simiones

If there had been a massive migration of hundreds of thousands of people, and even more so hundreds of thousands of slaves, from late bronze age Egypt (a powerful, old, highly literate kingdom), we would expect to find significant evidence of this (inscriptions, local stories, migration sites, etc). The absence of any such evidence, while not conclusive proof of course, constitutes evidence against this event happening.

We also know for example that the types of beliefs detailed in Exodus, especially the idea that the Israelites worshiped Yahweh alone as the only God, are not historical. Belief and worship of other gods were common in both the ancient kingdoms of Israel and Judah long after the supposed time that the Exodus happened - in particular El (who was later associated with Yahweh) and Asherah (who was sometimes seen as the wife of Yahweh). So at least this aspect of the Exodus narrative is directly contradicted by archaeological evidence.

This is similar to the reason we believe the stories in Genesis are not historical, e.g. the flood, - if they had been historical, we expect that they would have left behind certain marks; those marks haven't been found, so we have a reason to believe that they didn't happen.

The Exodus is an entirely fictional account though, it's not based on any real historical events. Even King David seems to be mostly mythical, though there is some vague evidence of a "House of David" being something some real kings claimed descent from.

Edit: I should say "almost entirely fictional". The main scholarly agreement is that it may record some stories of some small numbers (hundreds, at most some thousands - nowhere near the 600k in the Bible) real semitic slaves' escape from Egypt and migration to the area of Canaan, mixing with the local Canaanite population that were the precursors of the Jewish populations of later Israel and Judah.

I really wonder where this idea that the world was less polarized before social media is coming from. It's not even 100 years ago that we had some of the most extreme ideologies in history taking hold all over both Europe and the USA (fascism, socialism, and others). People literally went to war over these things. Another ~100 years before that, French people were cutting off the heads of their ruling class, and setting prisoners free.

If anything, social media has inspired far too much passivity in our societies. People feel relieved that they could vent their frustrations online, instead of taking to the streets and seriously threatening some of the power of those putting them down.

Also, a big part of why the elites of society dislike social media is the huge democratizing effect that it has had on information. Of course, not so much in the more authoritarian societies where our leaders were hoping for this effect, but in their own backyards. The biggest example of this by far is the information about the Gaza genocide - that is presented at best equivocally in the mainstream press (with some exceptions like The Guardian), but that was clearly visible on TikTok and other social media. This led to perhaps the single largest policy conflict between the vast majority of the population and the vast majority of government elites in the current day EU.

What is the false positive rate that you would be comfortable with such a scan having? What would be the risk of your personal photos and videos being recognized as CSAM and reported to your local police (and thus being shown to your local police) that you would be happy to accept?

Would you also be ok with not being allowed to send any mail unless you first scan the contents of everything in that envelope and include a generated signature that might tell the post office that you're sending CSAM? And then having the envelope delivered directly to police if the scan did indicate that?

FTA:

> What changes with the return of Chat Control 1.0—and what stays the same:

> What is coming back: US tech companies are once again allowed to scan private messages without a warrant or prior suspicion. This affects direct messages on platforms like Instagram, Discord, Snapchat, Skype, and Xbox, as well as emails via Google’s Gmail and Apple’s iCloud.

> What remains unchanged: Public social media posts and files hosted in cloud storage could already be scanned without this law. Furthermore, private messages can always be reported by users, or monitored by authorities using targeted, court-ordered wiretapping.

> What is still NOT being scanned: End-to-end encrypted chats, such as those on WhatsApp, have always been exempt from these scans. Additionally, European providers of messaging and email services have never implemented chat control measures.

> It’d be hard to pass its test suite and not be as robust as the original.

This is not true, even in principle, even for Postgres itself. You'd be right to say that it'd be hard to pass the test suite and not be robust at all to some extent. But even in Postgres, I bet that you can quite easily introduce a change that will pass the whole test suite but reduce robustness compared to the latest release (for a somewhat silly example, add a call to `exit()` on a timer that's longer than the longest duration test in the suite - that will significantly reduce robustness while still passing the entire test suite).

You answered at more or less the same time as an OpenAI engineer confirming this is the case, explaining why they did it, and pointing to their docs that also very explicitly explain this "made-up nonsense". You should really consider why you felt so certain about something you were plain wrong about.

Not always, it depends on the defect. There are many, many design defects that do not significantly impact the working of the device that will not get repaired, but would usually be patched in regular software (especially for visual design or bad UX like a literal rough edge on a physical product). Recalls in particular are very rare - they generally only happen for very serious faults, usually those that affect the safety of the device.

Also note that expectations can vary quite a bit between physical and software products. For an example, my vacuum cleaner fairly often shuts down if I stop and re-start it too soon. If this happened with an app, I would certainly consider it a pretty ugly bug, and might ask when they plan to fix it. But I would never consider sending the vacuum cleaner to a service station to try and get rid of it - and I think there is little chance they could fix it even if I did.

If you invest 100B dollars into compute while asking for 1T, but in reality would need 100T dollars to meaningfully move the needle, you're still significantly underestimating the amount of compute needed, despite not getting as much as you wanted.

The biggest tension between software and one-off purchases is related to bugfixing, and especially security. It makes sense that you don't expect new features for a product that you paid for once. But, what is the case with bugs? If the product mostly does what it promised to do, but sometimes crashes, do you expect those crashes to be fixed for your version, or not? What it you're using a 10-year old version and a new critical (say, unauthenticated remote execution) vulnerability is found in it? Do you expect to get it resolved as part of the price you originally paid, or would you be ok with being told you have to buy a new version if you want this ?

Interesting, I hadn't noticed... Though related to the critical mass, keep in mind that Waze has been acquired by Google many years ago now (13, specifically).

> Android Auto does try to distract me with attempts of engagement like "Is the construction site still there?" and other shit.

That's not Android Auto, it's Waze doing that (and it would do it on Apple too). And it's not "an attempt at engagement", it's a cost for keeping the real-time map and traffic data that Waze is chosen for up-to-date. Use Google Maps if you don't like the interruptions.

Oops, I confused some history, thank you for the corrections!

Well, it's not exactly engineering, but physics often uses quite informal math. For a pretty modern example, the Dirac delta "function" was used long before it was formally described; and I have heard it said that even today String theory uses some math that is not fully formalized - though I can't say I know what specifically, so I may be wrong. Newton expressed calculus in terms of inifinitesimals (the dx notation was simply an infinitesimal delta x, not merely notation for derivation), which were not not formalized until much later, after they had already fallen out of favor and had been replaced in formal math by the delta-epsilon limits-based constructions.

Edit: one better example from modern physics - the path integral formulation, used in both string theory and other areas of QM/QFT, is not fully formalized and formally proven to work. Also, a more concrete example of a widely used but actually still unproven conjecture in string theory is the famous AdS/CFT correspondence.

This is a bit reductive about what "proof" actually means in mathematics. Even in math, the kind of formal proofs that tools like Coq can automatically verify are an extreme, and lots of accepted and practiced math is not doing that. Proofs are often more abstract and even occasionally hand-wavy (for example not proving "obvious" statements or minor lemmas).

Mathematicians also occasionally build on top of unproven foundations (e.g. all popular asymmetric encryption schemes are built on top the assumption that certain problems such as integer factoring are hard, for which there is no formal proof), or at least explore both possibilities for statements with unknown truth value (e.g. you can find lots of work that explores the consequences of P = NP and/or P != NP).

However, there is a major separation between math and programs that I think mostly invalidates your proposal - most math we're talking about here is simply not applicable directly to the real world in any way. It's only studied for the interest of mathematicians. There is no real world consequence for Fermat's last theorem, for example - it was just a really interesting to prove theorem. In directly applied math, such as engineering, it is in fact much more common to work with unproven but well tested conjectures.

> How is the government supposed to be able to regulate them impartially when they're literally invested in them.

That is exactly the point of this move, especially during the Trump administration.

2500 kcal is waaaayyy above the TDEE of most people (especially most women - who hover around 1200-1500 TDEE; but even for men, 2000 TDEE is quite common for a regular office worker lifestyle). So most people would get quite fat quite rapidly eating that.

That's why Skittles-based diets tend to perform so well!

> a group who who broke a policewoman's back with a sledgehammer?

This was tried in court, and the conclusion was that this was an accident during a fight, not an intentional act [0] (after violence was initiated by security forces and the police during a PA break-in against an Israeli weapons manufacturer).

So this is quite deceptive framing.

[0] https://novaramedia.com/2026/05/05/four-palestine-action-act...

Maybe his views are more heterodox and he was counting Puerto Rico as 51st! (sarcasm, in case this isn't clear)

That's not the problem, and this is not just pedantry. It's just not correct to say that n/0 = inf, nor even to say that positive_n / 0 = inf, in any normal math context.

For example, if you accepted that n/0 = inf just like n/1 = n, then you'd conclude that n/0 + 3 = inf + 3 = inf, so n/0 + 3 = n/0, so 3 = 0. Or you'd want to do weird things like asking what is sin(inf).

There are more layers to this problem, if we want to get into the details. The LLM is defined in terms of floating point operations, and those are not actually fully deterministic, on most hardware and in most performant implementations.

IEEE 754 only specifies precision requirements for certain operations, not precise bit patterns (e.g. for exponentials). So, at least in principle, the same hardware performing the same operation could produce different results at different times, as long as they are close enough to the theoretical answer. I'm not sure if any hardware actually works like this.

IEEE 754 also specifies that many of the basic arithmetic operations are not associative - so any reordering (which is common when batching multiple queries at the same time) will introduce indeterminacy from the perspective of your own query (that is the result for your query will change depending on what other query happens to be processed at the same time, which is not under your control).

Finally, even if we take the case when a query is processed alone, and even if one particular hardware is completely deterministic, the result will be different on different hardware - which can again look like non-determinism if you're sending your query to a load balancer.

So, the math for LLMs is deterministic in theory, but implemented with non-deterministic approximations & optimizations in practice, and their results are then normally used only as a probability distribution to be sampled from.

This sub-thread was about cloud environments, where different requests may be served by different hardware. And it's in fact very likely that there will be a mix of different hardware from different vendors, in any particular LLM cloud for now.

In almost all forms of math, the value n/0 is undefined. It's definitely not infinity, for two reasons - depending on the value of n, it can be negative; and neither info nor -inf are numbers, so they can't be the result of an equation (unless you look at transfinite equations).

What you can do in math is talk about the limit of a series of fractions as the denominator approaches 0, and that's where you get some relation to infinity or -infinity. But the limit can also be any other number, if the numerator also gets closer to 0; or it can not exist, if the function oscillates.

The very concept is quite culturally contingent - both in how much buyers would care about getting such a service, and in how much celebrities would be willing to engage in this type of fan service. Going global would also require a huge amount of celebrity contractors, a big problem for scaling.

There's a huge difference between purely intellectual subjects, like organic chemistry and integrals, and the mix of emotional and intellectual depth of a novel. A lot of the meaning of literary works is built on top of shared human experiences, just like the meaning of integrals is built on top of more foundational math. However, we don't have any good way (and certainly make no effort in school) to teach this shared human experience to pre-teens and teens.

A good part of the value of some of these works basically comes from recalling similar feelings you felt in situations similar to the characters, maybe comparing your actions at such times to theirs, or the reactions of other people you knew, etc. It's simply not possible to experience this part of the work as a teen. Perhaps one of the clearest examples of this limitation is in Lolita - the nature of the relationship described, the power and life experience differential, the contrast with the reader's normal interactions with children - are impossible to be conveyed to or truly empathized with by teens.

I would say that Checkhov is mostly known for his plays, not for short fun stories.

I don't think it's conceptually difficult, it's just hard (for some of us, at least) to remember the names of a culture you haven't been exposed to. I speak a language with slavonic influence, so I didn't have this problem with Russian literature, but I remember vividly how hard it was to remember the names of characters in the first anime I watched, just because I was so unaccostemed to Japanese names (even though they were clearly very distinctive).

The recall of words you aren't familiar with tends to be pretty poor. This is also visible in how hard it can be to build vocabulary when learning a new language, and how you can completely mix up words at that stage - there's nothing about names that makes this any easier.

You mean GNU Hurd, I guess.

There are many crappy design decisions in Unix, Posix, Linux, MacOS, that have been known and worked around for decades (see the Unix Hater's Handbook). One example would be async IO, which has been famously bad in Unix, and Linux alone has tried 2 different generations of solutions - with the newer one, io_uring, being suspiciously similar to Windows' decades old IOCP. Fork/exec is infamous as a needlessly complex process creation API, requiring huge effort in the kernel just to handle the most common case of simply launching a new process. The traditional Unix security paradigm of simple user based file permissions is extremely weak and has required numerous solutions on top to handle realistic security scenarios - e.g. to protect a user's files from being tampered in unexpected ways by a process launched by that user (SELinux, jails, namespaces, etc).

I agree that Windows is crappy, but that doesn't mean that Linux and MacOS aren't also crappy in their own ways (not to mention iOS, Android).