Confidence in software

One of the best signals of quality software is user patterns showing high trust.

In middle school, we had this website where we’d keep a quarterly journal of our achievements and progress. Every few months, we’d answer questions and write essays and attach photos and such. I remember seeing classmates crying a few different times because the content didn’t save after hours of writing it up. Usually it’d be something like them uploading a file a little bit too large, an alert popping up, and then the page reloading from some unhandled error.

This happened to a lot of classmates, and the teachers started giving a warning to, “never write your essays directly into the site! Always copy and paste it over!”

I feel like this pattern is showing up all over the place. Everywhere I look there are small signs of distrust showing up in so many software tools. “Papercuts,” I think people call them.

• AI tools: I always write longer context prompts into a text editor and copy it into one of the many chat apps. Never directly into the chat app because almost all of them have issues.
• Apps and sites of all kinds: “Error loading the page. Please reload” Nice, they blanket denylisted the entire vpn range I’m on without any other traits.
• Ecosystems: I don’t respond to messages on my computer because of platform specific bugs. Phone only.

It kills me, because I get it. In a lot of organizations, priorities shift around so often that it feels like a waste of time to ship anything other than the minimum, since it might get thrown out anyways. Doing it “right” the first time doesn’t really make sense if the developer is in their own state of low confidence. And worst of all is that doing it the right way might never make sense for anybody in the entire creation process.

So to me, seeing these bugs all over the place in production now feels like a organizational problem, like a lot of things. Maybe devs don’t have the capacity or drive to “do it right,” because in the past, it got thrown away. Or maybe there’s too much focus on speed before all else. But it’s sad, because as users we are losing confidence in some of our most critical daily products, and the folks shipping this quality are probably doing so out of their own, separate lack of confidence in it being worthwhile.

The good part to this sad story is that I am seeing a lot more conversation about “craft” and “quality” lately. Since the past few months have been all about AI agents shipping entire apps in an hour, the narrative changed more to “oh yeah but their craft sucks!” So I don’t necessarily think most software will have a worse experience over time, but will likely continue degrading in places with organizational problems that cause high quality moments to not make sense for the builders.