- wat10000Good point, I momentarily forgot about the ability to specify a schema. In that case, you'd have a lot of places where there's only one possible output. Even if you have multiple fields and there are no ordering requirements, typical schemas won't take long to get to a unique field name prefix. If you've output `"us` then `ername"` is likely to be the only valid continuation in many cases.
- Is there anything in the JSON grammer that only allows one valid option? In any case, I also don't understand why it would be costly. The fact that tokens are typically multiple characters would complicate things somewhat, but checking that a given token results in valid partial JSON doesn't seem too hard.
- I thought structured output was done by only allowing tokens that would produce valid output. For their example of a missing closing bracket, the end token wouldn't be allowed, and it would only accept tokens that contain a digit, comma, or closing bracket. I guess that must not be the case, though. Doing that seems like a better way to address this.
- We can simulate a Turing machine, given storage. The infinite storage and infinite time is always a sticking point when comparing any real physical system to a theoretical Turing machine, so we tend to ignore those bits.
- You're missing the point of my rhetorical question. A school can't influence national gun policy. So if the school wants to not let kids have guns, then they have to check for guns.
- The problem isn't the houses themselves, it's the land to put them on. Technology doesn't help much with that.
- That's easy. The challenge would be finding something that has less than the lander, but more than zero.
- It wouldn’t be totally implausible to use that phrase between the wars. The name “the First World War” was used as early as 1920, although not very common.
- Those hardware-level locks are typically not considered because they work quite differently. A standard software mutex can cause other threads to block indefinitely if, for example, the thread holding the mutex gets preempted for a long time. "Lock free" isn't really about the locks, it's about a guarantee that the system makes progress.
In this sense, the hardware locks used for atomic instructions don't really count, because they're implemented such that they can only be held for a brief, well defined time. There's no equivalent to suspending a thread while it holds a lock, causing all other threads to wait for an arbitrary amount of time.
- 10-15 vs 15-20 is definitely comparable in my view, considering how fuzzy the numbers are.
I'd put Hitler much higher, though. That figure must be excluding a lot of war deaths. For non-war deaths, there are 6 million Jews, maybe 3 million non-Jewish Soviet citizens, 3 million non-Jewish Poles, plus a bunch of other groups with smaller numbers. Taking just those big ones, that gets us to 12 million non-war deaths.
But surely we should count at least some of the war dead. Deciding what to attribute to who is very subjective. Starting with the strongest case, 3 million Soviet POWs were killed in captivity, hard to blame anyone else for that. That's up to 15+ million. There's a good case for including Allied military deaths in the European theater, since they were killed by Axis forces. The vast majority of those are the Soviets, which accounts for another 5-8 million (not double-counting the POWs killed). I'd also include Axis military deaths under the general principle that you get credit for what happens when you start a war. That's another 5-6 million. That puts us at 25-37 million.
Then you can get really fuzzy. There's many millions of Soviet citizens who starved due to wartime disruptions, do they count? There's around a million German civilians killed during the war, or died in the immediate aftermath due to the Allies, do they count?
- I'd say it's a subset of fact checking it. You can check facts without doing anything else, but doing something with the knowledge is inherently checking it. If the lecture presents some programming technique, and I implement it, I'll find out pretty quickly if it's wrong.
- Packaging costs money. Gift cards make money. Easy fix.
Stopping the theft you describe is very easy. Don't have actual gift cards just sitting around. Require customers to get them from the cashier at the time of purchase. Have dummy cards on display if you want them to have something to hold, or make them ask.
Of course these solutions aren't free. Adding friction to the purchase process will reduce sales. Retails have clearly concluded, I assume correctly, that it's not worth the cost. Nothing wrong with that.
Don't confuse something being difficult to fix with something not being worth the cost of fixing. We can put a solid upper limit on the impact of fraud by looking at what it would cost to stop it, and conclude that the impact of this sniffing fraud is less than the impact of having cashiers exchange dummy cards for real ones at the time of purchase.
Note that this isn't a "this is easy, they must be idiots not to do this" sort of thing. The current approach is probably the smartest one, given how things currently work. If the incentives changed to make retailers bear more of the cost of fraud (say, legally putting the burden of proof on the retailer to show the card was used legitimately, otherwise they have to refund it if the customer alleges fraud), things would change quickly.
- And yet they continue to sell these cards. Why?
It's simple: they're essentially free money. The worst case for them is that the recipient of the card uses the full amount of the card. In that case, the issuer "only" makes the full profit on those sales. Often they do better: the card is used partially or not at all, then lost or forgotten about.
You can see how lucrative they are by looking at promotions. You can often find deals where you can buy a $100 card for $90, or similar. Why would you sell a dollar for 90 cents? Because you know that on average you're selling quite a bit less than a dollar.
As for the fraud risk... do they even care? When gift cards are used for crime, the issuer doesn't suffer. Maybe they have to deal with upset customers, but that's hardly new. Most of the time, the gift card is bought legitimately, given to criminals, resold, used by the secondary buyer, and the only one who suffers is the unfortunate scam victim who bought it.
It would be so easy to make gift cards more secure. Modern technology can do a lot better than an alphanumeric code under a sticky cover. The fact that they don't bother should tell you everything you need to know about how important fraud is for them.
- China is quite a bit worse. Not having an Apple or Google account in the US would be kind of inconvenient. Not having WeChat Pay or AliPay in China means you can't buy stuff most places. They've ensured that their de-facto-mandatory services are domestic, but they're a lot more mandatory.
I assume the Chinese government is quite happy with this, because they have no trouble bringing their large companies to heel, unlike the US. And centralizing payments like this gives them a great deal of information and control.
- Clicking through the link to the original paper, the point seems to be that qanats are inherently sustainable because they only produce as much as goes in. You may gradually exceed their capacity, but there won’t be a sudden “oops, no more water” crisis as can happen when you pump an aquifer dry.
- A nice thing with LLMs is that you can ask them for a more comprehensive and detailed translation, and explain the nuances and ambiguities rather than trying to match the style of the original. This is great for things like group chats in a foreign language, where it’s full of colloquial expressions, shorthand, and typos.
- How do you propose to not let kids have guns?
- Same coin, other side.
- That’s not the point. This isn’t about a professional association losing some government money. This is about the government’s war on vaccines.
- What nuance? HHS is run by a completely unqualified jackass who is trying to carry out a war against the most effective medical interventions ever devised. There’s no other side to this.