The Gaming Den

Something brought up in the "Evolution vs Economics" thread was how shitty virtually all science reporting is. Which is the truth. So I was wondering: where is there any science reporting that isn't bloody horrible?

I can only think of one place, which I found recently: Ars Technica. And unfortunately they aren't a science reporting place, they report mostly on goings-on in the world of computers and the internet and whatnot. But they have a section devoted to science, and most of the time it's really quite excellent.

Anywhere else out there that doesn't suck?

Science News is decent, although they've recently become more like Scientific American. In the mean time, SciAm has become more like Fox News, so their relative goodnesses have stayed unchanged.

I like reading ScienceDaily, but I make no claims to their quality. Even the MIT science newsletter has bad reporting. Either that or some of the best minds in AI have no idea what they're doing.

There are unfortunately alot of nuts that like science news.

-Crissa

As a computer scientist let me tell you this.

Cutting edge AI and VR are sitting right there on your current PC games.

Thats as good as they've managed in those fields. Which is to say little more than cosmetic differences in the last 15 or so years.

Right now "true" AI looks pretty much impossible.

And for all those guys in articles talk about "strides forward" in fields like voice or image recognition. Thats basically a lie.

Image and voice recognition is possible, to the limited extent it works thanks to pretty much computing brute force. But its "dumb" recognition, no understanding.

Its also unreliable recognition and many of the trails "proving" the supposed effectiveness of things like voice or face recognition software are basically falsified and privately funded unscientific lies. Accuracy on these things is like 30% if you are LUCKY.

There is some promising news in the most fundamental aspects of computing technology with new electronic logic components being experimented with. But thats basically in the "hey we might build a better artificial neuron!" territory rather than computer science having any idea what the fuck to do with that once (if) they have it.

AI has done some great shit, but it has the problem that any time something good comes out of it it stops being called AI. Probably because it doesn't fit the public perception of what AI is supposed to be.

The Language Log bitches about science journalism every now and then and they have some good points.

Here's just one of the latest posts on the subject.

But the phd comics image in the entry is pretty spot on.

I'll see if I can find some older posts that are very awesome on the subject.

edit: more annoyance on a specific science subject... If you read some of the early comments, it will link you to the university pr article that seems to be partially to blame.

There's a lot of people who will fall for the weirdest sciencey bullshit and I guess they have money to spend so you might as well write towards them.

According to some research I did, the pentagon forked over a shitload of money towards some weird ass attempt to make a data mining AI thing. I'd laugh at what the program director said they had accomplished, except that the fact this program even exists means that the pentagon's dick got super hard when someone tried to sell them the horrible bastard child of 1984 and Terminator 2.