@BordedDev a port wiggling is never a good sign.

But if it works, it works.

@BordedDev music, or musak?

think local act local > think global act local

for writing well-tread things, or following instructions on small specific modules, AI is pretty good.

It's like working with a competent junior programming partner.

They'll make some errors, and need some review, but with a little work and clarification, the system usually gets you where you need to be.

People get busy with life.

@Lensflare have everyone thats wants a new account to post an SHA256 hash of a new password.

Then when the new platform launches, they enter the password, and you match the backend hash with the hash of the password.

If accepted, you then prompt them for a new password.

I'm more than happy to give up my old comments and posts if the platform has a long-term viable replacement.

Tell him you used "whitelist" to remind cis-het-normative whites of their whiteness, and the term comes from "viva-la-revolucion comrades" in the 1819 jamaican revolt against british colonialism, a term they used to mark "british oppressors and sympathizers, in the struggle for equity" and that "supremacists and covert right-adjacents have astroturfed and colonized the word, so refusing to use it is unintentional reactionary tactics that give aid and comfort to those who have traditionally oppressed minorities."

or any old line like that.

Just bore him with word salad until he gives in.

Everything I've heard about NVME has been the same: They're unreliable, they fail sooner, they have a shorter lifespan, etc.

they call it a "practice" for a reason. Because you have to remember to do it.

Although, from my experience, "practice" is a synonym for "an oversight".

Library and language warts are the mother of "best practice".

Thats how we got javascript after all.

@jonathands the joys of things being so slow, you start traveling backwards in time.

If it is on the internet, you can be sure some search engine or bot is indexing it.

Whats worse is when it bleeds out of your coworkers headphones and they're the sort that listen to the same song on a loop for hours.

@BordedDev The power of music compels you!

@Lensflare It's the sign-of-life signal.

With all the talk of devrant 'dying', I half expected, coming back, that no one would be here.

Love that its still a sort of low-bandwidth backchannel for talking with people long-term.

And for all the complaints of spam, I still get less spam here than I do in my emails (or notif spam anywhere else for that matter).

When I post here, I know reliably I can get ahold of at least some of the old gang.

And thats all that matters.

Whats been going on on your end Lensflare?

@kobenz "Pretend that common space is called time and do it with stocks and BAM! You've got a risk free portifolio"

Is this actually a thing?

I assume it is done by balancing buys and sells using options? Or have I exceeded my own understanding here?

@Root We'll put the splinter in their eye together.

Stay alive. Every day you stay alive is a splinter in the eye of the soulless parasites who have ground generations into near-poverty.

@figoore hey thanks figoore!

Proofs in the pudding.

@Root you sound exhausted.

@AlgoRythm well something interesting you can do with your test data (the target your evolutionary algorithm is training on) is get the first order derivatives between a parent and its descendant mutants, if a child has a negative derivative then you can eliminate it, and keep the ones with smaller, but positive derivatives then the parent had with its grandparent, essentially enforcing a candidate ranking.

Don't know how much math you're versed in, but 1st order derivatives just tell you how much the output of a function changes instantaneously.

Sort of like 'how far would you travel in an hour if you went 10 mph faster, 1 mph, or even 0.1 hours faster?"

Your distance traveled in some amount of time, like an hour is a function of your speed.

@BugsBuggy don't worry, you'll get there. Marketing is a learning experience that very much resembles a crash course.

@AlgoRythm DR ate my prior comment and fucked up the spacing, but have you tried varying any of the numbers mentioned and looked at the slope of the convergence rates across different configurations?

Also those are mega cool papers Algol. Thanks!

@BugsBuggy I put it through archive.is first. First thing I noted is it wants technical information and doesn't specify what. I'd fill it out but I'm entirely uninterested in providing my ip address.

@AlgoRythm Any sufficiently large population over time, will be a sampling of the full distribution of some parameter sweep over the grid of some distribution.

@AlgoRythm thank you. Going through the first now.

Anyone wanting to read it can find it here:

https://arxiv.org/pdf/0902.1629

First thing that pops out

"The methodology proposed in [1] minimizes an influence of random circum-

stances and different power of the used computers. In particular, the com-

putation is run 100 times for each function of the test set. The number of

successful runs is then taken as the probability of the success (the compu-

tation is considered to be successful if the difference between the best value

found by the algorithm and the theoretical optimum is less than 1% of the

optimum value, or a distance is less than 0.1 if the theoretical optimum is

zero). If 500 generations pass and the optimum is still not reached"

I wonder if these numbers are empirical or if they just eyeballed them during test runs, or if theres some other theoretical reason they chose 1%, 0.1, and 500, etc?

@JsonBoa parameter sweep a grid, just randomize where you start for each parameter, and use some sort of bitmask for the combination of all selected parameters.

Doesn't get you partial activation of course.

Hey algorythm what papers you reading?

@BugsBuggy I'd love a platform that I could bitch about poorly made games on.

Anything steam-stats-esque would be great too.

@Wisecrack edit: after filtering, not before.

though I haven't implemented the full filtering regime either.

Why 7 btw?

@BordedDev RIP.

More meteor showers to come.

I'll make a post on DR when I hear about the new ones.

@AlgoRythm lol, alright, I'll give it a go. Number of candidates before sampling and a bunch of other filters, grows by a factor of about 2.4534 per digit (where 'one' digit here is the addition of both a digit to p, and a digit to q). So if you have 1400 pairs of 3 digits, at 4 digits, after filtering, you have 3,434 pairs, etc.