Math in general is something that you don't need to know until you need to know it, the problem with "just in time" learning is that if you don't already know it then you won't even be able to know when it's needed and which specific bits of math you need to learn for the specific situation you find yourself in.
If you don't know math, you often won't be able to even realize when math is needed.
There's a Steve Yegge quote about this:
> You'll gradually realize that your math deficiencies aren't just something that you might need to beef up on if you ever "need to"; you'll see that virtually every problem space has a mathematical modeling component that you were blissfully unaware of until Done, and Gets Things Smart gal points it out to you and says, "There's an infinitely smarter approach, which by the way I implemented over the weekend." You stare slack-jawed for a little while, and then she says: "Here's a ball. Why don't you go bounce it?"
Yeah, that's true, and I've been in that situation. Unfortunately the math I was missing was not calculus or anything I had exposure to in school, or any actual experience in practical application. You can't know everything.
>Peer reputation and self policing works in a system where no money is at stake.
It works in systems where money is at stake too. You need a high trust, conscientous society. Doesn't mean that having one solves the issue entirely. The ocassional bad apple will fraud there too, but it wont be rampant enough to be a systemic problem.
>When your career and income are at stake, why would you take the risk of opposing Dan Ariely? Only someone who was already rich could do that.
You don't need independently wealthy scientists to expose fraud.
You just need an environment where if regular salary dependent scientists expose fraud, they're not punished for it.
If you don't have that, sure, then you need independently wealthy scientists to dare.
Lightsail also does not offer any form of virtual serial console to access a VM where you've accidentally broken networking or ssh, when even the crappiest lowendbox-tier VM provider generally does. The only way to 'recover' a lightsail VM that's in that state is to export the storage device to full amazon services starting from a $25/mo VM.
Well yeah, obviously. I think the point was that uptime is not the only concern, and that a service sufficiently attractive in other ways can get people to overlook a lack of reliability.
> Kant goes one level further. He says, basically: "before you do something, ask yourself: if everyone else in the world did what you're about to do as a matter of course, will it make the world better or worse? If it makes the world worse, then don't do it".
This is not the Categorical Imperative. What you're describing is more like Rule Utilitarianism. Kantian ethics is deontological, not consequentialist.
Hence my qualification at the start. My problem with deontological ethics is that the justification of it abruptly hits an impenetrable wall, i.e. the rule or the duty, unless you appeal to some benefit or prevention of harm to rational beings. Even in the Metaphysics of Morals, the whole concept of duty would be arbitrary if not in reference to rational beings. But happy to be corrected.
> Even in the Metaphysics of Morals, the whole concept of duty would be arbitrary if not in reference to rational beings.
In what sense? There's a sense in which there's no moral law without rational agents, but Kant himself is explicit about the "good will" being entirely independent of intended or obtained ends. In other words, no rational agent needs to benefit (or be harmed) in Kant's system for individuals to be morally blameworthy or praiseworthy.
(See GMS 4:394 for the "jewel" analogy, which is most popularly cited as evidence for Kant's position on this.)
I understand that this is the common (and perhaps the prevailing) interpretation, but my question has always been: once you subtract action potential and expected ends, what remains of the concept of "good"? We cannot simply end the inquiry by saying "good will" because it still doesn't clarify what it is that one wills. Disclaimer: not a Kantian scholar.
I think the very short version of Kant’s view is that “good” is not impinged upon by any particular desire or action. What makes the good will “good” is that it operates in conformity with duty, and duty in turn is derived from reason (and reason is a property of our ability to self-legislate, i.e. give ourselves law without engaging in casuistry.)
It's been twenty years since I read Kant, but it differs from utilitarianism in that it has no concern for utility of others, rather it's "if I could pick rules for everybody to follow, those are the rules I should follow. Ergo if I never want anybody to lie to me, then it is always immoral for me to lie -- anything else would be a contradiction, since it assumes two rational agents coming to different conclusions.
A good point, but (a) this merely sounds like a modification of the golden rule: do unto all others, without exception, as you would have done unto you, and (b) the rule isn't an end in itself because the end lies behind the "should" (e.g. why should someone not lie to me).
I am by no means an apologist for Kant, but his ethics are clearly deontological, not consequentialist. He categorically rejected applying empiracism to ethics. Again, my memory is fuzzy, but I believe that Kant further argues that ethical rules are fixed, since any rational actor properly reasoning from first principles would eventually land on the same set of rules.
It should not be surprising that contemporary ethical systems yield similar results in everyday situations because nobody will get published with their ethical theory of "why you should murder at least one innocent person per day."
It's not about what you actually want in this case, but what you could want. It's impossible to want people to betray your trust, because if you wanted it then it wouldn't be betraying your trust.
So that type of lying is strictly forbidden. You do not want to be deceived. But if include in "lies" such things as reading fiction aloud, then that can be permissible as you can want that: you aren't really deceived or betrayed.
The Taliban was actually very rapidly pushed out of Afghanistan by the US working together with the Northern Alliance. The real problem was that the US couldn't reach into Pakistan to fully end the Taliban for good. Thus the Taliban was allowed to rebuild and reconstitute its forces over the years waiting for the US to finally pull out of Afghanistan.
But if they had really been a superpower they would have had the diplomatic ties and military strength to do that. I mean Pakistan was harboring Bin Laden himself no less.
However I don't agree, the Taliban was never really eradicated within Afghanistan either. That's why they came back so quickly. In fact even in provinces far from the border they came back every night. Our forces would build a water well for the local citizens, and the same night the Taliban would come and blow it up and rape half the village. They didn't withdraw into Pakistan every night.
What does that have to do with LLMs hosted by Anthropic. They didn't say the Dept of War couldn't make kill bots. They said no using Claude to make kill decisions. A quadcopter drone running on a Raspberry Pi isn't running an Anthropic model locally.
Fiber optic drones were last year's weapon of choice, this year it is drones with autonomous terminal phase.
This happened because the AI data center build-out increased demand for fiber optics, driving up the price to less affordable levels.
Autonomous terminal phase is also more effective; ~80% mission success compared to ~30% for fiber optic drone.
Fiber optic drones are real but the fully autonomous drones others are describing are also real (but less publicized due to the bad PR around killbots)
My counter argument is that you only need 1% of the population to be bad apples in order to destroy a high trust society. Hell, even a single serial killer on the loose will make a lot of people too afraid to walk alone at night. One or two burglars can be responsible for 50% of the burglaries in a town. And so in and so forth.
The other problem you have is that a lot of crime tends to be financially rewarding, so if left unchecked it will tend to grow over time as the criminals get richer and expand their operations.
For all those reasons, you need surveillance to catch criminals and restore trust even if 99% of the population are good people.
The thing is I think if you give people the choice between a 0.001% chance of getting murdered on the street or a 100% chance of being surveilled 24/7 you will find that people will take their chances with the murderers
0.001% chance of being murdered is 1 in 100000 which is about the murder rate of the UK. That does not seem to stop the UK from wanting more surveillance.
The problem with murder rate is it doesn't matter what the actual numbers are. Just bombard people with bad news 24/7 and they will believe the sky is falling and murderous mobs are roaming the streets.
I think people are for surveillance under the presumption of it catching major criminals. But once it's introduced, they realize its main purpose isn't to nab murderers, but to punish them (the people who voted for it).
Surveillance everywhere means you can't park in a no parking zone for 3 minutes just to drop something off real quick. You'll get a ticket mailed to your door. You can't go 5 mph over the limit when nobody is around. You can't slow down at a stop sign without stopping for a full 10 seconds without risking being busted. You can't sneakily sample a grape before deciding whether they taste good enough to buy. You can't not tip a bad restaurant without them plastering your picture on the walls to embarrass you. You can't say a mean thing to an annoying person online in the heat of the moment. Half the people on earth cheat on their spouses while pretending they never would and anyone who does should be imprisoned, and soon they'll have their secret revealed and regret their words. You can't trip and fall in a goofy way without your video being a global meme for a whole month.
All these pop into people's heads and make them think "screw that" after at first thinking it's great.
Unfortunately evidence suggests that people want the opposite, or at least surveillance for everyone else. And not just over murder, over petty crime. Nextdoor is full of people who want to be a surveillance network of humans.
I'm not sure we know what people want though. After terrorist attacks politicians seem to decide what people want, tell people what they want, but rarely, if ever, offer a referendum on the subject. And on this alone they also get bipartisian support because nobody dares look weak on terrorism, so there's no possible vote against it.
Reality shows otherwise. Take a look at Google and free-for-ads-and-surveillance everything. People are willing to make compromises on privacy for less important things than the change of them or their loved ones getting murdered, as mentioned as an example.
But back to Flock: public cameras are not 24/7 surveillance, as those don't surveil people at home.
ps: I'm not for a panopticon, enforcement needs to come with surveillance also, otherwise it will just be an anarcho-tyranny. So common in the developed world. And enforcement of law must also apply to companies, even (and especially) to vendors of the law-enforcement.
The main problem with these arguments, which are common and we also see in a lot of free-market discussions, is that they assume that every person has perfect knowledge all the time. Including knowledge of changes that might have just occurred.
That's... an extremely difficult claim to make, and one that people largely make for free, unchallenged. I'll challenge it: most people have absolutely no idea what's going on.
When you sign a terms of service, do you read it in full? When you get what feels like a weekly email saying the TOS has changed, do you read it in full? And, to be clear, you auto-consent when the terms of service changes, they don't pretend you have to read it.
This is by design. It's not a coincidence the TOS is extremely long, extremely vague, and constantly changing. And, it doesn't matter anyway. There are no competitors. Because guess what - they have the same TOS. And the TOS doesn't mean anything, anyway. Because the majority of the rights they explain they can exercise in the TOS, they never do. So you can't even use that as a baseline for "what is this company doing".
So not only are people not knowledgeable in practice, but they also face many barriers, by design, to being knowledgeable. And, one further, they actually CAN'T be knowledge even if they dedicate their entire lives to it, because the knowledge does not exist and is not available.
People do not "want" the surveillance state. Rather, they submit to it, under lies, omissions, and asymmetric power. This includes me and you.
aye, the average joe has no idea wtf data mining is, or how its risky for them
my educated, medical professional in-laws have some vague idea about it but love their instagram and lose chasing the related trends. they have no real, direct understanding as to how this could impact their privacy, or how marketing works, or how propaganda is similar to marketing, etc.
So the argument here is that people are too stupid to realize that they are being manipulated into being controlled, therefore it’s ok?
Sit someone down, a normal average person, and explain to them the actual implications and then do you think they will agree? Reality shows that when people actually understand what surveillance means they reject it wholeheartedly
No, they don't reject it. The realize they hate it, but make a compromise.. and use the free gmail, the ad ridden internet, etc., because they have no real alternatives. Rejecting surveillance capitalism is only realistic for the poorest and the richest. The average people must accept it.
It is not about percentage of the population, but about whether the bad actor rise to the top.
> The other problem you have is that a lot of crime tends to be financially rewarding, so if left unchecked it will tend to grow over time as the criminals get richer and expand their operations.
It already happened. The president, Miller, Karp, Thiel, Zuckerberg, Musk, hell the company named Flock are all bad actors who got richer over time and expanded their operations.
High-trust societies are based in transparency. Ironically you get high-trust by actually not trusting people. We can have surveillance and transparency at the same time... the point is that people actually need to be accountable.
It's the lack of accountability, especially on the meta-level where the people that need to be held accountable know that there is no accountability, that can make surveillance dangerous.
We need both police, and we need the police to have police, and at every level, there ought to be an antagonistic relationship.
Just deciding, fuck it, no transparency and no surveillance is just going to lead to organized crime and cronyism... and surprise, surprise, that's exactly what we're getting in America.
> My counter argument is that you only need 1% of the population to be bad apples in order to destroy a high trust society.
Counter-counter argument, it doesn’t actually matter what the proportion is, rather the perception of it. The likelihood of non-white collar crime affecting your life has dropped dramatically in all of our lifetimes. The chance of being affected by white collar crime has skyrocketed - who hasn’t been solicited for a class action lawsuit?
Which leads me to my hotter take: America isn’t turning into a low trust society because of crime and brown people (white replacement theory), it’s turning into a low trust society because of white collar crime, grifting, private equity takeovers, naked corruption, prosperity gospel, fuck-you got-mine, exurban development, oversized car culture, etc etc etc.
Edit: furthermore, it’s in these psychopathic fuckrats best interest that people believe that there is a “1% bad apple” phenomenon to justify their domestic security panopticon, and they would be laughing all the way to the bank that there are so many suckers who buy into their propaganda if they weren’t so high from huffing their own ketamine, cocaine, adderall, and HGH-laced farts that they believe their own bull shit.
It’s like blockchain but for racism, and it’s genius.
On-point surveillance by people would be a good start; I do not see why it needs to start with endless surveillance by machines run by a private company.
I see no problem with every private citizen having a video camera that records everything around their home or business, and when a crime happens the police can solicit the recordings.
Also it is fine if such a private surveillance system alerts the police when a crime is in progress.
What I find completely unacceptable is to have such ubiquitous surveillance means that are networked and available to the police or other such institutions for real-time surveillance of whatever they choose. It is absolutely certain that whoever will have access to the centralized control will abuse of this power.
I have grown up in a country occupied by foreign invaders, who have installed in power a government of their choice, who have robbed and murdered whomever they wanted, and then for almost a half of century nobody could do anything against them, because everybody was under surveillance, so that no opposition could be organized.
Americans do not really understand how frustrating it is to live in this way and they fail to recognize how quickly their country becomes more and more similar to the former "communist" countries.
> For all those reasons, you need surveillance to catch criminals and restore trust even if 99% of the population are good people.
Hard disagree. Everyone knows who the bad apples are, but the policies are made so that they get a slap on the wrist and the get sent back out on the streets.
If you don't know math, you often won't be able to even realize when math is needed.
There's a Steve Yegge quote about this:
> You'll gradually realize that your math deficiencies aren't just something that you might need to beef up on if you ever "need to"; you'll see that virtually every problem space has a mathematical modeling component that you were blissfully unaware of until Done, and Gets Things Smart gal points it out to you and says, "There's an infinitely smarter approach, which by the way I implemented over the weekend." You stare slack-jawed for a little while, and then she says: "Here's a ball. Why don't you go bounce it?"
reply