A few days ago I finally did the thing I’d been putting off for years: pulled back all the posts I lost when I migrated to WordPress in 2013. Before that I was on Tumblr, and before that, a couple of other domains. Posts going back to the early 2000s, just sort of… evaporated.
I used Claude Cowork to scrape the Wayback Machine across all the old domains. Archive.org had a fair number of IP and browser restrictions in place. Cowork found approximately ten ways around them. No complaints. It eventually produced about 1,000 individual text files, which I then asked Claude to dedupe, tag with existing tags from my site, cross-referenced against what already existed, and quietly imported without triggering a massive RSS blast to everyone’s feed readers. (If you’ve ever done a bulk import wrong, you know.)
What I didn’t expect: some of this stuff I had genuinely forgotten I’d written. Most of it is pretty rough. I was in my early 20s, it was the early 2000s, and the bar for what constituted a blog post was different. But reading through them brought back what I was thinking about, what I was using, who I was talking to. Little time capsules from a person who is technically me. I also definitely exercised an editorial muscle and deleted a number of posts that were not import-worth.
Done by hand, this would have been a weekend of grinding work I didn’t have the appetite for, and those posts would have stayed lost. The thing that changed isn’t the what, it’s the activation energy.
I did have one moment, staring at 1,000 recovered text files, thinking: congratulations, you’ve built a content pipeline for AI training. But that’s a problem for another post.
danielandrews.com
