators, or the web companies, to discriminate against one-another. I agonized over this one a lot. The operators would like to be able to charge web companies extra if they want reliable delivery of data (for example, in a time-sensitive app like video streaming), or if they want a guarantee of a certain level of throughput. I understand why they want to do this, because it would help pay for their infrastructure, and I do not think it is inherently evil. But I think it would cause too much collateral damage to the mobile market. In fact, I think it would put us on a road toward wrecking mobile data.
The first problem is that hidden back-end charges like this are essentially an invisible subsidy for cellular data. A user won't know the real cost of the data he or she is using, and this could end up increasing traffic on the cellular network artificially, contributing to data overload.
There are also big practical problems with implementing charges for quality of service. As Dean Bubley has pointed out repeatedly (link), there are huge drawbacks to this sort of approach. To give one example, there is no way to guarantee quality of service when you don't know how overloaded a particular cell site will be. If one high-priority video session comes in, does the operator shut down five other "regular" data sessions to make way for the high-priority one? In that case, the "regular" customers are not getting the service they paid for, and they won't even know it. They'll just think something is wrong with the web app they're using.
I agree with Dean that there's no way to make a system like this work predictably and fairly. Better to just charge users for the data they consume, let them know how much that costs, and allow them to adjust their own usage patterns.
The other reason we should ban quality of service fees is because in some cases they could produce in a destructive power struggle between operators and websites, with users caught in the middle. US cable television is a nightmare example of what not to do.
In cable television, it's common for network operators and content companies (the cable channels) to pay each other for services. For example, Home Shopping Network reportedly pays cable TV companies to be included in your service package, because they know they'll make more money if they're seen in more homes. They are, effectively, subsidizing your cable television service.
On the other hand, many of the most popular channels charge the cable companies a fee for the privilege of carrying them. For example, ESPN (the leading US sports network) reportedly charges cable companies about $4 per month per household; other popular channels are in the 5-20 cent per month range.
The same sorts of things could happen in the mobile web if the operators could charge websites for service. For instance, what if Facebook started offering video streaming as part of its services? If the mobile operators tried to charge Facebook for its network usage, what is to stop Facebook from turning around and demanding a fee from the operators for allowing them to carry Facebook?
Unless we're very careful, we could end up with a situation in mobile similar to the one in cable TV, where users get caught in disputes between the network operators and the content creators. Some of those arguments in the US have been incredibly ugly, with users tied into long-term contracts for cable service but unable to access the channels they thought they paid for. And remember, in cable we get these messes even though we have only have about a hundred channels to negotiate. On the web, you have literally millions of them.
The operators should not kid themselves that they would win in this sort of showdown. If Facebook cut off its traffic to Sprint's servers, what would happen? Would users abandon Facebook because it's not on the Sprint network -- or would they switch off of Sprint because it doesn't have Facebook? I think we all know the answer to that: there would be crowds holding pitchforks and torches outside the Sprint stores. The websites have far stronger brands and far more user loyalty than the operators. So it's unlikely that the operators will really be able to coerce money out of the most successful websites.
In practice, I think the operators would be able to get fees only from small startups that don't have brand awareness with users. That becomes a barrier to entry for those companies, which historically have been the source of most online innovation. To give a real-world example of what that could do to the web, look again at cable television programming: A small number of networks dominate the selection of channels, resulting in slow innovation and reduced choice. There is very low turnover in these channels.
If the web worked like cable TV does, we'd all still be using AOL for e-mail.
I've talked with people at small startup cable channels, and they are incredibly bitter about the barriers they face getting placement on cable systems. They're actually counting on the web to let them bypass the cable operators.
I think the only way to make the mobile market work efficiently is to make the payment mechanisms as clear and visible as possible. Make users pay for the data they use, and allow web and app companies to make their sites and apps toll-free if they want to, but don't start creating hidden layers of fees and subsidies. That will just distort the market and expose operators to retaliation. My operator friends, this is a war you cannot win -- so don't start the battle.
To formalize this settlement, government regulators should ban both operators discriminating against websites or types of traffic, and websites withholding their content from a particular operator or network.
6. Encourage open WiFi. As I mentioned above, we're not creating a standalone cellular network, we're creating an integrated wired and wireless network. WiFi has a critical role to play in that network, and we should make it even more central. Here's a question for you: How often have you tried to find an available WiFi network, and seen no networks at all in range? I can't speak for other countries, but it almost never happens to me in any populated part of the US. But how many times have you tried to sign onto WiFi and found only locked access points? That happens to me all the time.
We already have a very dense, well-populated wireless front end to the data network in most places that matter, but we can't use it fully because most of the access points are locked down.
There are good reasons for the lockdown. If you leave your WiFi router open, it can be hacked (actually, it can also be hacked if you keep it locked, but that's a topic for a different post). Also, in the US if someone downloads child pornography or does something else illegal on the Internet, the law often goes after the router owner because that's the only person they can find. You can read some horror stories here.
But getting those connections opened up would have huge benefits for the public, because it would take some of the pressure off cellular wireless. Rather than telling people to close off their connections, we should be encouraging them to leave them open. Regulators could help this in a couple of ways:
--First, we should require that the next generation of WiFi routers have a pass-through feature enabling public access to the Internet without giving access to the user's home network. Traffic from the user's private connection should have priority over the public one, and if public usage is excessive the user should be able to throttle it.
--Second, the law should be changed to protect people whose open wireless connections are abused without their permission.
Opportunities
So that's how I think the future of mobile data will look: unpredictable growth, always skating the line between overloaded and overpriced, and with a huge variety of users, almost all of them with some sort of limits on their data service, and many with budget plans that encourage very careful use of data. For the health of everyone involved in the market, I hope we'll also get regulations that make the market more transparent, and more open to new players.
It's a different mobile data world than many analysts have been predicting, but that's not necessarily a bad thing. Often the best business opportunities happen when conditions change unpredictably. I think this is one of those times. So I'd like to conclude by recapping the big opportunities as I see them...
For handset vendors, I think the most interesting new opportunity will be the smartphone designed for people with limited data budgets. How do you entice people into gradually using more data? This is an opportunity to do a fundamental rethinking of the smartphone user experience. Since different people will probably respond to different data features, I think it will also be an opportunity for smartphone vendors to stake out their own market segments, helping to insulate them from the intense commodity price pressure we're likely to see in generic smartphones as the market fills up.
Try to think like an automobile vendor in 1950. Do you want to compete with everyone else in midsize sedans, or would you like to dominate a smaller segment like station wagons or sports cars?
To target a segment, you'll need to hire people who know how to design integrated hardware-software systems rather than just devices, and you'll need to learn to partner closely with app and web companies as peers (rather than the serf-overlord relationships you're used to having).
In the last couple of days I've been contacted privately by some people who predict even more revolutionary moves by the handset companies, most notably the idea of selling a phone at retail bundled with airtime that you've bought from an operator. In other words, the phone comes with its own network service. That's what Amazon did with Kindle, and there's nothing in principle to prevent a handset company from doing the same thing.
I think there would be a lot of implementation challenges, most notably keeping access to that third party network if it starts to run out of capacity. But it would be intriguing to see what someone like Apple would do with this.
For operators, I think it's important to pick your battles. Although covert traffic-shaping and charging websites for service is very seductive, in the long term that will lead you into intense conflicts that you're not likely to win. It would also create more incentives for the handset companies to set up their own virtual networks, which really would transform your networks into dumb pipes.
I think it's better to focus on new business models that are a win for both you and your business partners. The most appealing of these to me is toll-free data. That would be intriguing to a lot of web and mobile app companies, allowing you to build cooperative alliances with them. And it's a whole new revenue stream that might become very large over time.
For web and app developers, the emerging segmentation of mobile data makes the idea of "enticement" even more important than it is today. How do you give people some software for free and then entice them into paying for add-ons or other apps? Already most of the mobile app developers I talk to are thinking along those lines, and obviously that business model is very well established on the web. But as smartphones reach down to more price-sensitive people who are less enthusiastic about data, there will be intense demand for apps and websites that can entice them into starting to pay for bits of mobile data.
These "data on-ramp" apps are not always intuitively obvious, and will probably differ by country (for example, mobile horoscopes were a major driver of beginning data use in parts of Asia). The companies that can find the on-ramps will be incredibly valuable to investors, handset companies, and operators.
What do you think?
That's my take on the situation. What do you agree and disagree with? What else would you add to the picture? How does it differ in your country? And most importantly, what do you think the opportunities are? Please post a comment and share your ideas.
There's a big nasty dilemma hidden at the heart of mobile computing: No one knows how we'll pay for all that mobile data we're supposed to use in the next few years. The question doesn't get much publicity, but it drives some of the most intense debates in mobile, including net neutrality and the wireless bandwidth "crisis."
This is the conclusion of a three-part series on the issue. In Part 1 (link), I talked about the tech industry's unlimited vision for the growth of mobile data, and why I think it won't come true because we'll run out of people willing to pay for data service
In Part 2 (link), I discussed the alternate scenario, in which everyone is willing to pay for mobile data and adoption of it continues to accelerate. In this case, the mobile operators will need to invest urgently in increased capacity, and even with that investment we'll eventually run out of wireless bandwidth.
The two scenarios leave mobile operators trapped between the need to expand their networks and the fear that they won't be able to pay for the expansion. So the operators are trying to get other parties to help pay for the network. I believe that's the real driver behind the net neutrality debate and the rhetoric about a wireless bandwidth "crisis." Ultimately, government regulators will decide who will pay and how the mobile data network is structured, which will have a huge effect on which companies win and what we can do with the network.
In this part I'll give my take on what we should do about the situation, and I'll talk about the opportunities all of this change creates for operators, handset companies, and developers.
The look of mobile data in the future
If you only took away two messages from the first two posts in this series, these are the ones I'd want you to remember:
1. The only thing we can predict for sure about the future of mobile data is that it's unpredictable. Maybe I'm right that it'll saturate soon; maybe Cisco's right that it'll go on growing explosively for years; maybe we'll average out to something in the middle. The variables in play are so numerous, and so complicated, that absolutely no one can predict for sure what will happen.
In that sort of uncertain situation, I think our top priority should be to keep the mobile market as flexible as possible, so it can respond quickly and efficiently to whatever the customers decide to do. That means we should ensure that market signals -- things like pricing and customer demand -- are as clear and unambiguous as possible, so we'll all know what the real level of demand is, and we can all respond to the same base of information. The word "transparency" gets overused these days, but goodness gracious we need as much transparency as possible in mobile data.
2. We should plan wired and wireless data together. We need to deal with the reality of the mobile network and market, not what we might want it to be. And the reality is that we're not creating a separate wireless data network, we're creating a single integrated wired and wireless network. A lot of the political rhetoric about mobile data talks about a completely cellular data future as some sort of public goal. It's more like a public fantasy. Every forecast I've seen from the wireless operators requires that they be able to offload a lot of traffic to the wired network. Forget about wireless replacing wired; what we need to do is make sure they both work together well, with each focusing on what they do best. That means wired is used whenever possible because in most cases it's cheaper and higher capacity, while wireless fills in the gaps.
We should set up a level playing field between wired and wireless so the market can sort out which traffic should go where. Artificial political goals for the penetration of wireless, or favoring one network technology over another, are incredibly dangerous because they may lock in a market structure that turns out to be unaffordable. In fact, because the market is so unpredictable, those sorts of goals are almost certain to be wrong.
So I get queasy when the US Federal Communications Commission, and even big companies like Google, argue that wireless data should have different regulations than wired data. I think that increases the risk that we'll accidently bias the overall network in the wrong direction.
What we should do
As I've said before, I am not a big fan of government regulation in business, because it's usually inefficient and slow. However, there are some situations in which you can't get the government out of the market, and I think cellular wireless is one of those cases because the public ultimately owns the airwaves in most countries.
So if we're going to have government regulation, let's do it right.
The grand bargain. The operators are asking for some mammoth benefits. In the US, some of the biggest operators want to merge. Okay, let's let them do it. I don't think TMobile US is large enough to be viable in the long term anyway, so we need to merge it with either AT&T or Sprint. If TMobile joins AT&T, which is the current proposal, the next merger in the US will be Verizon-Sprint; I think we have to accept that as well, for the same reason.
The operators in the US and Europe want more spectrum allocated to them. Again, I'd go ahead with it. In the US, the television networks aren't using the extra spectrum, so it ought to go somewhere useful.
But in return, we should demand serious changes in the cellular data market. I'm not talking about tweaks at the edges, I mean permanent changes in the rules of the game, designed to ensure lasting competition and a more flexible market that responds better to customer needs.
Here's what I propose:
Stop whining about the wireless "crisis"
The first step is to change our rhetoric. The bandwidth "crisis" is the tech industry's equivalent of the War on Terror: it's based on a genuine problem, it can never be completely solved, and it can be used to justify many actions that people might not otherwise consider.
The idea of a wireless crisis is an incredibly convenient tool for motivating government regulators. Elected officials assume they are responsible for solving a wireless spectrum crisis, since they allocate wireless spectrum. If it were called a "Verizon and AT&T don't want to pay for a bunch more cell towers crisis," I don't think President Obama would propose spending $50 billion on it.
This isn't just a US issue. Anything one government does in mobile data is played back in other countries as a justification for equivalent actions there. On a recent trip to Australia, I was surprised to hear a radio commentator complaining at length about the government's plan to supply broadband service to many Australians through landlines rather than wireless. You can make a good argument for using landlines, since (as we discussed in part 2) they can carry a lot more data than wireless. But the commentator was upset that Australia was failing to do "what Barack Obama is doing in the United States."
It's reasonable to ask what's so wrong with a little crisis hype and international competition. After all, governments move far too slowly in most cases, so if a bit of alarming rhetoric makes them respond faster, isn't that a good thing? The trouble is that we'll all have to live with the results after the "crisis" is "solved." In that world, no matter how much spectrum we allocate to wireless data, service will continue to have slowdowns, outages and service gaps, especially in the United States, because it's more profitable for the operators to run their networks right at the edge of overload (in this sense they have the same financial incentives as airlines).
We're lying when we tell people that the whole wireless data network could collapse. Although service problems are a certainty, there is virtually zero risk of a full network collapse, unless the operators cause it themselves by underpricing data plans and selling more smartphones than they can support. And we're misleading people when we say that prices will go up unless we allocate more spectrum. Prices will eventually go up no matter how much spectrum we allocate to data, because demand for cellular data is growing faster than supply.
By overstating the risks and talking about the "crisis" as a temporary, fixable thing, we create an unrealistic public expectation for the quality and price of cellular data in the future. That may well be advantageous for a couple of quarters or even a year, but in the long run it will erode public trust when we don't deliver the benefits we promised. The wireless operators, especially AT&T in the US, already have big image problems. Overpromising will make the problems worse. To the extent that government agencies, and mobile tech companies like Apple and Google, participate in the crisis rhetoric, they risk their credibility as well.
We need to ask ourselves as an industry if we want to have the same sort of public image in five years as the airlines have today. If not, we should be honest with people now. For example, I think there is a convincing, legitimate case for reallocating old TV spectrum for data services. Without it, mobile data prices will go up faster, and a lot of the features many of us want from mobile data may not be affordable. But we should also be honest with people that cellular bandwidth overload is a chronic disease rather than a crisis, the network is not going to collapse unless we're incompetent, cellular service will not be as fast or cheap per bit as a wired, and cellular data will generally be a supplement to our wired broadband, not a replacement.
Make the cellular data market transparent
The problem with the cellular data market as it's structured today is that it often hides from users the real cost of the network they use, so they can't make well informed choices, and it's hard for us to tell which buying patterns are genuine and which ones have been created artificially. For example, the cost of your smart phone is subsidized, so you don't realize what an expensive piece of hardware you're carrying in your pocket. You're told that you have unlimited data, but actually if you use it too much your operator will probably reduce your data speed without telling you.
By making cellular data seem cheaper than it is, we encourage people to use the network more, increasing the very overload that we're supposed to be fixing. Some of the proposals for the future of mobile data would further increase the overuse of cellular data by making it seem even cheaper to users.
The structure of the mobile market also limits competition among mobile operators (especially in the US), and reduces competition between mobile phone manufacturers.
I think this systematic distortion of the market must stop. If people could see the real cost of cellular data, they would make better-informed decisions about when and how to use it, and we wouldn't need secret back-end controls on traffic. Meanwhile, more competition in services and phones would mean faster innovation, more consumer choice, and more efficient prices.
Here are some specific steps I think we should take:
1. Ban covert traffic limits. Today some wireless operators (and some wired ones as well) are quietly reducing the quality of service they deliver to some users, without telling them. This is done through various techniques including "traffic shaping" (prioritizing or delaying certain types of data packets) and "throttling" (reducing the throughput of the network, or the speed of certain transactions). In effect it usually means reducing the connection speed of people or apps that use the network the most. For example, Dean Bubley recently wrote about an ISP who consistently reduced data throughput at particular times of the day (link).
There are some types of traffic management that make sense. E-mail spam can be reduced through throttling that limits the number of e-mails that can be sent by a single account per second. Throttling can also be used to limit malware attacks, by reducing the ability of a rogue app to flood the network with traffic. And I think it's fine to enforce the speed you paid for in your Internet connection. For instance, if you've paid for a 10 MBPS connection and the operator limits your throughput to 10 MBPS, I do not have a problem with that.
But in some cases the operators are limiting network performance to covertly restrict users, either by interfering with certain types of traffic, or by limiting the speeds of some users without telling them. For example, the current Verizon Wireless terms of service give them the right to reduce the throughput in your "unlimited" data plan if you're in the top 5% of data users (link). They can do this without notifying you.
This sort of hidden restriction is damaging to the market because people may sign up for a wireless plan believing they will get more service than they actually will. They can't make a fully informed decision between wired and wireless service because they don't know how much wireless data they're really going to get. This may misallocate resources and make the wireless network even more overloaded than it would be otherwise.
The answer to this is simple: Require operators to notify a customer when they have throttled or shaped his or her service (other than enforcing the promised speed of the connection). I am not against throttling in general, but it should not be done without notification. A text message would be fine. The Internet speedometer, which I discuss below, will also help with this problem.
2. Require a data gas gauge and speedometer in smartphones. Can you imagine buying a car that didn't have a gas gauge and speedometer? That's essentially what we do today with smartphones. For most smartphone users today, there is no easy way to tell how much data throughput you're getting from the network, and how close you are to any limits on your data usage. Some operators bundle apps to do this, some have more arcane ways to check, and some send you a text if you get close to the limit. But I think it's fair to say that most people are in the dark about their usage until they get their monthly bill, and if they do go over a limit they will have trouble figuring out why.
This is an easy problem to fix. We should require that every smartphone have an app, accessible at the same level as the Settings app, that tells the user how close he or she is to hitting any data caps in the service plan (for example, if you are a Verizon user, how close are you to getting throttled?). The app should also show how much data you're using at any particular time, so you can see how much throughput the network is really giving you.
We also should modify the signal strength bars to change color depending on how much data you're consuming at any moment. This would show you when you're using a website or app that uses huge chunks of data. When customers see that video or Flash makes their signal bars turn red, they'll be much more cautious about using those sites on the wireless network.
3. Decouple the phone purchase from the network. Currently in the US and much of Europe, if you sign a contract for a data plan, you get a discount of several hundred dollars on a new phone purchased at the same time. But you have to buy the phone through the mobile operator, giving them huge control over the selection and features of the phones they sell. Basically, users are not free to pick the phones they want; they have to take the phones their operator chooses to sell.
This operator lock-in is subject to all sorts of backroom manipulation. Weak phone vendors are forced to comply with a huge list of tests and requirements, while for stronger vendors the rules are often waived. I've also been told privately by some operators that they deliberately discriminate against some handset vendors because they just don't like them.
The handset vendors aren't completely clean either. A vendor with a hot handset may restrict its availability to a single operator in order to extract concessions from them. Can you say iPhone?
It's a wonder that some operator or handset company hasn't been sued already for restraint of trade. With the amount of operator shelf space shrinking in the US due to mergers, I think it's only a matter of time before there's a legal detonation.
In addition to the legal risk, these restrictions have the effect of restricting customer choice and competition, so they are bad for transparency. It's time to open up the handset market. To make that happen, subsidies should be separated from the purchase of a particular phone. When someone signs up for a plan, they should get a voucher for a discount on any phone. The voucher can be used at that time to buy a phone in the operator's store, or it can be used later to buy a phone in any other store.
This would encourage more selection and competition in mobile phones. It would create more direct competition between operator service plans. And it would put the wireless and wired networks on an even footing (can you imagine a wired data provider limiting the brands of PC that you can use with your cable data connection?).
In the US, I think we should consider one other step to open up the handset market. In most of Europe, and many other parts of the world, there is a vigorous retail market in mobile phones sold separately from an operator. Because everyone is on the same network standard, and because all the phones use SIM cards, it is easy to buy a new phone at retail and pop your card into it. You do lose the subsidy, but virtually all customers know they can at least switch phones if they really want to. This leads to a much larger selection of phones, and to higher competition between operators because it's easier to choose separately the phone and service plan you want.
The US market is much less open. Most mobile phones are sold only through operator stores, and it can be very hard to switch from one operator to another because they have different network technologies, and some of them don't even use SIM cards. Because it's so hard to switch phones, I think most US mobile users are barely even aware of what a SIM card is, and how to find it in their phone (most of them would probably confuse it with the SD card).
To open up the handset market, the US should require that all mobile phones use SIM cards, and that they be switchable between the major operator networks. That way someone could go into a consumer electronics store, buy the phone they want, and use it with any network. This will have to be phased in over time, but we're already moving toward it anyway. Verizon and AT&T are both moving to LTE, and there are very strong rumors that Sprint will do so as well. So some day we'll have one standard cellular technology base in the US. In the meantime, we'll have to buy dual-mode phones that use both LTE and either GSM or CDMA, depending on which operator you use. But the chipsets for smartphones are increasingly capable of handling several different networks, so they can switch between LTE, GSM and CDMA. I think it would be reasonable to require that future smartphones sold in the US be SIM-based and capable of operating on all three standards. I think the real question is how quickly we could phase in that requirement; if you have thoughts on that please post a comment.
4. Enable toll-free apps and websites. As I discussed in Part 1, we need the data equivalent of a toll-free phone call, in which a website or mobile app company would pay for the data traffic generated by a particular app or site. This requires changes to the operators' billing infrastructure, but I think it will be essential for enabling the growth of mobile data. It should be an extremely high priority for the operators, and it's in the interest of web and app companies to get together with the operators to define standards for these charges, so they'll be easy for developers to work with. I suspect there's an important role government regulators can play in helping to encourage these negotiations.
5. Do not allow the operators, or the web companies, to discriminate against one-another. I agonized over this one a lot. The operators would like to be able to charge web companies extra if they want reliable delivery of data (for example, in a time-sensitive app like video streaming), or if they want a guarantee of a certain level of throughput. I understand why they want to do this, because it would help pay for their infrastructure, and I do not think it is inherently evil. But I think it would cause too much collateral damage to the mobile market. In fact, I think it would put us on a road toward wrecking mobile data.
The first problem is that hidden back-end charges like this are essentially an invisible subsidy for cellular data. A user won't know the real cost of the data he or she is using, and this could end up increasing traffic on the cellular network artificially, contributing to data overload.
There are also big practical problems with implementing charges for quality of service. As Dean Bubley has pointed out repeatedly (link), there are huge drawbacks to this sort of approach. To give one example, there is no way to guarantee quality of service when you don't know how overloaded a particular cell site will be. If one high-priority video session comes in, does the operator shut down five other "regular" data sessions to make way for the high-priority one? In that case, the "regular" customers are not getting the service they paid for, and they won't even know it. They'll just think something is wrong with the web app they're using.
I agree with Dean that there's no way to make a system like this work predictably and fairly. Better to just charge users for the data they consume, let them know how much that costs, and allow them to adjust their own usage patterns.
The other reason we should ban quality of service fees is because in some cases they could produce in a destructive power struggle between operators and websites, with users caught in the middle. US cable television is a nightmare example of what not to do.
In cable television, it's common for network operators and content companies (the cable channels) to pay each other for services. For example, Home Shopping Network reportedly pays cable TV companies to be included in your service package, because they know they'll make more money if they're seen in more homes. They are, effectively, subsidizing your cable television service.
On the other hand, many of the most popular channels charge the cable companies a fee for the privilege of carrying them. For example, ESPN (the leading US sports network) reportedly charges cable companies about $4 per month per household; other popular channels are in the 5-20 cent per month range.
The same sorts of things could happen in the mobile web if the operators could charge websites for service. For instance, what if Facebook started offering video streaming as part of its services? If the mobile operators tried to charge Facebook for its network usage, what is to stop Facebook from turning around and demanding a fee from the operators for allowing them to carry Facebook?
Unless we're very careful, we could end up with a situation in mobile similar to the one in cable TV, where users get caught in disputes between the network operators and the content creators. Some of those arguments in the US have been incredibly ugly, with users tied into long-term contracts for cable service but unable to access the channels they thought they paid for. And remember, in cable we get these messes even though we have only have about a hundred channels to negotiate. On the web, you have literally millions of them.
The operators should not kid themselves that they would win in this sort of showdown. If Facebook cut off its traffic to Sprint's servers, what would happen? Would users abandon Facebook because it's not on the Sprint network -- or would they switch off of Sprint because it doesn't have Facebook? I think we all know the answer to that: there would be crowds holding pitchforks and torches outside the Sprint stores. The websites have far stronger brands and far more user loyalty than the operators. So it's unlikely that the operators will really be able to coerce money out of the most successful websites.
In practice, I think the operators would be able to get fees only from small startups that don't have brand awareness with users. That becomes a barrier to entry for those companies, which historically have been the source of most online innovation. To give a real-world example of what that could do to the web, look again at cable television programming: A small number of networks dominate the selection of channels, resulting in slow innovation and reduced choice. There is very low turnover in these channels.
If the web worked like cable TV does, we'd all still be using AOL for e-mail.
I've talked with people at small startup cable channels, and they are incredibly bitter about the barriers they face getting placement on cable systems. They're actually counting on the web to let them bypass the cable operators.
I think the only way to make the mobile market work efficiently is to make the payment mechanisms as clear and visible as possible. Make users pay for the data they use, and allow web and app companies to make their sites and apps toll-free if they want to, but don't start creating hidden layers of fees and subsidies. That will just distort the market and expose operators to retaliation. My operator friends, this is a war you cannot win -- so don't start the battle.
To formalize this settlement, government regulators should ban both operators discriminating against websites or types of traffic, and websites withholding their content from a particular operator or network.
6. Encourage open WiFi. As I mentioned above, we're not creating a standalone cellular network, we're creating an integrated wired and wireless network. WiFi has a critical role to play in that network, and we should make it even more central. Here's a question for you: How often have you tried to find an available WiFi network, and seen no networks at all in range? I can't speak for other countries, but it almost never happens to me in any populated part of the US. But how many times have you tried to sign onto WiFi and found only locked access points? That happens to me all the time.
We already have a very dense, well-populated wireless front end to the data network in most places that matter, but we can't use it fully because most of the access points are locked down.
There are good reasons for the lockdown. If you leave your WiFi router open, it can be hacked (actually, it can also be hacked if you keep it locked, but that's a topic for a different post). Also, in the US if someone downloads child pornography or does something else illegal on the Internet, the law often goes after the router owner because that's the only person they can find. You can read some horror stories here.
But getting those connections opened up would have huge benefits for the public, because it would take some of the pressure off cellular wireless. Rather than telling people to close off their connections, we should be encouraging them to leave them open. Regulators could help this in a couple of ways:
--First, we should require that the next generation of WiFi routers have a pass-through feature enabling public access to the Internet without giving access to the user's home network. Traffic from the user's private connection should have priority over the public one, and if public usage is excessive the user should be able to throttle it.
--Second, the law should be changed to protect people whose open wireless connections are abused without their permission.
Opportunities
So that's how I think the future of mobile data will look: unpredictable growth, always skating the line between overloaded and overpriced, and with a huge variety of users, almost all of them with some sort of limits on their data service, and many with budget plans that encourage very careful use of data. For the health of everyone involved in the market, I hope we'll also get regulations that make the market more transparent, and more open to new players.
It's a different mobile data world than many analysts have been predicting, but that's not necessarily a bad thing. Often the best business opportunities happen when conditions change unpredictably. I think this is one of those times. So I'd like to conclude by recapping the big opportunities as I see them...
For handset vendors, I think the most interesting new opportunity will be the smartphone designed for people with limited data budgets. How do you entice people into gradually using more data? This is an opportunity to do a fundamental rethinking of the smartphone user experience. Since different people will probably respond to different data features, I think it will also be an opportunity for smartphone vendors to stake out their own market segments, helping to insulate them from the intense commodity price pressure we're likely to see in generic smartphones as the market fills up.
Try to think like an automobile vendor in 1950. Do you want to compete with everyone else in midsize sedans, or would you like to dominate a smaller segment like station wagons or sports cars?
To target a segment, you'll need to hire people who know how to design integrated hardware-software systems rather than just devices, and you'll need to learn to partner closely with app and web companies as peers (rather than the serf-overlord relationships you're used to having).
In the last couple of days I've been contacted privately by some people who predict even more revolutionary moves by the handset companies, most notably the idea of selling a phone at retail bundled with airtime that you've bought from an operator. In other words, the phone comes with its own network service. That's what Amazon did with Kindle, and there's nothing in principle to prevent a handset company from doing the same thing.
I think there would be a lot of implementation challenges, most notably keeping access to that third party network if it starts to run out of capacity. But it would be intriguing to see what someone like Apple would do with this.
For operators, I think it's important to pick your battles. Although covert traffic-shaping and charging websites for service is very seductive, in the long term that will lead you into intense conflicts that you're not likely to win. It would also create more incentives for the handset companies to set up their own virtual networks, which really would transform your networks into dumb pipes.
I think it's better to focus on new business models that are a win for both you and your business partners. The most appealing of these to me is toll-free data. That would be intriguing to a lot of web and mobile app companies, allowing you to build cooperative alliances with them. And it's a whole new revenue stream that might become very large over time.
For web and app developers, the emerging segmentation of mobile data makes the idea of "enticement" even more important than it is today. How do you give people some software for free and then entice them into paying for add-ons or other apps? Already most of the mobile app developers I talk to are thinking along those lines, and obviously that business model is very well established on the web. But as smartphones reach down to more price-sensitive people who are less enthusiastic about data, there will be intense demand for apps and websites that can entice them into starting to pay for bits of mobile data.
These "data on-ramp" apps are not always intuitively obvious, and will probably differ by country (for example, mobile horoscopes were a major driver of beginning data use in parts of Asia). The companies that can find the on-ramps will be incredibly valuable to investors, handset companies, and operators.
What do you think?
That's my take on the situation. What do you agree and disagree with? What else would you add to the picture? How does it differ in your country? And most importantly, what do you think the opportunities are? Please post a comment and share your ideas.
Subscribe to:
Posts (Atom)




