Howdy.

01 · The spark

I typed “look” and the computer listened.

I had an Atari 2600, but my friend down the street had an Atari 600XL computer. We were playing Police Quest at his house when I typed “look” and the game actually did what I asked. That's when it clicked for me that you can tell a computer what to do, as long as you say it in a way it understands.

Later I bought that 600XL from him, upgraded it from 16K to 64K of RAM, and wrote my first programs in Atari BASIC. Then came an IBM compatible with an 8088 chip, and several upgrades later I was spending about as much time getting DOOM to run in 640K of memory as I spent playing it. It never felt like work, though.

Read the full story

02 · First job

72 hours to 7 minutes.

Right out of college, I wrote C++ code to pull patient data out of doctors' offices so a big health system could bring it into their main system. We had a two-hour window on a weekend to run the import, but my test run had already been going for 72 hours. It printed a dot for every record, and the dots kept getting slower.

After that meeting, about eight of their programmers crowded into my closet of an office, and one of them asked what sorting algorithm I was using. I said, “Right. Sorting algorithm...” and they left in disgust. Fortunately, I knew about the algorithms. It just hadn't occurred to me to use one. By that afternoon the same job ran in about 7 minutes.

03 · The side project

A train simulator in a real train car.

On nights and weekends, I built a 3D train simulator for a Norfolk Southern exhibit. It ran inside a real train car with a real throttle wired to the computer. I quoted $1,000, the scope grew, and I ended up charging $10,000 for something that should have been a $100,000 project, so I was really green.

It didn't run right for weeks, until I figured out the computer had half the RAM I'd asked for. However, the project caught the eye of one of the owners of the company the exhibit was for, and he asked me to come work for him. I told him no several times before I finally said yes.

04 · By day, by night

Enterprise systems by day. Game books by night.

By day, I started as employee number 12. The company grew and I was one of more than a hundred developers. Then I went to another company and was their only developer. I started using ASP.NET while it was still in alpha and architected a new e-commerce system, and we brought 60 clients onto it in six months. I became CTO, grew the team, and later rebuilt the system to be used by our Fortune 500 clients.

By night, I continued to work on games. I wrote three books on programming, starting with two on building games with XNA. Two of them were best sellers, and we don't need to talk about the third one. I was also a Microsoft MVP for ten years.

The downside was that 80 to 100 hour weeks started to feel normal, and that habit stuck with me for a long time.

05 · Out on my own

The right tech at the wrong time.

I started GlobalCove Technologies on 10/10/10 as a side business, and a few years later I left my CTO job to run it full time. When Microsoft announced HoloLens, I was sure it was an April Fools' joke, but it wasn't April. I got one of the first devices and started a membership site to teach developers how to build for it.

I personally spent more than 40 hours a week making content for that site, and it brought in about $1,500 a month. There just weren't enough people in the world who cared about HoloLens yet, so I shut it down.

06 · Back inside

A big bank and a small startup.

So I went back to a corporate job at a large bank, and I was pretty down, because my business had failed. My big goal that first year was earning an Xbox achievement every single day, and that actually took more energy than the system I was architecting.

Over six years there, I architected new systems on two different teams. I also served as CTO of a FinTech startup and led a remote team from six different countries. Between the two, I got to see how big teams and small teams actually work, and what slows each of them down.

Read the full story

07 · The turn

I almost quit software.

When ChatGPT started writing real code, I thought there wasn't going to be much of a future in software development. So after the bank let our entire team go, I tried to become a general business consultant instead.

Then Anthropic released the Model Context Protocol (MCP), and that changed it for me. I could see that AI was going to need good engineering around it, and I'd been doing that for a long time, so I started building a tool to help .NET developers work with it.