While the 3 technologies I just mentioned for portable social graphs are important, they don’t compare to the 10 Ways the World Might End, a TED talk by Stephen Petranek. 

Number 10: We lose the will to survive.
Number 1: Earth is hit by an asteroid.
(I’ll let you watch to get the rest. BTW, I recommend downloading the video to desktop.)

Number 10 seems unlikely … the Darwinian instinct to survive is too basic to filter out over generations.

Number 1 does sound a bit more likely now that I know a civilization-ending asteroid missed Earth by six hours.  Yikes.  Spend more money finding asteroids, NASA!

And RE: spending more money: The speaker’s solution to all of these problems was more government funding.  Personally, I generally prefer private foundations and donations funding such activities, since I believe those groups have more incentive to spend wisely and get a bigger social ROI.  However, in the case of terraforming mars (to protect against solar flares) — which would be a 300-500 year project — could a private foundation foot the bill?

I hope so … because very few governments (nor the tax-paying citizens) are forward-looking enough to fund a project that will affect their great, great, great, great, great, grandchildren.  Heck, we don’t even care about balancing the budget today, let alone 10, 20, or 500 years out.  Politically, it’d be infeasible to give up current pleasures for such long-range problems.  I’ll call it longterm-apathy, but there’s probably a better term.

At the end of my explanatory Portable Social Graphs post, I said that I would talk about ‘getting there’ in another post.  Here’s that part of that promise.

There are three technologies that are coming together to make portable social graphs happen: OpenID, Oauth, and Higgins.

• OpenID and CardSpace is the distributed framework for single-sign-on across multiple sites.
• OAuth is a protocol that lets users grant sites the ability to access their data on other sites using tokens (instead of storing their user/pass).  The first draft was just released.
• Higgins is a very loose data model for storing nodes, node data, and link information (ie: a social graph, among other things). 

Put these things together and users can very-easily login to numerous sites (OpenID), can allow those sites to share data amongst one another (Oauth), and the those sites can communicate data in a standard way (Higgins, in conjunction with a schema dictionary/translation service to explain the differences between site’s data models).

I should note that all of these things have been around for awhile.  OpenID is really old, Higgins is moderately old (2003), and oAuth is about six months old.  Why do they matter now?  Because they needed each other.

[One technology I’ve left out of this document is Microsoft’s CardSpace/InfoCard.   Even though I find it one of the most compelling and most-easily-understood identity concepts (we all understand business cards, credit cards, club cards, etc), it’s backed by Microsoft too strongly to succeed amongst web developers out there.   Higgins seems to have subsumed part of the idea on its own and although it might take longer to get the code going, I’d guess it’ll have wider adoption than Microsoft’s initiative. Note to self: Read more about I-cards and Higgins…]

So now that we’ve got the tech, all the big sites will switch over tomorrow we’ll all have portable social graphs?  Not so fast.  Users need compelling reasons to use the technologies above.  Time to create them.

I attended the Start-Up Project in San Francisco on Thursday, an event evangelizing Amazon’s web services (AWS), which provide storage/computation/queuing/checkout/etc services.  I have been following AWS for awhile but somehow expected the whole thing to be ‘under the radar.’  Boy, was I wrong.

[For those of you unfamiliar with AWS, the core/most-disruptive products are their S3 (Simple Storage Service) and EC2 (Elastic Computing Cloud), which offer reliable/persistent/secure online storage (S3) and on-demand virtual machines for computing (EC2).  Put these two together - along with their persistent messaging queue (Simple Queue Service, SQS) - and you can quickly scale up/down internet applications.  No wasteful capital expenditures, just the computing power you need.]

What surprised me most was the level of maturity and rapid/breadth-of adoption of AWS.  First, the AWS ‘tires’ have already been ‘kicked’: the services are mature.  The Internet Archive, for one, testified that the EC2 machines crash far less often and that the bandwidth has become faster/more-reliable.   Second — and relatedly — there are already many more start-ups using AWS than I expected.  Smugmug is the poster-child, and given that he started almost a year ago, I suppose I shouldn’t be that surprised that AWS is catching on so rapidly … but, just, wow, hearing the VCs say that they consider it with every investment and are actively looking for companies that are using AWS … this stuff is mainstream!

My conclusion: it’s already big and it’s gonna be huge.  EC2/S3/SQS (or something similar) will become the new paradigm of web application development, and fewer and fewer companies will bother buying servers.  Once Google/Microsoft/IBM/others-with-datacenters step up to compete against AWS, we’ll have some competition in the market and scalable computing will get cheaper, allowing previously too-expensive applications to live and breath.  (Similar to what happened when bandwidth got cheap and video/music become options … now computing power gets even cheaper.)

Of all people, Fake Steve Jobs has a nice insight into the consequences of utility computing. Here’s an excerpt from the fake-letter Sun CEO Jonathan Schwartz (AKA My Little Pony) sends to Fake Steve:

“Our industry is heading toward an enormous convulsion as we shift toward a utility computing model. Instead of selling to thousands of big enterprise customers and millions of SMBs, we’ll be selling to a very small number (like maybe a few dozen) of global utility computing operators. That contraction will cause great pain among IT providers as our customer base rapidly shrinks. Buying power will be concentrated in the hands of a few savvy and powerful customers who will demand low prices and quickly drive the cost of any new innovation toward zero, so that the generously wide new-product profit curve we’ve been accustomed to, and which we’ve used to fund new development, will be radically foreshortened. Instead of selling excess capacity and unnecessary duplication to thousands of witless IT managers who pay too much for our products, we’ll be selling at thin margins to operators who will wring every bit of capacity out of their systems.”

Fake Steve is usually hilarious, but there’s not even a joke in the above, just a spot-on analysis.  (If you read the rest of the post, it does get funny.  The above is just a set up.)

So, to recap: start-ups already recognized the value of AWS and Fake Steve already noted that utility computing is going to squeeze hardware providers as buying power consolidates.   Can I add anything new?  Of course!

• Utility computing is the best type of green technology play.  It’s all about using every cycle of computing power so that you don’t waste energy cooling idle servers.  And why do I call it the best type of green-tech?  Because you don’t even need to realize that it’s green tech to want to use it; it provides so much inherent value that you don’t need to care about the environmental benefits to use it.
• I don’t know enough about hardware design, but whatever company (Intel or AMD) best-supports utility computing will be the ‘winner’ in the enterprise market as this stuff takes off.  XXX-tuplet-core or whatever it is, bet on the processor that focuses on utility computing needs rather than the mid-to-high-end servers. 
• Speaking of which, Dell/HP really-outta get into utility computing as well, since the Amazon/Googles/IBMs/Microsofts that run the utility computing data centers will have enough buying power to skip-over the OEMs and go directly to the Asian factories for their hardware.  When running a datacenter to provide utility computing becomes a core-competence (as it will for the utility computing companies), the traditional server manufacturers (Dell/HP) may get skipped over.
• Another Dell/HP comment: If you’re not doing utility computing yourself, and if you’re not going to be selling as many servers because businesses will be using using utility computing … you’re left with the home computer market.  And how’s that looking?  It’s looking like Apple is kicking your butt.   And what about the ‘digital lifestyle’ built around TV?  Sony, Nintendo, Microsoft, and the set-top DVRs have better penetration.  Hmmm … I don’t see any niches left for the two currently-largest computer manufacturers?
• With the proliferation of ever cheap computing power, “here’s to” yet-another evolution of inefficient programming languages and techniques that take advantage of the low cost.  I’m not saying we need to go back to squeezing programs onto punch cards — and I can’t complain because I love garbage collection and other benefits of cheap memory/computing — but, well … it’s a little sad to see programming become less efficient.  It will be interesting to see what comes of it, of course; perhaps future “garbage collection” will mean “shutting down machines I had been temporarily using” rather than just disposing of variables I had allocated in memory?

And just a reminder: don’t invest off of my predictions … spend the money on yourself doing something in the field if you want to make money.