site banner
Advanced search parameters (with examples): "author:quadnarca", "domain:reddit.com", "over18:true"

Showing 12 of 12 results for

domain:natesilver.net

I listen to very little new music overall, so you're probably right: if I listened to more unfiltered Top-40 I would probably hate most of that much more than I hate a lot of modern country. Music essentially gets added to my library from the local college radio station, from my gym friends, from my wife. The only time I'm listening to unfiltered new music is when my father is listening to country radio while we drive somewhere.

Martino goes into this extensively in the book. While a lot of other teams were engaged in sign-stealing using replay rooms that bordered on illegal, but concludes that none were as extensive or as team-supported as the Astros.

I do think the Astros were also disliked for hitting "betray" on Baseball culture in other ways. The Lastros era was the worst MLB example of open tanking, which is a disease on American sport which I truly hope teams adopt the obvious solutions to solve.

Disparate Impact is going to be struck down. The GOP pressing the inclusion political party registration, veteran status and religion (Christian etc) in disparate impact laws is one of the smartest things they’ve done recently (not, admittedly, a long list) since it will accelerate their demise.

But in the long term they’re just not viable. They are pushed because explicit quotes were ruled illegal by SCOTUS, but the more they contradict each other (eg disparate impact against hiring Republicans Va disparate impact against hiring minorities) the more the courts are going to be overloaded with an endless series of these cases and SCOTUS is going to have to act. Even though companies may have legally sound defenses to why their new hires are 70% registered diverse Dems but retirees are, say, largely Christian Republicans (age and politics, changing racial demographics, whatever) the sheer onslaught of cases will become unmanageable.

Nybbler will undoubtedly have some kind of blackpilled spiel about why even this is doomed, but it seems to me that, uh, heightening the contradictions of disparate impact is the surest route to tearing it down.

I think it suits well as a modern sport since it's largely self-directed, supports a wide range of time slots, Instagram friendly and you can take a friend group of differing ability to participate without direct competition.

But that would be a very stupid thing to do.

Without the export bans China would have continued to buy instead of build the most advanced process chips for the foreseeable future. We've forced their hand to bet everything on build.

I think "disparate impact" is a ridiculous standard. The odds that any decision process will return the same results for two groups which differ on all sorts of socio-economic axes seems unlikely.

"Disparate impact" and other "equity" based arguments are farcical on the face. It only ever cuts one way, and the arguments are deployed extremely selectively. They will never, ever, in a million years apply "disparate impact" arguments on which identity groups pay the most taxes or which identity groups are mostly likely to be victims of crime from outside their community. When non-whites outperform whites, they are just better than whites and should be celebrated. When whites outperform non-whites, it's racism and the thumb must be put on the scale.

I don't want AI making hiring decisions, [...] or deciding verdicts in criminal trials. Anything that helps prevent that (even if imperfect and incomplete) is a good thing in my view.

What's better about a person making those decisions? The criteria I can think of (accuracy, speed, interpretability/legibility, compliance with standards) don't always favor human decisionmakers.

(I don't want anyone monitoring what I write on the internet for wrongthink, so I'm with you there)

I'd argue there's a third category of some people who just have absurd physical gifts and don't need to be particularly switched on or with it. NBA Centers tend to have more diverse personalities/interests than other positions since they have to be 7-footers to even qualify so the effective pool of potential NBA Centers is small.

I'm a very large person and got to fringe professional academy levels in Rugby. There were only like 10 people in my region who were effectively credible competition for what ended up being 4 low-paid slots on the development pathway. Meanwhile a normal-sized human position was hundreds of people competing for like 5 spots.

You think it's a coincidence that this is when that "man vs. bear" meme popped up?

(2) "ARTIFICIAL INTELLIGENCE SYSTEM" MEANS ANY MACHINE-BASED SYSTEM THAT, FOR ANY EXPLICIT OR IMPLICIT OBJECTIVE, INFERS FROM THE INPUTS THE SYSTEM RECEIVES HOW TO GENERATE OUTPUTS, INCLUDING CONTENT, DECISIONS, PREDICTIONS, OR RECOMMENDATIONS, THAT CAN INFLUENCE PHYSICAL OR VIRTUAL ENVIRONMENTS.

This is not limited to ML, this bill applies to any computer program.

The regulation is bad because even if you remove direct references to age, color, disability, ethnicity, genetic information, limited proficiency in the English language, national origin, race, religion, reproductive health, sex, veteran status and so on, you will still have disparate impact, either because the AI inferred them from oblique references (Latoya Washington living on MLK Boulevard) or because causes of disparate impact correlate with age, color, disability, ethnicity, genetic information, limited proficiency in the English language, national origin, race, religion, reproductive health, sex, veteran status and so on.

The bill opens the doors to non-stop litigation. When a real person or an expert system lower the credit card limit of Latoya Washington living on MLK Boulevard, they leave behind a trail that shows their chain of reasoning worded in a way that pointedly avoids any references to age, color, disability, ethnicity, genetic information, limited proficiency in the English language, national origin, race, religion, reproductive health, sex, veteran status and so on. Everyone knows this and isn't triggered by obvious disparate impact. When the same bank uses an ML model to do the same thing, there's an obvious way in for a lawsuit: disparate impact? check, AI? check, time to sue, good luck proving that your model didn't lower the credit limit because Latoya was black.

It is poor form - you can note that you know what someone's biases are and that you expect them to have a certain perspective (I have done so myself, because you're not wrong about @coffee_enjoyer) but don't just jump into a conversation to tell someone "Hey, don't waste your time arguing with this guy." It's the same sort of insufferable thing you see everywhere else on the Internet: "Reminder: JK Rowling is a transphobe, Thou Shalt Not Engage with her!" People talking to @coffee_enjoyer can usually figure out for themselves where he's coming from, he doesn't exactly hide it.