Platform On The Brain
Not a day goes by on AVC these days without some sort of post on platforms. That’s because I am thinking a lot about them and so are my colleagues at Union Square Ventures. As I said in my most recent post, the people who have the most to say about platforms are the developers themselves.
Albert Wenger, our venture partner at Union Square Ventures, has spent the past nine months building a web service and dealing with its growing user base (ie scaling it). He’s lost nights of sleep and some of his hairline, and has come out of that experience with a hunger for a new platform for building, hosting, and scaling web apps. He’s posted his thoughts on that new platform on the Union Square Ventures weblog.
I never forget I was in a meeting with a guy who was behind iTunes for Apple – he made one comment when I told him about our “platform”… he said customers don’t buy platforms – they buy applications.From that point on I changed our company pitch, and even though we are a “platform” for mobile the customers we’ve spoken to don’t understand that. They want solutions to their problems. And as you know the customer is always right.
One of greatest benefits of a building a platform and not just a product, however, comes from enabling mass customization and user innovation. There are a lot of possible ways of definining ‘platform’, but I thing the most important and exciting definition is of a platform providing people a domain-specific, higher-level toolbox of abstractions, algorithms, and data with which to solve their problems. For a consumer product like iTunes, this may not be so necessary (then again, last.fm and songbird may suggest otherwise… ;-)The easier it is for customers to use the platform, the more likely they will build stuff with it. No matter how great your C++ libs are, the population who is going to use them is limited. That’s why providing the plug-in API (level 2) and runtime (level 3) type platform is valuable, as pmarca so nicely laid out. And as the cost of providing those services come down with gains in processing and storage, the economics favour giving more users easy access to more resources.This sort of view of platform is pretty different from the low-level, magically scaling database infrastructure that Albert Wenger is asking for, but I suspect that in the grand scheme of things, scaling relational databases is an easily solvable problem and one that gets easier with time, whereas solving customer problems is trickier.
Have you read Joel Spolsky’s latest? He’s saying the time is ripe for a dominant Web 2.0 platform to arrive, and compares today to how Lotus 1-2-3 lost the war…
D’oh, this time with link http://www.joelonsoftware.c…
The human brain could be an interesting platform to build on.
Fred:The “platform” should simply be your own private domain name. Everything hangs off of it. How? Sub domains. Google’s doing it brilliantly with Google Apps. Their Calendar, Email, Docs, and Start Pages all have the option of using custom sub-domains. The key to making private domains the “platform” is to find a painless and non-technical way to change one’s CNAMEs. Go Daddy should be the ring leader here.Why not have GoDaddy and Facebook do a deal whereby Facebook users can choose between a hosted account (what we have now) and a self-hosted account (think about WordPress.com vs. WordPress.or for an example). Facebook could then let users use a private sub-domain (http://facebook.wordpress.com) as their public URL.Were this to happen, users would now actually be able to add other services to their private domain as sub domains. It would all be right there. Go to somebody’s private domain (www.fredwilson) and then choose which service you want to visit (delicious, Facebook, LastFM, etc).TGP
The sentence,”Facebook could then let users use a private sub-domain (http://facebook.wordpress.com) as their public URL.”should read,”Facebook could then let users use a private sub-domain (http://facebook.timothypost…) as their public URL.”Sorry.
thanks for the tip, fred. the comments thread over on Albert’s post is pretty interesting.