Kalium

At the scale of Meta, "down" is a nuanced concept. You are very unlikely to get every piece of functionality seizing up at once. What you are likely to get is some services ceasing to function and other services doing error-handling.

For example, if the service that authenticates a user stops working but the service that shows the login form works, then you get a complex interaction. The resulting messaging - and thus user experience - depend entirely on how the login page service was coded to handle whatever failure the authentication service offered up. If that happens to be indistinguishable from a failure to authenticate due to incorrect credentials from the perspective of the login form service, well, here we are.

At Meta's scale, there's likely quite a few underlying services. Which means we could be getting something a dozen or more complex interactions away from wherever the failures are happening.

You are absolutely correct. That would be a much better experience.

That said, getting there strikes me as pretty challenging. Automatically detecting a down state is difficult and any detection is inevitably both error-prone and only works for things people have thought of to check for. The more complex the systems in question, the greater the odds of things going haywire. At Meta's scale, that is likely to be nearly a daily event.

The obvious way to avoid those issues is a manual process. Problem there tends to be that the same service disruptions also tend to disrupt manual processes.

So you're right, but also I strongly suspect it's a much more difficult problem than it sounds like on the surface.

Every healthcare system I've ever encountered, read about, or heard about has some kind of rationing. This is apparently independent of questions of profit or non-profit, private or public. Socialized, not-profit-driven healthcare systems also have to ration resources somehow.

I'm sorry, I can see I have been unclear. I am in no way defending the worst system of rationing healthcare on the planet. It's awful, horribly, abominable, and must be changed to include some basic fucking humanity.

I'm saying that rationing is inevitable and our thinking about health care systems needs to account for that.

One the primary goals of any - and indeed every - medical system is to ration resources. Which is to say to privilege access to medicine. When we talk about medical care access, we tend to talk around this, but it's a needed part of any system. Any system that doesn't do this quickly learns, institutionally, that without some limiting mechanisms in place the system will be overwhelmed and its resources exhausted.

Obviously some approaches and systems work better than others and are more tolerant to some kinds of faults. Unfortunately, none of this gets away from the absurd, insulting, and grotesque reality that care has to rationed somehow.

There are probably better numbers somewhere. Likely several sets worth. One of the things that makes SF's numbers especially thorny is that SFPD engages in a daily campaign to deter reporting crime.

This makes the official numbers known unreliable, but also means there's lots of room to debate how much more reliable any alternative set of numbers might be.

Naively, yes, diversity avoids systemic risks. However, from working in auto insurance it was my experience that insurers reason about risk in considerably more sophisticated ways.

For one thing, in the US auto insurance is entirely a state-by-state market. This means that geographic diversity is inherently severely limited. Any auto insurance operation in California is going to be exposed to a lot of wildfire risk without the ability to geographically diversify. This is priced in.

For another, there some groups of people - remember auto insurance is really mostly about insuring people - who are statistically more expensive risks than other groups. Young men are measurably less safe drivers than middle-aged women, to pick an anodyne example. Individual premiums reflect this as well. I have no idea what kind of systemic risk would uniquely affect all middle-aged women across an entire state, but presumably you can think of one.

Most insurers have groups of customers they would prefer not to have. This is usually because the insurer cannot cover them at cost, never mind profitably. There's a series of ways they encourage those customers to find other carriers. This is where the idea of maximizing diversity breaks down most clearly - the overall risk pool is not improved by including these customers.

In a great many cases, insurance rates drop. Sometimes quite sharply. I've seen as much as a 50% drop.

Auto insurance is inevitably surveillance capitalism. Insurers always look at aspects of life your life and behavior to establish your risk profile. Age, employment, the kind of car you drive, your driving record, and more all have a measurable impact on your risk.

Telematics mostly makes things cheaper for good drivers. It also sometimes gives insurers a chance to try to nudge drivers towards safer behaviors, both through direct messaging and through higher prices for drivers with higher-risk behavior. Whether this is a good idea or not is a personal question, though for my own part I tend to think that underpriced insurance for operators of heavy machinery is not a human right.

From the perspective of a private insurer, you do not want as diverse a risk pool as you can get. There are groups that you can reasonably anticipate are at high risk of being a bad bet. In many cases you want to encourage them to either change their behavior or seek an alternative provider. Either way, this looks like a high rate.

Telematics often makes it easier to figure out who those people are.

How do you imagine the auto insurance industry is driving this change in customer behavior and outcomes? I cannot imagine any way in which it's the fault of the insurers, but I know that this is merely a failure of my imagination.

Bear in mind that the auto insurance industry is quite distinct from the auto manufacturing industry.

Yes, I do think it's ineffective. It's needlessly ponderous, excessively emotive, and cares far more about its activist mission than questions like "How is the business structured?". It's not an accident that the central character is someone carefully chosen to be maximally sympathetic to a well-meaning, well-off, educated liberal white person.

Moreover, a similarly sympathetic (and horrified progressive middle-class-assuming) portrait of a check cashing business and its customers would almost certainly read as incredibly condescending to someone who finds them a fact of life. The article quite clearly assumes its readers have never been temps, never been crammed into a van to work in a warehouse, and will be upset to learn the conditions in question. It just doesn't say so briefly, opting instead to bury the assumption.

Do you want to read an article that portrays your younger self as a victim so that readers can feel bad and maybe learn about the financial structure of check cashing? At the risk of making a large assumption, that does not seem like it would help you feel included.

What, in your opinion, is the sensitive and inclusive way to discuss a category of business that the author can fairly assume a significant fraction of the userbase has never interacted with? Should the author discard that assumption and presume every reader is deeply familiar with check cashing instead? We know, with reasonable statistical evidence from the FDIC, that this alternative assumption is untrue.

I ask because there is, as far as I can tell, no obvious way to do it. You could just as easily write an essay talking about carrier-provided phone financing, and it would expose a similar cultural rift where most Americans don't do that and are only very vaguely familiar with how it works.

Or is the glaring thing that you dislike here the clearly stated assumption that most of the readership is not like you? In which case perhaps we should simply request that authors not state such assumptions while they make them anyway in the interests of being maximally informative.

It's also worth remembering that the world of LotR has people in it who were there for the advanced civilization. It's not beyond living memory. A number of those people are important minor characters, and the sense of loss helps form the backdrop for the ever-present grief of much of elven culture.

The past cataclysm was very much a part of the current world in LotR in many ways.

Mountains of expository prose are a crutch for an unskilled author. There are a number of storytelling techniques to build out a world as it becomes relevant that allow for hinting at greater depths. These allow for exposition, characterization, and drama while still hinting at greater depths without dedicating entire chapters to the socio-economics of grain production.

Most people already have something to lose. Giving them the fear of losing it has more to do with enforcement.

Where I am, I get a lot of aggressive behavior from drivers because I am sticking to the traffic regulations and they don't like it. The laws here very explicitly give me a clear right to use almost all roads.

Many drivers imagine that bicycles are not permitted in roads and some take it upon themselves to make my life dangerous in an effort to enforce that.

The VC-oriented financial model (that is, your typical startup) is fundamentally incompatible with Mastodon's non-profit financial model. As a result, very different choices around licensing, funding, and business structure are good fits.

Each tool has its use and serves its specific purpose. What works for Mastodon might not work for others trying to advance different goals.

I would hesitate to characterize a service running on grants and donations as a good model for your typical startup to imitate.

The effects are, if any, are not observable.

Let's go with a change of phrase, then. What's the practical difference between suggestion and a clear and explicit instruction that is unenforceable and where the user offering the instruction cannot tell if it is being executed faithfully?

What's the practical difference between an unenforceable and unobservable explicit instruction and a polite suggestion?

My point is not that Google is good or bad or that all the other kids are doing it. My point is that DNT is at current fundamentally incapable of delivering what we're asking it to do.

Your browser is doing that to the extent that it carries over information about downloads and bookmarks. That's not the same as Google associating those with the cookies used to track your Google account.

I've yet to see anything so much as suggesting that Google is using that kind of Chrome user local data to track people. Did I miss that?

Do Not Track is a suggestion at best. In the best of all possible worlds where Google does exactly as you suggest in forcing it on and respecting it, every other server out there is perfectly capable of ignoring a polite suggestion.

Your other idea is interesting. Google should lobby for regulations that would hurt their competitors far more than them. You wouldn't regard this as a big company engaging in regulatory capture?

> Incognito mode is supposed to not send your regular cookies so sites shouldn't be able to track you unless they use dubious stealthy tricks rather than the explicit well-known and documented method (i.e. a cookie). While I know that it's possible, I think it's morally despicable and I expect better from established (and ostensibly honorable) companies like Google.

Yeah, that's what it does. Instead you get a new set of tracking cookies, because to Google and any other random website you look like a new browser they haven't seen before. By design, they have no idea that you're using "incognito mode".

With that in mind, what do you think Google and others should do differently?

You can use older images, collected from before the "poisoning" software was released. Then you don't have to.

This, of course, assumes that "poisoning" actually works. Glaze and Nightshade and similar are very much akin to the various documented attacks on facial recognition systems. The attack does not exploit some fundamental flaw in how the systems work, but specific characteristics in a given implementation and version.

This matters because it means that later versions and models will inevitably not have the same vulnerabilities. The result is that any given defensive transformation should be expected to be only narrowly effective.

That works if you're willing to have your system be DoS'd. If every request takes half a second, that just means the attacker needs more systems sending requests in parallel to get the required number of samples quickly. In today's world, you'd probably just scale up a bit to ensure you could still serve requests. If you let your system go down under load, you can severely restrict how much data attackers can get at the cost of being useless to whoever your intended users are.

The attacker would also likely need more patience, but timing attacks are not for those looking for quick wins.

The preferred way to handle this is to make sure your operations are constant-time. This eliminates the timing side channel. This is harder, and often slower, but solves the problem entirely.

Bottom line: Yes and no. Yes, you can do that. No, it won't really do what you want.

The best way I've found to imagine a timing attack is as a set of graphs produced from thousands to millions of data points, each of which is a time measurement. Any given high-level input takes a certain amount of time to process (example: trying a login). Curves - and thus timing side channels - are distinguishable from one another when the curves have different peaks. The width of the (probably bell-shaped) curves are going to be determined by the amount of variance.

What happens if you add a little predictable random variance to each operation that's the same in each operation? Each curve probably gets a little wider, but you don't move their peaks relative to one another. At most, it might mean the person carrying out a timing attack needs to gather more data. Timing attacks already work with network noise implicitly doing what you've suggested doing explicitly.

You can simulate this yourself. Graph a thousand points of [0-1] and another thousand points of [0-1]+0.1. Superimpose the curves on one another. You may have to use some bucketing to make the curves visible.

It's a neat possibility for a science fiction author. I've seen the idea used a couple of times.

One of the key questions for life is where all the external energy to overcome entropy comes from. Being close to a star without being fried by it is a good way to accomplish this, but there are fewer easy energy sources on a rogue planet.

It's worth noting the critical details: how to prevent scurvy was understood, but the underlying mechanisms were not. This mattered because it meant why the treatment worked was not understood, with the result being a resurgence when a supposedly effective treatment turned out to be ineffectual.

Basically, it's easy to think we understand something when we have a solution to it, but the two should not be automatically conflated.

In some setups, you're required to trust both the maintainer and the mirror, which are not always the same party. If someone can generate a collision, it means a mirror can mount an attack even when the maintainer is corrected trusted.

Long story very short, history. Unions in the US are organized shop by shop.

You can go to a union's homepage and click "JOIN TODAY!" if you wish. It just doesn't create any obligation for your employer to care.