The Official Blog of Cogmap, the Org Chart Wiki


Archive for May, 2011


Programming Cults in the Baltimore/DC Area

Friday, May 13th, 2011

If you are a non-technical founder of a company similar to the kinds of companies that I might start, your job is to be a talent magnet. You have to convince awesome people to quit their jobs and join your start-up. If you can’t do that, then you probably aren’t doing your job. Your job as a non-technical founder in a very early stage software start-up is three things:

  1. Get critical talent to join the team
  2. Sign up a few customers
  3. Raise money

If you can’t do #1, #2 is hard – customers like products.

If you can’t do #1, #3 is hard. As I told a group of entrepreneurs at a recent Founders Institute event: “If you go to a sophisticated investor and ask him to put money in, but none of your friends have joined your company, that is a huge red flag: If you can’t convince your friends to quit their jobs and join your company, why would a complete stranger trust you with their money.”

One thing I think about a lot when I think about what my next company might look like is Steve Newcomb’s “Cult Creation” essay. One section in particular is very compelling to me, so I am going to quote it in its entirety to give context to the rest of my blog post:

Create a Dominant Market Share – one of the things that I took notice of was Google’s move to develop a lot of their tools in Python.

Curious, I thought.  Why would they do that?  At the time Python was a new(ish) language, although growing quickly in popularity.  Then it hit me.  They were going for a dominant market share in a specific talent pool. If you can get in on a new talent pool trend, the benefits can come back ten-fold.

Here’s the strategy.  Get the first luminaries in the field, then as that language grows in popularity you are labeled as the de facto place to go if you want to code in that language.  Then hiring get 10 times easier.


In 2005, when we founded Powerset, we realized Ruby was the new Python, so we went after some A-level people in the Ruby community.  The top two we went after were first, Kevin Clark (a 20 year-old wiz-kid who we were trying to convince to quit school) and second Tom Preston Werner (now the founder of GitHub).

We got both of them, and within a matter of months, we had one of the largest Ruby teams on the planet.

Anyone who wanted to code in Ruby knew about Powerset simply from the Ruby meetups which were dominated by either Powerset or Twitter people.

We then did the same thing in the field of computational linguistics.  At one point we estimated that of the 200 or so people that really understood computational linguistics in the world, we had about 40 of them.

What’s the benefit?  Once we knew we had this level of talent market share penetration, we had almost a guaranteed worst case scenario that most startups would dream about.  We knew that our talent pool was so strong, that even in the event that we just ran out of money, one of the big three search engines would simply buy us for our team.

At that time we knew that a talented engineer in a tough to get tech was worth about $1.5 million per head.  Thus, I knew with relative assurance that since we were going to hire at least 70 people with our Series A money, that our worst case scenario was about a $100 million exit.

If anyone is paying attention, you are now saying, wait a minute!  Didn’t Powerset sell for $100 million to MSFT?  …. Yup, we nailed our worse case scenario!

That really struck home with me. Recruiting engineers is incredibly difficult and this implied that it might be easier: All you have to do is pick a language that you can dominate the market in and you are positioned to easily recruit engineering talent – one of the hardest parts of growing a company.

So what language should I choose if I am starting a company in the Baltimore/DC area?

Interestingly, Baltimore has a very large Rails community: So many Rails people that RailsConf has come to Baltimore several years in a row. But of course, that means there are already places that are Rails nexi for the community: 410 Labs and Smart Logic Solutions come to mind.

I think of Clojure or Scala as relatively specialized languages – probably inappropriate for whatever start-up I may one day have in mind. What choices do I have? PHP? Python?

I am interested in hearing people’s thinking here. What are the best languages for me to try and build a cult of ninjas around in this area?

I Love Math

Thursday, May 12th, 2011

Some of my favorite people have already responded, but I had been planning to write a blog post for so long that I feel like I still have to get it off my chest:

An article in Ad Age Digital, “The Dangers of Online Advertising’s ‘Math State’“, struck me as verging on irresponsible. I know people love a contrarian view point – look, I am blogging about it – but c’mon!

I had two immediate reactions:

  1. I am as online quant geeky as it comes. But even I recognize the value of great creative. Making great creatives has not gone down in value – I predict it will be the primary driver of the growth of online advertising over the next 20 years. But there is science there. Everyone who has written software knows that “great process provides a framework that can unlock creativity”. We have not yet determined how great creative can be expressed online – there will be an element of math and science to that – but that is no reason for a creative person to be scared.
  2. Kendall devalues what we have done online in the worst way. If I told you that we had developed tools that could determine with amazing precision, person-by-person, at an individually targeted level, the effectiveness of TV ads, would Kendall have said, “All the creativeness has been removed from TV advertising.” Not at all. We are introducing amazing new targeting and measurement in a new advertising world. That is good. The fact that the creative format is not as good today as it should be does not take away from the value of what has been constructed.

One of the things that I loved about this business from the moment I got started in 1993 was that we have the opportunity that John Wanamaker dreamed of: The chance to figure out what advertising is working and what advertising is not. And make it work better.

And I love math. Math is beautiful. Working with numbers is a beautiful act of creation and discovery that is fun and makes you a better person. Math is a gift to humanity that we have been given to explore our universe. Take that.