“I hate reality but it's still the best place to get a good steak.”

― Woody Allen

@hill_hobbit That only happens when we extend our definition of the tribe, which by default only includes the closest to us. We can do it by cultural training, and in civilized worlds it often happens, but it can be easily undone by applying certain ideologies, which unfortunately have a tendency to spread.

@dennis Actually, humans are among the species best adjusted to temperature changes among mammals, if not all animals – good temperature regulation is part of our long-distance running ability. For comparison, take a dog or a cat to a sauna, and you will kill them.

Humans are horrible by default. We can learn to behave acceptably, but there is an effort in that (totally worth it, by the way). Whenever we fall back to the default, no-thinking behaviors (like me right now, being sick), we turn into absolutely despicable creatures. Why does it have to be so hard?

- meta Show more

- meta Show more

@pettter It can doffer from 50% even with completely random answers. I assume anyone who deals with statistical data understands that.

@amikigu Depends on which problem you are talking about. The problem I was hinting at is that your company is missing out on good employees because its hiring practices are biased.

@pettter So if the age of the applicant is a good predictor of whether they are hired or not, globally, across all jobs in the company, then you still have a problem.

@pettter But you don't want to only be hiring older people, even if they are more experienced. Or only young people, for that matter.

Yiss, it got approved! If you can please score my design "Please. Keep talking" on @Threadless! "Passive-aggressive challenge". Or don't. It's fine. No hard feelings, really :<
#mastoart #creativetoots #artistsonmastodon

@Bzdata That can be tricky, because there could be actually correlation with the skill/experience there. Better stick to the obvious biases.

Here's a fun experiment you can do with your HR department's hiring data:

Take all applicants data from a few years back, and remove from it all information that could be possibly relevant to the job at hand. Leave only things like the name, sex, age, address (converted to coordinates, probably).

Now split the dataset in two, and train a NN model on the first half to predict whether the person was hired or not. Test your predictions on the other half.

If you get more than 50% hits, you have a problem.

I gave up and installed new system and recovered the backups on a temporary laptop. It's too big and heavy to lug around, but it's much more convenient than a phone. Hopefully the replacement laptop will arrive soon.

Show more
Mastodon for Tech Folks

This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!

We adhere to an adapted version of the TootCat Code of Conduct and follow the Toot Café list of blocked instances. Ash is the admin and is supported by Fuzzface as a moderator.

Hosting costs are largely covered by our generous supporters on Patreon – thanks for all the help!