Jon Radoff’s Entrepreneur 2.0 Blog

When I read Steven Baker’s write-up on Facebook’s server farm–10,000 servers and counting–I was amazed. Why the heck does Facebook need that many servers? It isn’t apparent to me that Facebook needs more than a few hundred servers. Some other commenters observed the same. But perhaps there’s more to this than meets the eye.

Google’s server infrastructure was covered in Business Week’s Google and the Wisdom of Clouds in December 2007. I was awed by the huge amounts of money being spent to build their server infrastructure. Even the electric bill is running $25MM/year per datacenter! Nevertheless, it makes sense for them: Google runs a huge number of applications, and making their server infrastructure available to developers is a big new business for them. Amazon discovered the same thing–why not make the cloud available to everyone, and profit from the world’s hunger for server capacity?

The biggest “problem” with Facebook’s f8 platform (that’s the platform for creating Facebook applications) is that you need to provide your own servers. This has led to quality-of-service issues with a lot of applications, and also creates another barrier to getting developers to make things for you.Might Facebook be ramping up server capacity because they plan to enter the cloud business as well? It would make a hell of a lot of sense. They’ve got a platform for building social applications with a built-in distribution model; if they could also integrate it with ability to offer flexible scaling of server capacity, then they’ve got a win-win with developers that could potentially leapfrog Amazon and Google’s clouds.

I have no idea of Facebook’s intentions, but this certainly suggests some of the amazing possibilities with social media businesses as they continue to scale.

Share this post

For the blog specific to my company, check out GamerDNA, my new startup. The content of this blog is going to be more about my occasional observations about technology, entrepreneurship, games and social media–stuff that doesn’t really fit into the corporate blog.

Share this post

I wrote an article for Gamasutra on the subject of viral games. You can read the full story here over at Gamasutra.

For those who want the Reader’s Digest version, here were my main ideas:

• Design games that sell themselves; make them watchable and easy to engage with. Magic the Gathering (the card game, not the online version) is a great example.
• Design games with fansites in mind. World of Warcraft has done a great job creating an ecosystem around the company, supported by data from the Armory.
• Support guilds and clans; in other words, support the social groups who pick up games.
• Support player-created content. This is like an adaptation of the “supporting fansites” idea, but it expands on the idea. Games need to be larger-than-life–and larger than themselves. They need to allow members to participate in the whole lifestyle of the game, through machinima, player-created maps and mods, customization, etc.
• Avoid level segregation: I think this is the biggest mistake of many MMORPGs. Yes, this has worked fine for World of Warcraft…but they now have amazing scale, and they were relatively early to market. Unless friends have easy ways to play with each other, you aren’t going to engage them.

Share this post