You may have seen my post that a couple of weeks ago I decided to try twitter. Feel free to follow me at www.twitter.com/kmerritt. My first 10 days or so have been mostly positive, other than the annoyance of the service being down fairly often and the mental distraction of the engineer in me wanting to solve their scaling problems.
Today it was revealed that twitter raised $15M in venture capital. Congratulations to them. Hopefully now they can rebuild their messaging architecture by hiring some people who can solve these kinds of scale problems. It’s a very solvable problem technically. My concern is the time it will take relative to their exponential growth. It’s a real world race condition. Can their current network survive for as long as it takes to build a more scalable one?
In all candor, they should have worked harder to not disclose their funding. Of course the funding itself is going to drive usage at an even faster rate. Why do I have that position? There’s no doubt in my mind that twitter hit a network effect inflection point recently. By that I mean at some point within the last 60 days or so, fence sitters like me observed that the twitter network itself was large enough and growing fast enough that the quality of the network was improving simply by the presence of new people. Linked In hit a similar inflection point 18 months or so ago. I think the only thing holding a lot of people back from using twitter was company viability. Raising money isn’t a business model, but it does buy twitter some time and I think more people will try it now.
There isn’t a requirement to disclose a financing. There’s a very minor risk to investors to not file with the SEC. It only comes into play if you try to go public and have unaccredited outside investors. That seems unlikely assuming twitter has had decent legal counsel. Under the circumstances, I think twitter might have been better served to keep the financing quiet and focus 100% of their energy on stabilizing the platform in part by not exposing it to undue stress. They could always spend 90 or 120 days, fix up their infrastructure, and then come out with a much more meaningful announcement proclaiming both network reliability and a significant financing.
Twitter built most of its stuff in RoR. They rewrote the memory manager to be more efficient as well. But perhaps when it comes to scale, you can push the limits of RoR only to a certain extent.