I thought the concern that justified shutting down inaccurate posts about vaccines was that it would *spread* inaccurate information, i.e., that people who didn't know any better might see such posts and suddenly start worrying that vaccines could cause harm. If they are speaking in code then it sounds like that's not an issue.

It sounds more like people an enraged that the 'bad people' aren't being punished and, at least in theory, that wasn't the justification for banning such speech.

That's not a good argument. Sure, ultimately the CIA cares about acquiring information and they'll make the calls that best achieve that end. However, you could say the same thing about the military: they'll make the calls that best achieve their strategic objectives not what's best for the individual soldier.

And yet, the US military puts great stock in their commitment to No Man Left Behind. Not because the military brass are letting their emotions get the better of their strategic sense. Rather, because, in the long run the fact that our military personal know they can count on the military not to abandon them allows them to be a much more effective fighting force. The lack of such confidence in the Russian forces is part of why setbacks turn into routs.

Similarly, the CIA knows damn well that their ability to cultivate new assets depends on the perception that they'll be taken care of and won't be exposed. Sure, there are more cases where the CIA can get away with screwing over an asset without it becoming public but that's not what's going on here. This is just incompetence.

The linked article only includes a very superficial summary. It's based on the reuter's report which the article claims to link but that link is broken. Here is the link to the report.

https://www.reuters.com/invest...

I was originally a bit skeptical since there is always the possibility of delibrate security failures to leak misinformation but based on the details in the report it seems like the CIA was just stupid. For instance:

We fucking have a giant agency which is supposed to have expertise in this kind of thing, the NSA (not to mention the military cybercommand). Are our agencies so siloed or subject to petty rivalry the CIA couldn't just ask for expert assistance?

The suicide is tragic and I don't doubt that media can have a causal role in pushing someone into despair and depression.

However, we seem to react very differently depending on the type of media. I have no doubt that there are people who enter depressive spirals reading about horrible tragedies in the newspaper everyday or consuming books from the library about horrible events like the holocaust. Yet, when such tragedies occur, we don't lay the blame at the feed of the library or newspaper even though it might be true that in an earlier age they wouldn't have had to deal with that kind of troubling content. Hell, the content doesn't even need to be troubling, novels, (and especially air-conditioning) make it more appealing to be socially reclusive.

We tend to treat whatever kind of media that's seen as new or scary differently than established parts of our media eco-system. When novels, newspapers or (now) TV and comic books play a role in someone's depression we tend to regard them as just the background of modern life. But we saw concerns about comic books when they were new, gangster rap when it was new etc.. etc.. and now we are seeing concerns about social media.

True, social media is a bigger change than a new kind of edgy music but people (really!!) warned about the harms of the written word and widespread literacy as well. And yes, we may have to learn how best to integrate social media into our lives just like we learned with all sorts of other media innovations but it never really helps when we treat the latest media innovation so differently than our existing media in terms of scare mongering and assigning blame.

The problem with all the concern over the dangers and bias in technologies like facial recognition is that the technology it is replacing is even worse. People are incredibly biased and error prone when trying to recognize faces and there is plenty of research about the problems with cross racial IDs and the problems that result from unblinded identification.

At least using facial recognition means we can get a much better handle on any bias/limits the system has and moderate our usage accordingly and improve the technology. We can't patch the human visual system and, while we know it's subject to bias/error, it's much harder to understand exactly when and how that error creeps in.

Having said that, I understand the concern about using facial recognition as a broad screen (e.g. running it against everyone who enters a stadium or subway). However, there are better ways to achieve that goal than this broad brush. For instance, require that law enforcement fill out a paper form, perform a captcha or other small task that takes a minute or two each time they wish to run a picture against the database.

It may not be a perfect protection but facial recognition is only going to get better over time while human beings will stay just as biased and error prone as ever and I fear the effects of this bill in the long term will be to keep us from benefiting from the improvements in fairness and accuracy facial recognition can offer.

This idea misunderstands human psychology at a very deep level. Yes, driving at over .08 is a bad thing but it's far from risk free to drive at .07. You start putting a little device in cars that tells you if you are over the legal limit how many people who now play it safe and won't drive after more than 1 drink start offloading that question to the car?

Never had them recced to me once...well I vaguely remember once way back I think I saw mrbeast a single time. Get friends with better taste.

So I get that it's about ad sales and totally see reasons not to strongly avoid disliked subjects given risks of mistakes etc (and could care less about it...I'll just not watch the vids I don't like) but why isnt the algorithm way better at positively recommending me videos? I stop watching all the time bc of this frustration.

I know it's possible to do alot better using modern ml methods so why don't they? Why do so few signals seem to matter (eg no apparent negative effect of presenting and getting ignored many times or having been already watched). More generally it feels very unpersonalized. It plays videos based on my recent videos but seems to learn nothing about me or my watching patterns.

WTF, I actively give you as much of my data as I can so you can better predict what I want so why are u making such a hash of it here? Are the content producers kicking up too much of a fuss...well fuck em, lots of them aren't that great but just happen to have some super narrow niche that works well with such a simple seeming rec engine!

A big issue here is that not only do ppl not understand (and, reasonably, most ppl will never want to waste the time to learn) the rec system but there are also lots of accidental votes. Oops, forgot I was watching on my wife's laptop. Ppl's kids hit them and I'm clicking things by accident on YouTube all the time.

Given that, the big problem with highly sticky recs is that lots of ppl will end up being locked out of content they would like to watch with no way to realize or fix the problem. You can go through a more involved process to say the content is offensive.

--

It feels like calling this a "digital dollar" has confused many people. The real meat of the proposal is a shift from only allowing large banks to have deposits with the fed to allowing everyday citizens to have accounts with the fed. In an ideal implementation, this would mean that everyone in the US had access to a free account with the federal reserve solving the problems of unbanked individuals, scammy overdraft fees etc.. etc..

I mean why should you have to put your money in some random bank when the federal reserve lets big banks directly deposit with it? And this has the advantage of letting everyday citizens take advantage of the lower risk (yes FDIC insurance is nice but a bank failure can tie up your money for quite awhile and it only protects 250k) and convenience that the fed already offers to large banks.

Of course, I wouldn't be surprised if the final proposal ends up only allowing citizens to have accounts indirectly (e.g. the fed has your money but you have to access it via bofa, citibank etc..). But it's still an improvement over the situation today.

---

As an aside, it would be an interesting reversal for the fed. There's a bunch of economists and investors who have been trying to launch a bank for some time that simply takes all customer deposits and sticks them in the bank's fed account. The fed has thrown up roadblocks and kept them from doing this.

This has nothing to do with the proposal. It's about letting people have bank accounts *directly* with the fed (well maybe intermediated by a bank but you would have a direct claim on money in a fed account) rather than merely FDIC insured. Basically, it lets regular people take advantage of the lower risks involved with having your account balance backed directly by the federal reserve.

Yah, I really hate that cash stuff. I wish we were still in the era where you had to use notes from private banks and worry about whether the notes you had/accepted would be honored and businesses had to guess at whether some note made by a minor bank in some backwater was genuine or forgery despite having never seen it before.

The government may be bad at lots of stuff, but you have to admit that providing physical currency is a pretty strong competency of governments. We'll have to see about digital currency.

It's called Fedbow and it's supposed to debut in 2023 (as mentioned above it's not actually instance ACH but it's an alternate settlement service that works instantly so you can send money to any bank account in real time).

https://www.digitaltransaction...
https://www.frbservices.org/fi...

Actually, the fed is working on this. Technically it's not speeding up ACH but they have a program to offer essentially instantaneous settlements services which would basically create a universal version of venmo/paypal. I don't remember the exact name but it's on the fed website if you dig under their settlement programs. I think it's supposed to become live very soon (if it hasn't already) but I'm not sure if bank participation is mandatory or optional.

This article gives a misleading sense of what's going on in this lawsuit. It's not just an argument that anyone who provides a service that's used for copyright infringement thereby becomes liable for the infringement. That's absurd.

Link below is a better summary that explains that the pleading alleges that these ISPs haven't fulfilled the conditions necessary for the DMCA safe harbor (because they only count violation notices on a per month basis). That's not an obviously frivolous argument.

Now, if they don't qualify for the safeharbor that doesn't automatically make them liable for infringement. However, the argument is that *by engaging in some degree of content moderation* (e.g. booting users who post certain kinds of content etc..) they no longer qualify as a mere conduit for the information the way the phone company does (they don't cancel your phone account if you post certain material). When it comes to websites that's a tougher question but I suspect that SCOTUS would eventually rule that merely cancelling accounts upon reports they were used for certain kinds of bad behavior doesn't make one the publisher of that content in the way that choosing which letters to the editor to publish in a newspaper does.

However, that's a long expensive legal slog and involves a fair bit of risk so the ISPs will want to avoid it and keep the cheap safeharbor defense.

https://deadline.com/2022/09/v...