I think it was a joke. It sounds like you read it as append-only, like most LSM tree databases (not rewriting files in the course of write operations), but I think GP meant it as write-only to the exclusion of reads, roughly equivalent to `echo $data > /dev/null`
Plants build three-carbon sugars during photosynthesis by fixing a CO2 molecule onto a two-carbon chain with an enzyme called RuBisCO. In a typical "C3" plant, this happens relatively directly. But RuBisCO can screw up and fix an O2 molecule instead, and the erroneous result costs the plant energy to repair.
As the temperature rises, so does the error rate. At a high-enough temperature, the plant loses energy overall, which it can't survive long term.
C4 plants separate this process into two steps spatially. They build a four-carbon molecule in a much less error-prone way, then move this to a part of the cell where it's broken down into CO2. RuBisCO is again used to build the three-carbon sugars, but because the relative concentration of CO2 to O2 is so high, the error rate is low. There's some additional overhead to this process, but it pays off in warm climates.
Incidentally, there's another warm-climate metabolism: CAM (crassulacean acid metabolism). CAM works by temporally separating parts of the process. At night, they open their stomata, and use CO2 to build an acid. During the day, they close their stomata, cleave CO2 off of the acid to increase the concentration, and let RuBisCO its thing.
I believe RuBisCO is the most common enzyme on Earth by weight. I find it striking that Mother Nature has had to find all these hacks to get around its shortcomings, but hasn't found a way to simply fix the enzyme so it doesn't make so many errors.
I’m not a huge fan of Uber’s corporate policies in general, but help me understand what’s wrong with this. Isn’t this what any company would do: maximize revenue from customers while minimizing expenses to their suppliers? Most businesses don’t tells us how they do this.
My grocer sells me a can of beans at some price. I have no idea how they arrived at that price, how much they paid their wholesaler, or that they may have a sale on beans next week. I buy or don’t buy beans based on whether I feel they’re worth the cost. And whether I feel like beans.
If I understand TFA correctly, what is going on, is that the grocer is going
* This person is wearing a suit, I'm going to charge double
* This is a regular that always buys the same thing every week, I can charge 30% more without breaking his routine
* This one is buying the ingredients for a recipe to do tonight, I can charge double more on one product because she won't want to go to another grocer just for one missing item.
Or in economic terms it is doing price discrimination to turn the consumer surplus into profit for itself. I think it's obvious why consumers wouldn't like that. Although they can also do "this one is a cheapstake with lots of free time, I have to offer a 20% discount to keep him coming"
"* This is a regular that always buys the same thing every week, I can charge 30% more without breaking his routine"
This one is getting to me more and more. When I grew up, you got the best deals as a regular customer. Nowadays it's the opposite. Loyalty is something that can be exploited. If you don't switch insurance regularly, you are paying way too much. If you stay at a job for longer, you get paid under market. If you use a service regularly, you get charged more.
I think it's really corroding society when loyalty and trust are viewed as an exploitable weakness.
I don't think its as much "corroding society" as it is a symptom of scale and corporate consolidation.
The mom and pop store making $200k/yr revenue selling physical goods gives regulars a deal because if their $100/mo spend goes elsewhere it is a demonstrably negative business impact, and there are 2 or 3 other mom and pop stores close by. The owner's kids also probably go to school with the regular's kids or there's some similar relationship. Plus, even if they wanted to maximize profit - they probably don't based on the previous point - they simply don't have enough data points to figure out how exactly to do it.
The multinational corporation has billions or tens of billions in revenue, tens or hundreds of thousands of locations, and so many data points they can test any price discrimination scheme they want and get data almost instantly. But this by itself isn't a problem (IMO, certainly some here would disagree). The root of the problem IMO is that there's no relationship. The manager of the location has no power, the district manager is in a city an hour away, the people making these pricing decisions live in another state or country, and they are judged on making the graph go up and to the right. Their bonuses depend on it, their livelihood depends on it.
In the mom and pop scenario, the customer interacts with someone whose livelihood depends on them being happy and has the power to make sure they are. With the multinational, they interact with someone who doesn't give a shit if they're happy or not, and has no power whatsoever, and the lowest person with power doesn't have any idea who they are and is actively incentivized to screw them.
Next: Cheapskate as a Service. Are you a cheapskate that can get discounts? Sign up on our app and use your "he gets great discounts" status to buy things for others, and earn a percentage of the savings!
We literally target advertisements in lookalike audiences based on behavioral profiling from on-site and in-store actions. It's been like this for more than a decade in my job; it's literally what I do for a living. But this has been happening with RFM direct mail for DECADES.
One major clothing retailer I worked with for 4+ years was big on promotions, particularly via direct mail. Those scratch-off coupons were not random; they were specifically targeted to individual customers based on their prior behavior.
For example:
- Customer A historically buys when they get a 30% coupon; test them with a 20% and see if they bite, but if not, give them the 30% and get some money.
- Customer B will always buy with almost ANY discount but never without one; give them the lowest one most of the time, but occasionally give them a higher one because it keeps them interested.
- Customer C will buy any time, but we need to keep them hooked so we send them the lowest one all the time.
---
I just don't know why people are surprised that this is happening at a bigger scale in app; it's baffling to me.
Versions of this already exist with "this one lives in a low cost country, buys Steam games, Netflix subscriptions, (...) and shares them with family in high-cost country"
The problem is most people don't know. In my country you can't change the price for a given person. So if you don't know it's done, you can't change your behaviour (like do legal actions).
This is not the law in all places sadly. I read that restaurants in Japan give a different, cheaper menu to locals and more expensive menus to tourists. Most tourists don’t know and the restaurant doesn’t want to price out the locals.
Costa Rica the vast majority of tourist attractions have a resident price and a tourist price. I have mixed feelings about it -- on the one hand, it makes sense for a country reliant on tourism to charge tourists more, since tourists are much richer, and a lot of the money goes to ecological protection, research, the arts, etc. On the other hand, it's kind of a bummer for a nominally cheap country to have quite expensive museum and national park admissions - it's hard not to feel like you're getting screwed, and it's not an ignorable difference for my budget.
It's an interesting dilemma. Personally, I prefer the version of price discrimination where you introduce high-margin premium value-adds that people can opt in or out of - alcohol or steak/lobster at a restaurant, rooms with views or additional packages at hotels, table service, etc, which can allow wealthier customers to subsidize less wealthy ones without necessarily compromising the core service. Though that's still a bummer when adding a view to a room is prohibitively expensive for something that cost the hotel nothing more to provide, and you feel like either you're getting screwed or you'll always have an alleyway view from your hotel.
That just sounds like racism though - how do they tell who qualifies for the ‘local discount’ versus the ‘tourist premium?’
Betting it’s based on stereotypical appearance and language, not checking IDs.
A more charitable approach might be to charge an extra fee for foreign credit cards - that way you get to effectively upcharge tourists, while encouraging conversion to and payment by local currency, additionally saving yourself transaction fees in the process.
Japan specifically is extremely xenophobic, they actively discourage people from immigrating, they do not want foreigners there except as tourists. If you are not born in Japan, you can never become Japanese as far as the locals and the government are concerned.
It's kind of reductionist to take that environment and that culture and just default to "giving the white guy the expensive menu is racism." You'd have to do that to probably 1500 people before you hit the person who has actually immigrated as opposed to a tourist.
> Presumably people are fine with getting "ripped off" by Uber, otherwise they wouldn't keep using Uber and paying for it.
They're fine with it because of the lack of real alternatives. There's effectively a duopoly with Lyft in most cities. Duopolies usually don't present every customer with at least some sort of solution that allows for both parties in the business transaction to gain value.
Which is why taxis were a regulated industry before some cokeheads in SV decided things needed to be "disrupted".
> Presumably people are fine with getting "ripped off" by Uber, otherwise they wouldn't keep using Uber and paying for it.
I’m not sure why we should presume people are “fine” with this just because it’s something that happens.
Plenty of things happen in this world that are not “fine” and make people upset, but continue to happen because of market forces, lack of reasonable alternatives, something being the “least bad” option, etc.
I think one of the most glaring issues is that Uber has established dominance in the category, which gives them power of their users and allows them to implement pricing strategies that are user-hostile with less chance of repercussions.
Right, people keep using Uber because it solves a real problem for them and provides convenience in their life.
If someone creates the same or better service at a more reasonable price, the consumers will switch. There is no vendor lock-in for Uber and no monopoly.
I don't see anything illegal going on here, just good old business.
Anyone can use coupons. Even if they don't want to spend the time to do it, they could. Same with store brand products made by the name brand manufacturer the choice is up to the consumer.
Uber's price discrimination is opaque. Even if they aren't doing dastardly things with it, people don't like feeling ripped off. We have no way of knowing when we are.
But most big box stores have moved to digital coupons that are indeed customized based on their creepy individualized spy dossier on you. At our grocery store, my partner and I get different coupons or even different deals for the same items.
Actually I may have a theory on why this is so annoying. Most people in Western countries are used to not haggling, and hate dealing with haggling for the two B2C products that are frequently haggled for, cars and housing, and frequently come out of the transaction feeling ripped off. And now Uber is trying to apply a faceless unilateral "take it or leave it" haggling process.
Yeah sure, but Joel Spolsky does point in this article how it pisses consumers off. And I think this sort of algorithm "pure" price discrimination based on mind reading how much the consumer is willing to pay is much more annoying than doing it by having cheaper and more expensive options.
Challenge: Implement price discrimination in a setting where all prices are published on the shelf under the product without knowledge of who is viewing the price.
Possibility: Develop a "loyalty" program where customers provide a member id like a phone number at checkout. Anonymous shoppers get the worst pricing. Everybody else voluntarily signs up for price discrimination disguised as rewards.
Also the low-tech version: farmer's markets, bakeries, anything with perishables they sell daily will often give steep discounts or freebies if you show up close to closing, rewarding those that are in-the-know or are willing to modify their schedule to save some cash.
I’d agree - but is it? Where I live (not a top-ten US metro area) we also have Lyft and a number of traditional cab companies. Uber is big, but by no means a monopoly.
I was thinking the same thing! Also, I have a Lyft app on my phone right next to Uber. Most places I go are served by both customers. It's easier to switch than groceries.
One data point: I’ve taught a few times at a community college here in the US. Obviously, that’s through an established institution. It paid about $7k for a ten-week course.
It's worth mentioning that while some parts of law can be really arcane (parents, terms of service, etc.), Supreme Court decisions are generally pretty readable.
Agreed! And the typography is perfect. I've enjoyed bypassing all the angles and bias of coverage and just reading the majority decision PDFs in full. As a layperson the implications of phrases do escape me, though. SCOTUS can get awfully terse.
Late in law school, I ended up writing a study guide for a multi-day Constitutional Law seminar for non-lawyers, and found that unpacking all of the legal turns of phrase and items that would carry huge implications when read by lawyers took between 3x and 15x the space of whatever segment I was unpacking, with an average around 7.5x-8x. And according to feedback that still ended up being a bit dense for most readers. Worst volunteer gig I've ever agreed to!
The dissenting opinions are also quite enlightening because they point out weaknesses in the majority and concurring opinions that might not be apparent to those of us outside the field.
Justices deliberately write their dissents in the hope that people will read them, be persuaded, and then those will eventually become established law. Ginsburg and Scalia were masters of this.
By the time you see a published SCOTUS decision clerks have definitely checked all citations. The problem is that occasionally the source or interpretation is questionable/contentious.
There's something very funny and simultaneously chilling about a majority opinion authored by Gorsuch[1] who has said “[o]nly the written word is the law” [1] totally mixing up laughing gas and a toxic pollutant emitted by cars.
[1] I know the clerks actually write them, HN pedants.
Clerks have much more limited time than one might think, and research is basically an added-on function that they never staffed for once courts started engaging in it.
“Facts” from amicus briefs make it in all the time without an apparent attempt at verification.
Perhaps they do check and then ignore their findings, but why bother with that?
You may be referring to the court's (and other appellate courts') long-standing practice of not second guessing a trial court's finding of what did or didn't happen, unless a "clear error" is demonstrated to have been committed by the trial court.
SCOTUS decisions and opinions, therefore, should not be construed as to concur with the trial court about what actually happened. They take it as granted that those things occurred, and provide the best legal resolution they can under those assumptions. The cases they hear aren't about determining "who did what". That's done and settled usually. They are about "what now?"
Gorsuch mixing up nitrous oxide and nitrogen oxide over and over is a recent fun example. The court also has engaged in fact finding despite the rules. Kennedy v Bremerton is one example. Alexander v. South Carolina State Conference of the NAACP is another.
You also see absolute garbage lies. FEC v. Ted Cruz for Senate has the majority saying "we can't find examples of X happening" while the dissent has a big list of examples of X happening. Or just straight up falsehoods about original understanding. DC v. Heller is a great example of this.
I mean that they make errors in statement of fact supporting their rulings and, worse, incorporate those facts into their ruling in material ways.
Say the Supreme Court majority writes, as guidance to lower courts, that interpretation of a certain amendment should feature consideration of laws enacted early in the country’s history, and before its founding. They further write that for the specific question before them, the total absence of similar laws in that history means they must rule a certain law unconstitutional.
Further suppose they were simply factually incorrect to the point that such laws were in fact common and are very easy to find, if you look like at all.
Now what? If you apply their guidance on how to analyze these questions, you’d have to reverse their ruling on laws similar to the one they struck down, should they come before you. But they ruled that specific one unconstitutional… but their ruling was contrary to the guidance they gave.
So we end up tied in a bit of a knot. Had these facts been argued rather than pulled out of some damn amicus brief without examination, perhaps the government would have presented a large pile of examples to rebut the simply-false claim that no similar laws existed in the country’s early history. But the court injected these “facts” as a key part of their reasoning when writing their decision, instead. Would it have changed the outcome? No. Would it, perhaps, have made it too embarrassing even for these clowns, to include that particular bit in their ruling? Maybe! And future lower court cases might take a different course, as a result.
[EDIT] The take-away for the casual reader of Supreme Court opinions, then, is that if they write something like "no examples of such laws exist until [YEAR]" don't be surprised if that turns out to be hilariously wrong. A "fact" making it into a Supreme Court opinion is not a strong indication the fact is... an actual fact. Their opinions are far less well-researched than one might suppose, emphasis on far, it's not that they're just imperfect like any people, their fact checking is outright poor by any standards.
My specific example is just Bruen with the serial numbers filed off.
Factual errors are downright common (a Google will turn up many efforts at fact-checking the "facts" in Supreme Court opinions) and aren't a new problem, dating back basically forever.
One key problem is that they simply don't have the resources to check all their facts. Their sources of facts are often amicus briefs from interest groups and specialized court-lobbyists, drawn from large piles and quickly skimmed for relevance by overworked early-20s law clerks. It would be surprising if they didn't get things wrong all the time. They do—luckily, a lot of times, it barely matters, but sometimes they get facts wrong that were central to their opinion.
The point is, don't believe a "fact" you read in a court opinion without double-checking. Even if it's about legal history.
[EDIT] As my sibling commenter points out, it's also the case that sometimes they just lie on purpose. But even absent that, the circumstances under which opinions are written would generate factual errors by accident, with some frequency.
FEC v. Ted Cruz for Senate is a great example of a case where the majority says "we don't see examples of X" while the dissent cites a big list of X.
Heller is also a famous case where basically every historian (as well as the defense) points out clearly that Scalia's interpretation just isn't the original understanding of the text.
I was making a barely-veiled reference to Bruen, actually. "Can cities require demonstration of need, to carry firearms?"
The bump stock ruling's stupidity had more to do with resting entirely on silly, plainly-motivated reasoning than its relying on objectively wrong "facts".
Incidentally, I was wrong about it mattering whether it was argued: the historical evidence was argued, and the majority simply went "uhhhh those many examples don't count, because I don't want them to". What's been turned up since the case is examples that fit the much narrower criteria they said would have been needed (neatly carved out to evade the provided examples) but positively asserted don't exist (to any notable degree), in such volume that it's beyond clear that restricting carry of firearms in towns on a need-basis was common throughout the country's history, and not just in the West (their reasoning for rejecting those examples, I shit you not, was "um, those don't count because they're from one area, never mind it's a giant area"—if it's starting to look like they were playing Calvinball with that ruling, it's because that's exactly what they were doing)
Mostly agree about Constitutional cases. Some of the less sexy statutory construction cases are written for a smaller audience, and even as a lawyer, if you don’t practice in that field, they can be pretty opaque.
I only read the big cases since law school, so take this with a grain of salt -- I've only read one or two KBJ decisions and a handful from Kavanaugh and ACB.
In terms of legal clarity alone, Gorsuch and then Kagan are ahead of the pack by some margin, followed at some distance by Roberts and ACB. Sotomayor is great as a writer, but not in terms of legal clarity. I think Thomas is middle-of-the-pack, and I think he gets underrated in this regard because people dislike his opinions. Again, experience of Kavanaugh and KBJ's writing is limited, but they're at least not far behind the pack here.
Alito stands alone as the only one I'd say is bad for a Supreme Court justice. And that's relative, so it doesn't mean he's awful, but I do actively avoid his writing.
The variance gets way higher the further from SCOTUS you go. At the state level it's basically roulette.
I'd agree with this. Kagan is the best overall, Roberts is up there when he wants to be and isn't forcing it, and yes Thomas writes fairly well but what he writes is abysmal.
Alito is hands down the worst writer and one of the worst to ever sit on the court. He's not the brightest but thinks he's a savant and writes with a smugness and conceit that drips off the page on top of being as disingenuous as they come. Reading anything he writes is torture.
Yeah, when he's like this his writing goes off a cliff. I feel like it peaks any time Sotomayor wrote whatever he's disagreeing with, but that may just be me.
I suppose that they are free to focus on readability when they don't have to structure their writing in order to defend their decision-making from anyone, being the highest court of the land.
A few points to the contrary: The primary purpose for their writing of opinions is not to defend the decisions, but to instruct lower courts on how they should reason on similar cases. So it very much matters on a practical level that their reasoning is sound. Second, if the goal is a cogent, defensible argument, readability is in support of that objective, whereas you suggest that readability is somehow (?) at odds with cogency and defensibility. Thirdly, the justices are keenly aware that the interested public consumes their opinions too, and that they are, in fact, ethically and morally bound to defend their decisions to the public.
> you suggest that readability is somehow (?) at odds with cogency and defensibility
All I am saying is that writing can be clearer and more concise when you don't have to include asides and digressions to preemptively defend it from criticism.
This is a technical point on power dynamics and writing style. When the authors are in a position of power, then they can focus more on readability (and cogency) and less on playing defense, because their court seats are secure for life.
By contrast, justices in lower courts are influenced by organizational politics and must focus more on defensibility, which may come at the expense of clarity.
So I do not find it surprising that the supreme court is able to produce more readable decisions than lower courts, because they have unique incentives.
> The primary purpose for their writing of opinions is not to defend the decisions, but to instruct lower courts on how they should reason on similar cases.
Correct. This is different from lower courts, which also explains why they are able to focus on readability.
Being the only non-democratically elected branch of government in a democracy, the courts don't enjoy direct approval from the people. So while the structure of the institutions seem to give them power, their authority ultimately hinges on whether people perceive their decisions fair and just.
When courts make unpopular decisions and can't explain why, there's always a risk of some constitutional crisis. If courts keep making decisions that they can't rationally defend, at some point their authority will begin to erode. (And if you followed the USSC rulings in recent years you might start to understand why. Maybe the resentment not yet directly targeted towards the courts, but those decisions did add fuel to the already divisive politics in the US.)
That was my first thought as well, "The camera will be used to take photographs" seem pointless, if it's a camera app it's obvious if it's not it does not help the reviewer.
I’m guessing reviewers have very limited time for each review. I don’t know what the internal process looks like, but I’m guessing an automatically generated screen listing the app’s permission requests, the justifications for each and other stuff? I would be surprised if there would be time for them to actually use any of the apps that they review or if they even have a phone to try apps on.
This seems exactly the kind of result I would expect from LLM automation. I would never trust any system that used LLM output without human review. Actually, I don’t think I would trust any system that used LLM output even with human review.
From what the Hudson Rock article shows, they were able to use an SE’s creds to access their demo account. This is not a customer account and shouldn’t (but of course could) contain sensitive info. It’s not clear to me how this snowballed into a larger breach.
Perhaps customers had granted this SE access to their accounts and the data within. Or perhaps there’s a deeper hack. But this isn’t clear to me from what I’ve read.
I was just going to post the same thing. The files that they show in the screenshots are things like PROGRESSIVE_BID_CHANGE_202405271129.csv. Looks suspiciously like the Snowflake Sales Engineer's data for their job role closing a deal with Progressive, not Progressive's own data. And there's no reason to think that a SE would have broad access to customer data. There may be some overlap, but I doubt it contains sensitive customer data owned by Progressive.
You’re thinking “bid” is in reference to Snowflake bidding for progressive as a client?
I’d say thats not likely, I work in fintech and the first thing this filename indicates to me is a CSV feed of market data for bid prices (https://en.m.wikipedia.org/wiki/Bid_price)
This is a common type of dataset a firm would dump into a datalake to use as reference data lookups against other more sensitive data (for pricing trades, etc.)
Something similar happened at a previous employer. Contractor was hired to do a big data PoC, and they managed to cajole access to a prod data dump for a more impactful demo.
They then managed to load all this PII data into an ElasticSearch instance that was open to the internet and was discovered by threat actors.
I wouldn’t be surprised to find that something similar happened here, where an unscrubbed prod dataset was shared for a better demo.
