Do the simplest thing that could possibly work

> What’s wrong with doing the simplest thing?

Staff. You've got developers and they will continue working on a product oftentimes way past the "perfect" stage.

Case in point: log aggregation services like Sentry/etc. It always starts with "it's so complex, let's make a sane log ingestion service with a simple web viewer" and then it inevitably spirals into an unfathomable pile of abstractions and mind-boggling complexity to a point where it is literally no longer usable.

This is an interesting point, but there's slightly more to it than that. When something is simple and does the job well, it has limitations. The problem is that adding each subsequent feature has a small benefit and a small, but immeasurable cost. Sometimes that cost outweighs the benefit, but knowing that before the fact is very hard, and removing features is almost impossible as people shout disproportionately loudly about losing things.

It's similar to the problem of regulation. Looking at each individual law, it often seems reasonable. It's only when there are 10,000, and everything grinds to a halt, that people realise there's a problem.

True.

IMHO the log services example is excellent to illustrate this: the path that leads to their hairball of complexity is perfectly clear and every solution they do is extremely logical and obvious.

This is why these services are so much similar.

But the end result is always too complex and it seems to me that "perfect" point does not exist for this line of products.