Tim Bray: "I'm having a little trouble understanding Android; the business side I mean, not the technology."

Android itself isn't the entire point, though as a developer it's cool to imagine a mobile environment that doesn't require so much damn code porting.

It's Big! How Big? Really Big!

Sometime in the next two years there will be 4B mobile subscription plans. Marc Andreeson is really excited about hundreds of millions of new web users per: "Look at it this way: most users on the Internet (1.3+ billion, with 100 million joining every year) are not yet using any social networking service."

What Marc is saying is that about there are about the same number of Internet users today as the entire world popluation circa 1850; but what we know is that there are same number of mobile plans today as the entire world popluation circa 1970. Internet usage has roughly matched population growth in the last two decades but mobile usage outstrips that; the number of new mobile plans is about 1000 per second, exceeding the birth rate. 1.9 trillion is the ballpark figure for mobile messages sent in 2007, most of those were in Asia, 95% of which were not spam, unlike email. 10 year olds get mobiles, not PCs. The number of mobile devices will double in the next five years. They are premium source of the fastest growing kind of data on the planet, User Generated Content (UGC). 95% of UGC is unstructured content, and about 75% of the total amount of it is replicated.

Mobile operators have sticky relationships, however tense, with customers in ways webcos can only dream of, but the application experience is not great, the form factors are tricky (the iPhone is a bit bigger for exactly that reason), and the fragmentation of platforms is a killer, making it a expensive place to develop and innovate on. Browser variation doesn't seem so bad compared to getting anything cool to run across handsets.

Android+webkit and iPhone+webkit, and most recently Nokia+Trolltech (home of khtml, of which webkit is a fork) potentially rationalises a lot of development effort that goes into porting apps; more important it leverages millions of web developers bored to tears doing yellow fade outs and portals. Android potentially lowers the cost of all that, assuming it ever gets widely deployed.

So, here's a gross simplification. All the cool stuff happens on the web, not on phones (unless you're in the far East), but mobiles are where the paying customers are. Webcos are obsessed with mobile eyeballs, telcos are obsessed with web sociality.

The web as lemonade stand. 

Tim: "I think that the problem in the mobile space has always been the business model, not the technology."

I would say it's not that simple, but you can if you want look at mobile as the commercial inverse of the Internet. The networks (not network, more on that in a minute) are monetised, but the apps are not. You can't count ringtones and games and chatliness as applications, they're content. When I say monitized I mean people pay good money for voice and data. SMS is certainly the most expensive way to communicate data that I know of.

Instead of competing with Android directly, operators will probably make carrier deck deals with 'pure play' SNSes like facebook, myspace and bebo using proxies for content upload and transcoding. And maybe build decent search apps into new phones, or buy some of the search startups for location based searching - that would send a counter message that the operators understand Google's model as well as Google understands their's. Nokia are already moving into in that game, having paid $8.1B for Navteq. And since I wrote the draft of this, they bought Trolltech. Clearly they're thinking about platforms and not just handsets.

Google seem to be doing a good job on developer relations; expect to see operators and handset makers put more effort into that this year. And in Europe at least, Google are hiring up on handset developers.

Android is a strategic play and not a 20% time thing that got lucky. Google might do really well by disrupting mobile development and coordinating the smaller handset providers in the Open Handset Alliance (when you are not the dominant player, or are negatively impacted by fragmentation, you encourage standards). But it's not obvious that the ad model is the right one for mobiles. Remember this is a world where people will pay four dollars for a 30 second ringtone of a track, but on the Internet they'll expend no small effort 'stealing' the whole thing. People might conceivably pay for content on mobiles that they would never dream of paying for on the Web. At the moment tho', Android doesn't ship with a video support (or flash), just sound.

Sh1ttabytes

"But as a Sun shareholder, I see another huge upside in the mass of server infrastructure everyone will need when the network operators unclench and we get an explosion of creativity and Mobile/Internet apps."

Further up I said "networks" and not "network". Even though mobile phone usage dwarfs web usage, networks tend to be conveniently partitioned by operator and by region. So there isn't a true network of networks like the Internet; the ecosystem are siloed, the technology and protocols are complicated. Incidentally these are probably contributory factors as to why it's hard to find any cool innovative IMS apps.

One way of looking at is that the lack of end to end openness of mobile networks and general poorness of mobile apps acts as a important throttling factor. What happens when half or even a quarter of the world's population gets onto a converged IP/telco network, and starts hitting the web hard over MMSCs or XMPP gateways? Other than a lot of downloads of squid, it beats me.

If we're talking about Sun shareholder value, I would say focus on storage. We're reaching a tipping point where our ability to generate content surpasses our ability to store it, and without sounding like Bob Metcalfe, I'm not sure enterprise datacenter models (ie, a SAN) will cut it if the telco networks really open up. What you tend to see are companies that have traditionally served enterprise IT producing bigger and bigger boxes (like Sun thumpers running ZFS). These are kickass for big Oracle RAC installs or MySQL read farms, or a billion media files. But consumer content is where the growth is, there's too much of it to fit into "one" filesystem, the stuff isn't like structured IT data at all. Arguably most of it is never going to go into a database. Following this tenuous line of reasoning, what Sun should be building is "storage as a service" - a lot of little boxes chained together, data grids not compute grids. They certainly have the filesystem in ZFS, what's missing is a service layer. In that case Sun should be paying devs to work on Hadoop/HDFS and MogileFS. Because one thing is for sure, if you believe even half of what people are saying about data growth, voip, ip-TV and all that, storage and network infrastructure providers are going to make untold billions on the back of UGC traffic. 

If you had to pick a company that "gets all this", it's Amazon - which is why Redmonk James is right - someone should buy them before they turn into the 21st Century's answer to General Electric.