What are everyone's thoughts on moving away from the concept of Story Points?

In a nutshell, while I do believe estimation is fundamentally hard (even in traditional construction/engineering, and even more so in software by at least some significant margin), I don't think that's really the problem with estimation.

The real problem with software estimation is that I think it's not worth pursuing. What is worth pursuing is speed of delivery, which necessarily means you have to build your engineering muscle around things that are vastly more important than estimating, like:

• Building small things that are good enough
• Iterating on them rapidly
• Running experiments

It's not so much that software can't be estimated, it's that I don't think it's really all that useful to do so.

The number of times that specific dates for software releases actually matters should be small, and even then, the importance of the date itself should be minimized.

When tech companies release software these days, even if it's a big splashy public announcement, usually the release date is extremely fuzzy, and that's because by the time you read the announcement there's probably already been people using a dozen different versions of it and when it gets "released" to you it's mostly just random chance of how the experimentation framework includes you in the treatment group.

What are everyone's thoughts on moving away from the concept of Story Points?