I was in Enterprise software and even though I didn’t visit users, I dealt with them regularly eg through video calls or engaging with them via support forum if support escalates an issue.

And yes we were judged on how pleasant to use our software was. If we miss a feature or ship a feature that customers intensely dislike, best believe that we’ll get a torrent of negative feedback on our support channels

In my experience enterprises do not want to pay for excellent software, simply because it is expensive. I would love to work on software that was:

- in more than two AWS regions

- required screen reader / disability support

- required multi-language support

- required multi-cloud

- actually needed a big Hadoop cluster - most enterprise data processing can be done on a MacBook Pro M4.

> in more than two AWS regions

As someone who has had to deploy services in every AWS region, of every AWS partition, I fantasize about two regions being sufficient.

My second major project at $ENTERPRIsE was 15 years ago and incorporated the first 4 (well multiple servers on different continents, why would I care about AWS)

It was internal shadow IT at its finest. It’s still used today, despite the people at the top not liking it the users think it’s the best thing since sliced bread.

Agreed. Another point IME is that the problematic requirements are also often non-essential, such as using a certain in-house framework made by some director, or building the whole project inside some abomination, like the workflow system of the Enterprise Content Documentation system.

This is how I’ve seen two-month projects becoming multi-year multi-team behemoths.