The first version was hand written HTML with a turquoise rule under every heading, and it stayed exactly as it was for seventeen months. The one project on it, a Connect4 game, got commented out at some point and replaced with the word WIP. Three of the four links in the nav pointed nowhere. None of that was a coding problem. I had nothing worth showing yet and I was putting my time into the wrong parts of the thing, and because my taste was improving faster than my skills were, I could see clearly that what I had was not good enough to put in front of anyone. So I did not put it in front of anyone. I left it alone and I did not share it for a very long time.

The next version borrowed a navigation idea I had liked on another site. You clicked and dragged the heading on the hero and a carousel of projects opened under it. I was pleased enough with it to send it round to friends, and most of them either could not work out how to move through it or found it behaved badly on whatever they were holding. I had not thought about any of that while I was building it. The commit history for that version is mostly me going back and patching it afterwards: keyboard support, a hold and drag gesture because a plain tap kept firing by accident on phones, a contact form that stopped being usable the second a mobile keyboard opened over it. Accessibility is not a pass you make once the interesting part is finished.

After that I started again on something modelled very closely on ning-h.com. I liked that site enormously, and what I was really doing was rebuilding it and calling it mine. I never got it to the same standard, and the nearer it got the more obvious it was that the parts I liked best were not my ideas. It lasted sixteen days and it never went live on my domain. I would still do it again. Working through somebody else's decisions taught me more than reading about them would have, as long as the copy is not the thing you end up shipping.