Re: Proposed HTTP field name registry updates - feedback solicited

Hope I can be granted a touch of leeway bumping this, pertains to an uber-issue brought up by the "Draft for Resumable Uploads" thread.

Initial Data/Lore rock-paper-scissors code was random, never hung.

Then I taught Data to play not to lose, preferring a defensive "paper" strategy; and taught Lore to play to win preferring an offensive "scissors" strategy.

I don't know if the results mean I'm really good, or really awful, at AI coding. Ha!

But invariably, once Lore goes up by 2, Data and Lore just throw rocks at each other until Data's AA batteries die because neither one is in any hurry to win, or not lose, whichever the case may be. Making "max score" totally irrelevant. I wonder if I shoulda seen that coming?

I *could* impose some artifice on the AI's to break such a stalemate (the easy way). But I'm pursuing a strategy of teaching them *not* to stalemate (the hard way). As in, Lore has to reach the winning score before Data's batteries die, so it's up to Lore to take the initiative. Now I just need to figure out how to code "initiative".

That's what I call an "organic" solution, and that's how I think HTTP development should proceed. The protocol allows any number of problems to be solved with artifice. I would rather try to use what's already there vs. working around what isn't, even though that's the hard way.

Otherwise, we're just chucking rocks at each other until the cows come home. :)


Received on Wednesday, 6 April 2022 05:47:56 UTC