With the exception of code under the old 4 clause BSD license, all BSD licensed code is fully compatible with the GPL.
That's certainly a popular claim, although by the text of both licenses, no form of BSD is GPL2 compatible, since both the 3-clause and 2-clause BSD licenses include terms that are not part of the GPL that are required to be followed in any distribution, and the GPL2 specifically prohibits adding additional terms. One of the changes with GPL3 was allowing additional terms of the type in the BSD licenses, so 2- or 3-clause BSD licenses are GPL3 compatible with appropriate additional terms.
The best thing Google did marketing-wise was not try to put the Google stamp on Android.
In many other cases they try to work the brand in where it doesn't belong, and every time they fail they dilute it a little more (I'm looking at you, GoogleTV).
They put the brand on everything they own.
They don't own Android. They did buy it, but then they turned around and sold it to the Open Handset Alliance.
Releasing 4 major versions in one year is immature, and Mozilla should no better. What motivation do they have other than competing with the other browsers that have higher version numbers?
The motivation is the motivation for everything in a lean operation: eliminate waste, in this case, specifically reducing the waste of work that is done but not delivering value to the userbase. This is done by reducing the time between the time features are first implemented and the time they are in generally-available, stable software.
This means more stable frequent feature releases with smaller sets of new features, and tighter focus in each new release.
Like accelerating the version number major releases suddenly makes the release cycle better.
Of course not.
What makes the release cycle better is reducing the number of major features that each release depends on, and reducing the impact of bumping major features from one release to the next; doing those two things means that you get new features to users quicker and cleaner -- quicker, because new feature A, once its complete, doesn't spend a long time waiting on a major release because new feature B, also slated for that release, isn't ready; and cleaner because once the major release is much later than planned because New Feature B has taken so much longer than planned to implement, there is no incentive to accept a less-than-stellar implementation of B to get out the release with New Feature A.
Accelerating version numbers is just a natural consequence of a release cycle that does both of those things.
I've said this for a long time. ISPs currently make a LOT of money reselling IP addresses, a connection with 5 static IPs costs almost double one with 2, and a static IP costs significantly more than double what a dynamic IP costs.
My current ISP has static IP a flat $5/month premium over dynamic, and does not offer any service level with a dynamic IP price low enough that that $5 premium makes the total cost anywhere close to double what a dynamic IP costs. (I think they still offer dial-up at a price where a $5 premium would be 50% higher than the dynamic IP price -- still far less than double -- but they obviously don't offer static IP on dialup.)
(I think its also $5/ea. for additional static IPs, but its effectively lower if you buy business plans that bundle in a number of static IPs.)
Why would Apple want to publish a plan if they have one?
Perhaps because a substantial fraction of Apple's shareholders are=, with reasonable grounds, concerned about the risk the absence of such a plan poses to the continued success of the company, and because Apple (that is, Apple's management team) works for the shareholders, not the other way around.
That's like my kids demanding that I tell them what we're eating for dinner next July.
No, its not.
First, because the relationship between a corporation's shareholders and the corporation's management team is not analogous to a child's relationship to a parent; its more like the relation between the proprietor of a small business to the employees of that business. Second, because what they are asking for isn't analogous to knowing an unimportant detail about the distant future, its about knowing the plan to mitigate a signficant risk that appears to have a substantial possibility of being realized in the near future.
Or not. As pointed out a couple other places in the thread, Google has since clarified, saying “The version of Honeycomb we’ve shown is optimized for tablet form factors. All of the UI changes are the future of Android. Yesterday’s event focused on tablet form factors, which is where you’ll first see Honeycomb.”
So on a platform that (supposedly) is already rife with fragmentation they are going to have completely different versions just for tablets?
One: Note the word "supposedly". Two: Nope, they won't. While Honeycomb is for tablets, Ice Cream will be for phones (or phone/tablet convergence) has been previously reported, followed by the statement leading to the misinterpretation in TFA, Google has clarified, stating “The version of Honeycomb we’ve shown is optimized for tablet form factors. All of the UI changes are the future of Android. Yesterday’s event focused on tablet form factors, which is where you’ll first see Honeycomb.” (emphasis added)
First, are tablet PCs *REALLY* the future of computing?
No, they aren't the future of computing, anymore than any one of servers running could-enabling software, traditional laptops, smartphones and so on is the future of computing.
Like each of those other things, tablets are part of the present of computing that is bound to have a role for quite some time in the future.
is it really likely that tablets will be the de facto laptop replacement in five years?
No, its likely that tablets will replace laptops for some users in 5 years (and, for some, they already have) and that they will fill serve new roles that laptops don't currently serve for other users. The set of niches for computing devices to fill is not fixed with new devices competing over the same limited set of niches. When laptops were introduced, some of them displaced desktops, but more of them opened up new roles.
Second, and more relevant to the topic, what's the major difference at an OS level in Honeycomb that makes it ideal for a tablet that's either 1.) unsuitable for mobile phones, or 2.) optimized for a tablet?
The ActionBar and some other UI changes are pretty much the only tablet specific parts. Other bits may be more resource intensive and not appropriate for current smartphones, but I wouldn't be surprised to see all of the features make it into Android versions targetting phones eventually. (Probably many of the features will come to phones relatively quickly in Ice Cream.)
One of the selling points of the cloud is the ability to spin up new server instances whenever they're needed. So why then would anyone need to buy a bunch of approximately equivalent servers for local development/testing/staging when all they have to do is set up a new environment in the cloud? Seems like a product designed for people who fundamentally misunderstand the whole paradigm.
Maybe you don't understand how the paradigm is actually concretely implemented. Servers remotely hosted in the cloud aren't magic, somewhere, there is an actual data center, with actual servers, running software that provides the "cloud" features.
Some enterprises with many functions want to have the benefits of the cloud (e.g., dynamic provisioning of resources among the various applications the enterprise is running), but prefer in-house hosting.
Some operations, additionally, want to actually host cloud services for their clients.
Either of these operations need to have actual servers in-house, running software that provides the "cloud" features.
Well, let's be honest: whatever authorization google grants or doesn't grant is irrelevant: they don't own the data (fortunately!) - which is merely keywords.
Its not merely keywords, its a mapping between search terms and search results. And ownership is irrelevant, because Google isn't filing a lawsuit against Microsoft claiming a violation of Google's legal rights. Google is pointing to what Microsoft is clearly doing, and using it to score PR points.
Users - namely google employees - apparently agreed by means of an absurd EULA (don't we love em) to send click-stream data, and Bing is merely correlating that "freely" given data.
Yep, and that's all Google has accused them of. Microsoft could just admit it and stand behind it in just the way you have, or they could deny it and make outrageous accusations of "click fraud" as a distraction. They've chosen the latter course, because the former course, while more honest, would admit that the only way Bing offers approximations of the benefits certain Google features (like the spelling correction function) is by trailing behind Google parasitically, not by having a strong search engine.
It's also an approach that, as Microsoft's OS/browser dominance wanes and mobile platforms without IE and the Bing Toolbar even available become more prominent, isn't sustainable, and yet another sign of how dependent everything Microsoft does is on the perpetual dominance of Windows, IE, and/or Office.
Now, whether an EULA should suffice to permit scraping this invasive (not just the link, but form fields or text elsewhere on the page) is kinda dubious
Because of the way Google dynamically rewrites links when they are clicked to enable their own tracking, clicked-link tracking alone (with the backend having logic to parse out the google links) is sufficient to acheive the observed results (since it gives you both the search terms and the target URL.)
In fairness, how should they prove that Google's accusations are false?
At a minimum, if they didn't do what they were accused of, they should present a plausible alternative explanation for the results of Google's experiments. (I would say they could provide some reason to doubt that Google's experiments produced the results Google claimed, but since they've acknowledged that with their click-fraud characterization of the Google methodology, it's pretty clear that they don't doubt the results themselves, so that would obviously be a dishonest PR gambit.)
But I think its pretty clear that they do exactly what Google has said they do, so what they really should do is fess up to it, and present the argument for why it isn't a bad thing. Stand by your decisions rather than denying them.
What proof is Microsoft supposed to offer? "Here's a look at how our search engine works, see, no cheating!" is just not going to work because A: that information is proprietary and very, very important to maintaining good search results, B: Nobody would understand it, and C: Even if they did that, what would that prove?
Well, nothing, because if they actually presented how the search works, it would reveal exactly what Google points to, as there is no other plausible explanation for how Bing reproduces Google results that Google produces through its spelling correction engine than that it uses the combination of Google search terms and results (whether its all the results or just the clicked-on results for those search terms doesn't matter) that it gathers from Bing Toolbar to associate results with search terms in Bing.
Flat denial with no alternative explanation for the evidence combined with hyperventilating misdirection about "click fraud" is not a particularly convincing argument, but its what Microsoft is using here, not because if they actually weren't doing what they are accused of they couldn't do better, but because they actually are doing what they are accused of, so they have nothing else to use.
The problem with the scenario you're spinning is that the toolbar that collects this information is Microsoft's Bing toolbar... a toolbar that adds a Bing search bar to IE.
That's important, because your theory makes the assumption that all users of this toolbar are Google users... but why would they install the Bing toolbar?
Because they bought a computer from a vendor that has an agreement with Microsoft to preload the Bing toolbar as part of its shovelware (from a quick web search, it looks like, at a minimum, Dell, HP, Toshiba, and Lenovo do this.)
Even if you tracked a user's click-through activities, it would only increase the strength of the webpages they go to, not correlate that data with unrelated topics.
While Google search results look like links directly to their targets (because they are, right up until they are clicked), Google uses javascript to dynamically rewrite the link target to google URL which includes the target page URL and search terms, which is how Google tracks the click throughs (this Google page then redirects to the real target with a 302 response.)
Consequently, if you track what link is actually followed (rather than what the link looks like before it is clicked), you will get the actual search terms used and the URL that the user decided to go to after searching for those terms.
Google has been careful to release software that is good for them and good for the Web. Microsoft releases software that is good for them but potentially damaging to the Web
Anthropomorphizing something that is not only not a sentient being, but not even a concrete entity is somewhat pointless. "good for the Web" and "damaging to the Web" really just mean "supportive of my desires for the web" and "opposed to my desires for the web".
Like you, incidentally, I like what Google does for the web much better than what Microsoft does. If you read upthread, you'll note that the point of similarity being discussed is about an issue raised with Microsoft's priorities in fixing what they perceive as problems in Google's product rather than devoting all of their resources to their own products, something Google very much did to Microsoft's product with Chrome Frame.
No comparison of the desirability of the motivation for those approaches was expressed or intended in the comparison.
Except that there's a critical difference between the two plugins. Chrome Frame was an attempt to bring standards-compliant CSS and fast Javascript to websites that still have IE6 users.
Right. Chrome Frame was built to fix what was, from Google's perspecitve and in terms of Google's business strategy, a problem with IE.
Microsoft's H.264 plugin for Chrome was built to fix what was, from Microsoft's perspective and in terms of Microsoft's business strategy, a problem with Chrome.
Google's preferences in this area might be more in line with mine or yours, to be sure.
I'm not sure why Apple would want to get involved in this manner with the greatest evil in our world today, News Corp.
You aren't? I thought its obvious: money.
If they want to make an app for the iPad, that's fine, but I don't see why Apple would want to publicize this new app as forging some kind of relationship between the two companies.
Because they want other publishers to take a look at what the Daily offers for News Corp and also choose to do the same kind of thing partnering with Apple.
Perhaps the filing of a complain to the FCC would be in order, maybe one to the DOJ antitrust division as well.
It worked for Google when Apple tried to pull the same thing with ads.
You are confusing the Federal Trade Commission (FTC) with the Federal Communications Commission (FCC).
And, AFAIK, the only thing that "worked" is: 1. There are rumors the FTC launched an investigation, and 2. The FTC did approve Google AdMob acquisition, citing competition from iAd.
What hasn't happened, AFAIK, is any change to what Apple is actually doing with iAd.
Nothing in the article shows how it couldn't have been the phone itself doing it, not AT&T doing it.
I see what you are doing, using product liability principles to shift responsibility from the data service provider to the merchants in the chain of selling the phone to the customer, starting with the direct seller and back up to the manufacturer.
While that potentially puts a other people on the hook, given AT&T's involvement in that chain, though, that doesn't exactly get AT&T off the hook, since in addition to providing the data plan, they also sell the phones.
Slashdot Mirror
That's certainly a popular claim, although by the text of both licenses, no form of BSD is GPL2 compatible, since both the 3-clause and 2-clause BSD licenses include terms that are not part of the GPL that are required to be followed in any distribution, and the GPL2 specifically prohibits adding additional terms. One of the changes with GPL3 was allowing additional terms of the type in the BSD licenses, so 2- or 3-clause BSD licenses are GPL3 compatible with appropriate additional terms.
The best thing Google did marketing-wise was not try to put the Google stamp on Android.
In many other cases they try to work the brand in where it doesn't belong, and every time they fail they dilute it a little more (I'm looking at you, GoogleTV).
They put the brand on everything they own.
They don't own Android. They did buy it, but then they turned around and sold it to the Open Handset Alliance.
Of course, they still develop Android.
Releasing 4 major versions in one year is immature, and Mozilla should no better. What motivation do they have other than competing with the other browsers that have higher version numbers?
The motivation is the motivation for everything in a lean operation: eliminate waste, in this case, specifically reducing the waste of work that is done but not delivering value to the userbase. This is done by reducing the time between the time features are first implemented and the time they are in generally-available, stable software.
This means more stable frequent feature releases with smaller sets of new features, and tighter focus in each new release.
If you don't understand the concept, I suggest you read Implementing lean software development: from concept to cash.
Like accelerating the version number major releases suddenly makes the release cycle better.
Of course not.
What makes the release cycle better is reducing the number of major features that each release depends on, and reducing the impact of bumping major features from one release to the next; doing those two things means that you get new features to users quicker and cleaner -- quicker, because new feature A, once its complete, doesn't spend a long time waiting on a major release because new feature B, also slated for that release, isn't ready; and cleaner because once the major release is much later than planned because New Feature B has taken so much longer than planned to implement, there is no incentive to accept a less-than-stellar implementation of B to get out the release with New Feature A.
Accelerating version numbers is just a natural consequence of a release cycle that does both of those things.
I've said this for a long time. ISPs currently make a LOT of money reselling IP addresses, a connection with 5 static IPs costs almost double one with 2, and a static IP costs significantly more than double what a dynamic IP costs.
My current ISP has static IP a flat $5/month premium over dynamic, and does not offer any service level with a dynamic IP price low enough that that $5 premium makes the total cost anywhere close to double what a dynamic IP costs. (I think they still offer dial-up at a price where a $5 premium would be 50% higher than the dynamic IP price -- still far less than double -- but they obviously don't offer static IP on dialup.)
(I think its also $5/ea. for additional static IPs, but its effectively lower if you buy business plans that bundle in a number of static IPs.)
So why does it seem as if everybody wants to make us dependent on a 24/7 connection to the
web
Web apps don't necessarily depend on a 24/7 connection to the web.
Why would Apple want to publish a plan if they have one?
Perhaps because a substantial fraction of Apple's shareholders are=, with reasonable grounds, concerned about the risk the absence of such a plan poses to the continued success of the company, and because Apple (that is, Apple's management team) works for the shareholders, not the other way around.
That's like my kids demanding that I tell them what we're eating for dinner next July.
No, its not.
First, because the relationship between a corporation's shareholders and the corporation's management team is not analogous to a child's relationship to a parent; its more like the relation between the proprietor of a small business to the employees of that business. Second, because what they are asking for isn't analogous to knowing an unimportant detail about the distant future, its about knowing the plan to mitigate a signficant risk that appears to have a substantial possibility of being realized in the near future.
And this is just Google's official stance.
Or not. As pointed out a couple other places in the thread, Google has since clarified, saying “The version of Honeycomb we’ve shown is optimized for tablet form factors. All of the UI changes are the future of Android. Yesterday’s event focused on tablet form factors, which is where you’ll first see Honeycomb.”
So on a platform that (supposedly) is already rife with fragmentation they are going to have completely different versions just for tablets?
One: Note the word "supposedly".
Two: Nope, they won't. While Honeycomb is for tablets, Ice Cream will be for phones (or phone/tablet convergence) has been previously reported, followed by the statement leading to the misinterpretation in TFA, Google has clarified, stating “The version of Honeycomb we’ve shown is optimized for tablet form factors. All of the UI changes are the future of Android. Yesterday’s event focused on tablet form factors, which is where you’ll first see Honeycomb.” (emphasis added)
First, are tablet PCs *REALLY* the future of computing?
No, they aren't the future of computing, anymore than any one of servers running could-enabling software, traditional laptops, smartphones and so on is the future of computing.
Like each of those other things, tablets are part of the present of computing that is bound to have a role for quite some time in the future.
is it really likely that tablets will be the de facto laptop replacement in five years?
No, its likely that tablets will replace laptops for some users in 5 years (and, for some, they already have) and that they will fill serve new roles that laptops don't currently serve for other users. The set of niches for computing devices to fill is not fixed with new devices competing over the same limited set of niches. When laptops were introduced, some of them displaced desktops, but more of them opened up new roles.
Second, and more relevant to the topic, what's the major difference at an OS level in Honeycomb that makes it ideal for a tablet that's either 1.) unsuitable for mobile phones, or 2.) optimized for a tablet?
The ActionBar and some other UI changes are pretty much the only tablet specific parts. Other bits may be more resource intensive and not appropriate for current smartphones, but I wouldn't be surprised to see all of the features make it into Android versions targetting phones eventually. (Probably many of the features will come to phones relatively quickly in Ice Cream.)
One of the selling points of the cloud is the ability to spin up new server instances whenever they're needed. So why then would anyone need to buy a bunch of approximately equivalent servers for local development/testing/staging when all they have to do is set up a new environment in the cloud? Seems like a product designed for people who fundamentally misunderstand the whole paradigm.
Maybe you don't understand how the paradigm is actually concretely implemented. Servers remotely hosted in the cloud aren't magic, somewhere, there is an actual data center, with actual servers, running software that provides the "cloud" features.
Some enterprises with many functions want to have the benefits of the cloud (e.g., dynamic provisioning of resources among the various applications the enterprise is running), but prefer in-house hosting.
Some operations, additionally, want to actually host cloud services for their clients.
Either of these operations need to have actual servers in-house, running software that provides the "cloud" features.
This is the market for local cloud servers.
Perhaps Microsoft could be so kind to tell everyone how IE users can opt out of this tracking so we can all avoid "click frauding" Bing?
Uninstall the Bing toolbar.
For extra protection, don't use IE at all.
Well, let's be honest: whatever authorization google grants or doesn't grant is irrelevant: they don't own the data (fortunately!) - which is merely keywords.
Its not merely keywords, its a mapping between search terms and search results. And ownership is irrelevant, because Google isn't filing a lawsuit against Microsoft claiming a violation of Google's legal rights. Google is pointing to what Microsoft is clearly doing, and using it to score PR points.
Users - namely google employees - apparently agreed by means of an absurd EULA (don't we love em) to send click-stream data, and Bing is merely correlating that "freely" given data.
Yep, and that's all Google has accused them of. Microsoft could just admit it and stand behind it in just the way you have, or they could deny it and make outrageous accusations of "click fraud" as a distraction. They've chosen the latter course, because the former course, while more honest, would admit that the only way Bing offers approximations of the benefits certain Google features (like the spelling correction function) is by trailing behind Google parasitically, not by having a strong search engine.
It's also an approach that, as Microsoft's OS/browser dominance wanes and mobile platforms without IE and the Bing Toolbar even available become more prominent, isn't sustainable, and yet another sign of how dependent everything Microsoft does is on the perpetual dominance of Windows, IE, and/or Office.
Now, whether an EULA should suffice to permit scraping this invasive (not just the link, but form fields or text elsewhere on the page) is kinda dubious
Because of the way Google dynamically rewrites links when they are clicked to enable their own tracking, clicked-link tracking alone (with the backend having logic to parse out the google links) is sufficient to acheive the observed results (since it gives you both the search terms and the target URL.)
In fairness, how should they prove that Google's accusations are false?
At a minimum, if they didn't do what they were accused of, they should present a plausible alternative explanation for the results of Google's experiments. (I would say they could provide some reason to doubt that Google's experiments produced the results Google claimed, but since they've acknowledged that with their click-fraud characterization of the Google methodology, it's pretty clear that they don't doubt the results themselves, so that would obviously be a dishonest PR gambit.)
But I think its pretty clear that they do exactly what Google has said they do, so what they really should do is fess up to it, and present the argument for why it isn't a bad thing. Stand by your decisions rather than denying them.
What proof is Microsoft supposed to offer? "Here's a look at how our search engine works, see, no cheating!" is just not going to work because A: that information is proprietary and very, very important to maintaining good search results, B: Nobody would understand it, and C: Even if they did that, what would that prove?
Well, nothing, because if they actually presented how the search works, it would reveal exactly what Google points to, as there is no other plausible explanation for how Bing reproduces Google results that Google produces through its spelling correction engine than that it uses the combination of Google search terms and results (whether its all the results or just the clicked-on results for those search terms doesn't matter) that it gathers from Bing Toolbar to associate results with search terms in Bing.
Flat denial with no alternative explanation for the evidence combined with hyperventilating misdirection about "click fraud" is not a particularly convincing argument, but its what Microsoft is using here, not because if they actually weren't doing what they are accused of they couldn't do better, but because they actually are doing what they are accused of, so they have nothing else to use.
The problem with the scenario you're spinning is that the toolbar that collects this information is Microsoft's Bing toolbar... a toolbar that adds a Bing search bar to IE.
That's important, because your theory makes the assumption that all users of this toolbar are Google users... but why would they install the Bing toolbar?
Because they bought a computer from a vendor that has an agreement with Microsoft to preload the Bing toolbar as part of its shovelware (from a quick web search, it looks like, at a minimum, Dell, HP, Toshiba, and Lenovo do this.)
Even if you tracked a user's click-through activities, it would only increase the strength of the webpages they go to, not correlate that data with unrelated topics.
While Google search results look like links directly to their targets (because they are, right up until they are clicked), Google uses javascript to dynamically rewrite the link target to google URL which includes the target page URL and search terms, which is how Google tracks the click throughs (this Google page then redirects to the real target with a 302 response.)
Consequently, if you track what link is actually followed (rather than what the link looks like before it is clicked), you will get the actual search terms used and the URL that the user decided to go to after searching for those terms.
I can't think of a single BOGO sale for an iPhone. Ever. Not so for Android
This past December, Best Buy had a buy ZERO get one free sale on the iPhone 3GS, which I think beats a BOGO sale.
Google has been careful to release software that is good for them and good for the Web. Microsoft releases software that is good for them but potentially damaging to the Web
Anthropomorphizing something that is not only not a sentient being, but not even a concrete entity is somewhat pointless. "good for the Web" and "damaging to the Web" really just mean "supportive of my desires for the web" and "opposed to my desires for the web".
Like you, incidentally, I like what Google does for the web much better than what Microsoft does. If you read upthread, you'll note that the point of similarity being discussed is about an issue raised with Microsoft's priorities in fixing what they perceive as problems in Google's product rather than devoting all of their resources to their own products, something Google very much did to Microsoft's product with Chrome Frame.
No comparison of the desirability of the motivation for those approaches was expressed or intended in the comparison.
Except that there's a critical difference between the two plugins. Chrome Frame was an attempt to bring standards-compliant CSS and fast Javascript to websites that still have IE6 users.
Right. Chrome Frame was built to fix what was, from Google's perspecitve and in terms of Google's business strategy, a problem with IE.
Microsoft's H.264 plugin for Chrome was built to fix what was, from Microsoft's perspective and in terms of Microsoft's business strategy, a problem with Chrome.
Google's preferences in this area might be more in line with mine or yours, to be sure.
But there is a "Liberal Media" Problem
Evidence?
You spend your whole post providing an explanation for the cause of a "problem" which you provide no reason to believe actually exists.
I'm not sure why Apple would want to get involved in this manner with the greatest evil in our world today, News Corp.
You aren't? I thought its obvious: money.
If they want to make an app for the iPad, that's fine, but I don't see why Apple would want to publicize this new app as forging some kind of relationship between the two companies.
Because they want other publishers to take a look at what the Daily offers for News Corp and also choose to do the same kind of thing partnering with Apple.
Perhaps the filing of a complain to the FCC would be in order, maybe one to the DOJ antitrust division as well.
It worked for Google when Apple tried to pull the same thing with ads.
You are confusing the Federal Trade Commission (FTC) with the Federal Communications Commission (FCC).
And, AFAIK, the only thing that "worked" is:
1. There are rumors the FTC launched an investigation, and
2. The FTC did approve Google AdMob acquisition, citing competition from iAd.
What hasn't happened, AFAIK, is any change to what Apple is actually doing with iAd.
Nothing in the article shows how it couldn't have been the phone itself doing it, not AT&T doing it.
I see what you are doing, using product liability principles to shift responsibility from the data service provider to the merchants in the chain of selling the phone to the customer, starting with the direct seller and back up to the manufacturer.
While that potentially puts a other people on the hook, given AT&T's involvement in that chain, though, that doesn't exactly get AT&T off the hook, since in addition to providing the data plan, they also sell the phones.