Posts

Alternate title: Google to world: Quake in fear and awe before our mighty data centers

One of the features in the current issue of Wired magazine is an interview with Eric Schmidt, Google CEO. You can read the interview online, along with most (if not all) of the rest of the current issue.

Aside: It's getting increasing difficult to justify subscribing to magazines these days ... but that's another post for another day.

I keep telling myself, "more and shorter posts", but then it happens. While I'm minding my own business, along comes an interview with Eric Schmidt in Wired and I get sucked right back into fighting the good fight; toiling in anonymity to protect the good and decent people of the world from the reckless and misguided few who would do the rest of us harm. I'm not unlike Spiderman really, without the web-themed costime and popular appeal.

on with it then...

There are a lot of things to comment on and criticize about Schmidt's answers to Wired's questions, but the general sentiment I want to express is this: much of what Google is doing doesn't make a lot of sense if we're going to judge their actions by what they say about them in interviews like this one. I am of course working under the assumption that we're not being given the whole story, but I'm also assuming that we're not being wholly lied to.

I'll try to lift as little of the interview as possible, but because I do want the reader to be able to understand what I'm writing there will be quite a bit quoted from the article. All quotes are clearly indicated. Please do go read the original at wired because this post does not cover, let alone quote the whole thing.

Wired: Google gets its revenue essentially from one source - online ads. One could argue that it's not diversified enough.

Schmidt: The criticism is valid ... But there are some new revenue models on the horizon. The most interesting is probably Google Apps, where we're already beginning to get some significant enterprise deals.

OK, we start off with a little honesty and then head in a bad direction pretty quickly. As I've already said in a post titled Google is an ill-mannered child with a sharp stick

'there is, exactly one [business model] - advertising. Well to be more accurate, Google's business model is advertising and everyone else's is Google.'

I won't say any more about why I think it's troubling that Google sees nearly all of its revenue from ads, and furthermore that it hasn't proven that it can deliver even one other profitable service. The fact that Schmidt acknowledges the problem is sufficient for the purposes of this post. That's the good start.

But let's hope that when Schmidt says that Google Apps are "the most interesting" potential source of revenue on the horizon, that he doesn't mean that they are the best potential source of revenue. I'll just say it, hosted productivity apps are a bad idea, especially for the enterprise. That will always be true unless there is some fundamental change in how internetworks (the internet in particular) work.

I'll talk about why I think this is true after the next quote.

Wired: Google Apps lets users do word processing, spreadsheets, and email in a Web browser. Then you store all the files in a giant server farm somewhere. Will businesses really be willing to work that way?

Schmidt: Corporations are tired of dealing with the complexity of the old model, and our products are now strong enough to serve business needs reliably.

Schmidt makes two statements here and both are unsound.

First, 'Corporations are tired of dealing with the complexity of the old model, ...'

The cause of the complexity of enterprise IT is not the productivity applications, it's the infrastructure. It's true that there are some training issues related to use of applications, but hosted applications are no answer to this. There are some unnecessary complications due to mismanagement of licensing and compatibility issues related to running multiple versions of apps within the same enterprise, and other more minor issues. The solution to these problems is adequate in-house IT resources and competent staff. Automated administration and remote management tools help quite a bit too.

Because these things (appropriate in-house IT resources and competent employees) are a requirement of any successful business, large or small, then the successful business is already well prepared to handle these relatively minor issues related to support of productivity apps. The reality is that IT managers and CIOs, even the support staff who routinely deal with application related issues, do not stay up late at night worrying about Word, Excel and PowerPoint. On the contrary, even problems that seem related to these apps are typically symptoms of other issues. Hosted apps may change the situation somewhat but trading new problems for old is not in any sense a solution. In fact in the short term moving to hosted apps might upset the support resources you have in place now. Hosted apps mean more pressing end-user support issues related to network infrastructure. Do you have enough network administrators to pull them into tier 1 support? Regardless, hosted applications have their own licensing and compatibility issues in addition to privacy, security, access, liability, management of active files and archives, and many others. Each of these is complex enough a topic to be the subject of not just magazine articles or books, but careers and entire industries. And all of those issues, if I took the time to list the dozens of other inherent problems, are only a small subset of the total . There are in fact far larger complications to consider. Unfortunately I can't continue with a protracted discussion of the complications of enterprise IT here. So I'll end this part of the post by saying that running a locally installed application on a standalone PC is a relatively uncomplicated affair.

Sidebar: I say relatively uncomplicated because if we consider everything involved there is essentially nothing uncomplicated about modern computers. From hardware, which should rightfully be discussed as a collection of components, each of which is a complicated device on its own, to the operating system, again a collection of very complex subsystems, device drivers, and applications.

Adding a network interface and an active connection to a local network complicates things, even if the user isn't intentionally or actively using the network.

Throw in some servers and the requirement of communication among devices on the local network and things get quite a bit more complicated.

Tie that local network to an internetwork, that joins together a vast number of other networks, and the complexity ratchets up significantly.

Now add to the mix several hundred million hosts and thousands of services with different agendas and requirements all competing for use of the network and the situation is radically more complex.

Factor in that some of those hosts are actively trying to disrupt or compromise your use of the network and things are more complicated still.

Finally take your local application and your data off of your computer, where you and your support staff have complete, unrestricted, and immediate access to it, and put it at some far off point on the network where it is entirely out of your direct control.

We don't have to evaluate this situation in great detail and we don't have to decide in favor or against hosted apps to consider Schmidt's statement. Would you say the new model is more or less complex than the old? Again, I'm not arguing here that hosted apps are a bad idea, I'm just saying that it is ludicrous to describe them as uncomplicated. Schmidt knows that, but he's hoping you're gullible and ignorant enough to be persuaded to let Google gamble with your data and your business.

Second, '...and our products are now strong enough to serve business needs reliably.'

This is an insincere claim. It shouldn't be trusted because even Google doesn't believe it (and who would know better than Google about the state of its own apps). How can Schmidt expect us to believe that Google Apps are 'strong enough to serve business needs reliably' when all of the services that make up Google Apps suite are in beta or worse:

All of these apps are considered beta apps by Google: , Docs, Spreadsheets, Gmail, Google Calendar, Google Talk.

Page Creator is still considered pre-beta as part of Google Labs.

Start Page, the last service Google offers, isn't a service at all. It's just a portal, similar to, but with fewer features and less customizable than the other major portals you'll find online, for ex Yahoo's portal offerings.

Even if it is demonstrated somehow that Google's Apps are reliable (getting rid of the beta tag would be a good first step) this fact is almost completely meaningless because Google's products are not the only things involved here. I made this point above when discussing the complexity of the situation but here I'd like to emphasize the fragility of it. Fragile? What the internet? But the internet is capable of withstanding a devastating and wide-spread natural disaster or even directed attacks, isn't it? That may be true but it is the internet that is robust not your connection to it. Aren't they the same thing?

No, they are not. In fact if we say that the internet is resilient that is because it is able to tolerate and adapt to that fact that your network will frequently be unavailable for any number of reasons. And that is great for the internet, but it doesn't help you when you're on one side of a disruption of network services and your data and applications are on the other.

Regardless of Google's statements, from the perspective of your computer or your network their apps are no more reliable than the $20 network interface card in your PC, the cheap firewall box you bought sitting in a closet somewhere with no environmental controls, or the guy outside the building operating the backhoe about to cut through the cable connecting you to your ISP.

We've all experienced the inconvenience of disruption in network connectivity. Without email, web access, or other important services that necessarily depend on the internet, we're limited in the work that we can do, but it's usually possible to accomplish something precisely because we have local data and local applications. In a world of hosted applications, any interruption in network activity becomes extremely difficult to deal with. Of course we can make exceptions, build in redundancies, and have backup plans but there's no significant advantage to doing any of this. The fact that it might technically be possible to have hosted applications when it wasn't before is not enough of a justification for doing it.

The last thing I want to say about this part of the article, and this really is the subject of a different post, is that the web browser is not the ideal container for all of these different applications that we're shoehorning into it. The web was not designed to be a framework for the development and delivery of other apps. That doesn't play to the strength of the web and it doesn't do justice to the new applications either. You can see this all the time with new web services, and with Ajax especially. These apps are described as "more like standard desktop apps", meaning more like standard desktop apps than traditional web pages, but they're certainly not more like desktop apps than actual locally installed desktop apps. The browser has its own behaviors, its own interface elements, its own patterns of use; all of these different applications don't fit that model. A back button has no meaning in the context of a spreadsheet application and neither do the address bar, bookmarks, etc. There's no eliminating the fact that these are integral to the browser. All that we're accomplishing is making clumsy applications while at the same time making the browser more confusing and more difficult to use. There's a simple reason why I would argue that this is necessarily true. The web is already an application!

Part of the interview deals with YouTube, and more specifically the issues related to copyright, litigation related to copyright, and how this issue affect Google's plans for YouTube.

"Wired: Viacom's argument is that you're not working hard enough to keep infringing clips off of YouTube.

Schmidt: ...under the Digital Millennium Copyright Act, there is a shared responsibility. The law says that the copyright owner monitors - and then we expeditiously remove - offending clips. We've done that. In fact, YouTube's traffic has grown since we did. So Viacom's argument that YouTube is somehow built on stolen content is clearly false."

Again, I have two issues with this. Actually, one comment and one issue. The comment first.

I can't say that in recent memory I found myself arguing for copyright owners. This doesn't mean that I don't respect intellectual property rights, I simply haven't agreed with many of their arguments. But I will say that in the case of something like YouTube the DMCA and its shared responsibility puts copyright owners at a disadvantage. As a short term solution it makes sense that if I'm a copyright owner and you're Google, and you have a bunch of infringing content on your site, then I would submit to you a list of material that I allege to be infringing so that you could look into the issue and remove content when appropriate. But, if this is how it is going forward, it doesn't make much sense for me, the copyright holder.

No matter how diligent I am there's going to be some non-zero amount of time from the moment the material appears until I'm aware of it, notify you about it, and you pull it down. For the time that the content is available, it's valuable to you. In fact, it's possible that because users know that my content will only be available on your site for a very short amount of time users will be more active and you will see more traffic than you would have if there was an expectation that the content would be available for longer period. It's not hard to imagine. If there's always something new and everything that's new is only going to be available for a very short time, then, as a user, if I want access to all of it I need to be very active all the time. This seems pretty beneficial for YouTube and Google. Meanwhile the copyright holders are expected to do a lot of work, for which they aren't getting any compensation. In fact there is cost to this for copyright holders, beyond the time and effort spent inventorying content, and that is resentment that users might feel for what they perceive as something being taken away from them. Users are unlikely to blaime YouTube, afterall YouTube is what made the content available in the first place. Of course owners want to defend their IP but they should be wary of alienating potential customers.

I take issue with the second part of the question. Namely, that because YouTube's traffic has grown while they've been removing content, this somehow implies that YouTube has not benefited from the same content. Or as Schmidt puts it, 'that YouTube is somehow built on stolen content'.

Of course this content has contributed to YouTube's popularity and success. If YouTube wasn't built on stolen content, then it was built with stolen content. It's really just semantics. Although it's more difficult to do after Google removes the content in question, all you need to do is look at the popularity of the pirated material and the sheer amount of it to gauge its value to the service. Time will tell if YouTube can be successful without it. But the fact that YouTube continues to enjoy popularity in the short term is not a good indicator one way or the other. A simple explanation could be momentum. A ball that is pushed continues to roll for some amount of time after you stop pushing it. If the conditions are right, it may even gain speed for a short time - maybe it starts rolling down a slope, or the wind picks it. Momentum isn't the only explanation. The fact that people get lung cancer after they stop smoking, does not imply that the cigarettes aren't the cause of the cancer (or a significant contributing factor). Maybe if things don't work out for Schmidt at Google he can go work for one of the tobacco companies. He seems to have a knack for the type of broken logic they employ when faced with sketchy legal issues.

"Wired: You thought there was a good chance of litigation when you bought YouTube. The deal set aside $200 million to cover the cost of law suits. Why did you make the acquisition if you anticipated so much hassle?

Schmidt: ...Because we think it's fantastic."

Schmidt goes on to say that they are integrating search, which is both obvious and doesn't add much to the argument. Of course, they must have thought it was fantastic to spend $1.65 Billion. I think the intention of the question was, 'what about YouTube is so fantastic?'.

To me it seems like they got their numbers reversed. They should have made the deal for $200 million and earmarked $1.65 billion to cover the legal costs.

Nobody is doubting the appeal of video. The value of moving pictures was established before the end of the nineteenth century. The problem is that there's nothing about YouTube that justifies the acquisition. (Except maybe the fact that YouTube had already built a very large audience based on stolen content, something that Google could never do itself, but Schmidt isn't willing to acknowledge that of course.) Google had a video service before YouTube. There are half a dozen other major video sharing services. These other services offer more in the way of features and functionality than YouTube. This is true of Google's own branded video sharing service. Furthermore, there are more options for online video now than there were when YouTube first came to light. If you ask me, it's a crapshoot. I wouldn't bet on this play.

"Wired: Let's talk about those data centers. They're changing the way the world thinks about computing. Explain.

Schmidt: It's pretty clear that there is an architectural shift going on. These occur every 10 or 20 years. The previous architecture was a proprietary network with PCs attached to it. With this new architecture, you're always online, every device can see every application, and the applications are stored in the cloud."

This is what people always say when they want somebody to buy into their kooky, bet-the-farm idea based on nothing more than their reputation. This is Schmidt playing the part of the luminary. It isn't much different than some cult leader predicting the end of the world. That's a pretty safe thing to do as long as you're smart enough not to say that the world is going to end tomorrow or next week. 5 years from now, maybe 10, that's a better time to say the world is going to end.

I've seen the same argument before about every piece of what Schmidt mentions in his quote. He's just mixed them together a little. Apparently the previous architecture was propriety networks with PCs. Well that's true, but the time has passed for proprietary networks for the most part. There's no point in predicting it. That was early to mid 90's. This is about as radical an idea as predicting today that rap will have significant influence on the music industry. PCs on the other hand have survived that architectural shift and are still going strong. People have been predicting the demise of the personal computer for a very long time. I suppose that at some point PCs will look sufficiently different that in some sense this prediction will come true. But I'd be willing to bet Google won't be around to see it.

So apparently, according to Schmidt, the current phase is PCs and proprietary networks, and the next next phase is 'where you're always online and every device can see every application and applications are stored in the cloud'.

This is just blather. There may be a point at which people feel as if they're always connected, but it's not going to be in the next few years.

Today we have wifi, a situation that's getting worse not better as everyone runs out and buys access points, and all of the redundant wired connections we have results in a lot of noise. I can't access networks now that I could 6 months ago, and it's not because the networks went away. It's because there are dozens more of them.

The cell phone companies have working solutions in their high-speed 3G data networks but rate plans are $60-100/month. Usage, at least in this country, certainly isn't mainstream.

No one has proposed a plan that will result in 'every device' being 'always online' much less a scheme whereby 'every device can see every application', whatever the hell that means. These are predictions in the same spirit as HG Wells' The Time Machine.

"Wired: How many data centers have you guys built?

Schmidt: In the dozens. There are a few very large ones,... but in a year or two the very large ones will be the small ones."

I don't even know how to address this. It's ominous. If you didn't see reason enough to be wary of Google before, that quote should pretty much do it.

"Wired: Google controls its own fiber optic network. Why?

Schmidt: We can tune it. One of the neat things about the internet bubble of the 90's was that people built all of this fiber, and now it's essentially free."

I have a couple of things to say about this as well. First, apparently Google controls its own fiber-optic network because owning the technology your business depends on makes sense to them. Say, as opposed to relying on some service provider and the reliability of the the public internet for business critical functions. I agree with his reasoning. The problem is it runs counter to the same arguments he made when discussing Google Apps in this same interview. Apparently Schmidt is saying, 'you should trust the internet with your business critical applications but we know better than to do the same thing'. That doesn't make me feel good about Google.

This next part must be my favorite of the interview.

'...one of the neat things about the internet bubble...'

Yeah, the internet bubble, remember that? That was so neat! The most significant loss of personal investments in history, ruined lives and careers, an industry left in turmoil, the stockmarket left in shambles, computer science as a field still hasn't recovered, and WorldCom. Ah WorldCom, now there were a bunch of neat guys. I guess if it wasn't so tragic, it would be funny.

Listening to Schmidt talk about how neat it is I can imagine a time in the not too distant future when the CEO of some future Google will be sitting down for an interview (maybe an interview for Wired. Of course by then the format will be such that the only people who capable of reading it are specially trained diviners) and offers this: "One of the neat things about the disastrous implosion of Google and the internet in the early 21st century was that they built all of these massive data centers and now they're essentially free."

"Wired: How should we think about Google today?

Schmidt: Think of it first as an advertising system."

How am I supposed to make sense out of an advertising system that wants to control all of my data and my applications, and wants to dot the face of the earth with dozens and dozens of data centers so large that the only way to convey the enormous size of them is to say that in a year or two the very large ones will be the small ones?