I think predictions are hard, especially about the future...

Yeah, that's much harder than predictions about the past.

Sorry, $80m. My maths was off.

From the figures in TFS it appears that only $70m of the $100m is accounted for. Where did the rest of the prize money go? Was it sequestered in basalt, perhaps?

Yes, including the environmental costs of mining and transporting said rock.

Why not use both? That way we've got some built in redundancy when one of them goes rogue. (And at this point in time, it's a coin flip as to which one it would be...)

A branch is just a reference pointing to a commit. Deleting the branch deletes the reference, not the commit. Eventually, unreferenced commits will be permanently deleted, but not immediately.

I wouldn't suggest changing that.

However, knowing that so-and-so deleted the reference to commit such-and-such at a particular time is incredibly useful, even if all you have is the hash of the now-deleted commit.

It would be even better if you were able to add commit messages to branch deletion. If someone wrote a message like "Deleting the 'sort_optimisation' branch as I've realised that this approach isn't going to work. We can't sort in place as the calling code may require the original ordering to be retained in the local reference." rather than simply disappearing the branch, that would be a useful bit of communication to the rest of the team and might save some wasted effort from someone attempting a similar misguided solution in the future.

I don't really see how you misunderstanding my point about you misunderstanding other people's points proves your point about how you know better than everyone else what they need, but whatever.

Actually, taken at face value, this study is a strong argument for multi-layered interfaces, or possibly even multiple interfaces, for a given application, with different target audiences. You can see it already in things like settings panels, where an 'advanced' mode adds a lot of extra settings that most people won't understand or be interested in, but where what we used to call 'power users' can tweak the things that matter to them. They are somewhat primitive at the moment, when they exist at all, but I'd love it if this study prompted a move in that direction. It won't. But I can dream.

If you delete a file, it's gone. Oh no, wait, we track it in the history.

Why are branches any different?

People often want stupid things. It says more about them than about the rest of the world.

People who dismiss other people's needs as stupid should probably take a long hard look in the stupidity mirror.

Who's going to write the other 5% and how do you plan on training their replacements, once they retire?

Prepare for immediate de-resolution.

"When we were first married this was a problem."

I bet! There's a whole slew of new and specific vocabulary that comes into play on your wedding night!

And tomorrow... "Man Survives With Titanium Heart For 101 Days - another World First".

One thing to consider is that there are two types of novel ideas. The first are those where you come up with something new that you've never heard of or come across before. The second is where you come up with something new that *no-one* has ever come up with before. A lot of our individual discoveries and novel ideas fall into that first category without falling into the second. These are totally the kind of ideas that AI can come up with and is the reason AI is impressive; it comes up with ideas that seem novel to you. With the amount of training data they have ingested this is quite a probable outcome as they have so much more information to draw on than one person's limited experience. However, whilst AI can come up with completely novel ideas by smooshing stuff together, it has no way of knowing which ideas are actually useful/worthwhile/interesting vs. those which are just hallucinations or AI slop.