Mikhail_Edoshin

After one of updates my MS Edge decided to start showing me the MSN page on a new tab. I had EN/US settings at the time so I received a page in English, either US or international.

So once it did that I glanced around the page and noticed that of top stories was about a model or actress of some fame, presumably. The heading told us that it could be that the actress appeared on stage or somewhere with little or no underwear.

Now, that kind of "news" would be impossible in USSR. Due to state censorship, of course. It did limit what you could watch or read. For example, C. S. Lewis was rather underrepresented. But at the same time it edified the whole field, protecting it from underwear news. (And that is a mildest example of what pollutes the cultural atmosphere of the West.) So while few Soviet people read Lewis, many were ready for that because they were raised in an environment where Lewis's topics and questions were important and serious. I remember Graham Green said that some of his most deep critics were "marxists", that is, came from the Soviet block; the West wasn't that interested.

I really doubt you've read much of Soviet writers.

About propaganda. When I was learning English, I watched movies. There were few, so I re-watched what I had several times. (Turned out to be an excellent method, by the way.) One of them was "Philadelphia", I've watched it six or seven times. Very well done. Now, tell me: isn't it a gay propaganda movie? I think it is. I see the whole set of talking points straight from political debates, masterfully dramatized and carefully curated not to reveal anything that might work against the desired goal.

I recently re-watched a few movies by Anthony Minghella (I liked his Ripley). I was surprised that the theme of two movie (and he didn't made that many) was treason and the author sided with the traitor. (Cold mountain, The English patient.)

I myself write loops with goto. It gives precise control, allows to reuse the knowledge gained from operations inside the loop, avoids artificial "while true" and is generally more convenient. It is also makes the language simpler as you can forget 4 to 5 loop-related keywords and hard-to parse operators designed primarily to look nice in loop conditions while also modifying their arguments.

This translates into "trust some authority", which is also a task. And it does not work when the canned solution does not quite fit, e. g. involves a bulky lib. The true answer should be "use your software engineering textbook and the standard reference of RNGs". Thing is, there is no such reference.

And RSSable. But they aren't. If most web content was in easily processable formats, many people could use that for various purposes, e. g. for search. This would diminish the market position of search gigants. But if web formats were hard to process, only big players would have resources.

I am not an expert, so I'm retelling what I myself read.

A Starlink terminal has a detail that reports its location. There is a restriction on location, so if a terminal reports a restricted location, it gets blocked. A terminal may get blocked either because it actually got off the territory or if it the GPS signal was somehow distorted. So all terminals that go into Ukraine are hacked: they get a chip that reports a fixed location. This allows both to fight the distortions and to ignore the restrictions. (For some time this hack allowed use of Starlink on either side; then a white list of terminals came into effect.)

Ukraine never used (maybe didn't have access to) neither the military subsystem, nor the high-speed tariffs (which are way more expensive). The tariffs they use limit the moving speed of a terminal to 100 mph (160 kmph). For some time the limit was not enforced so drones just flew at any speed.

Around the end of 2025 the limit started to get enforced. But initially the technology relied on the data of a GPS chip. Since the terminal was hacked, the GPS chip reported the same data and thus appeared static. The enforcement thus did not work for these terminals. So although the limit was always there and has been technically enforced for a good part of a year everything was quiet as things simply continued to work as they always did.

But it is possible to detect the hack if one checks how frequently a terminal changes the satellite it connects to. This improvement was made recently and rendered the hack unusable on high-speed targets. This is the point we are at.

Exactly that. Pictograms are never a decoration.

A more effective use of a cow would be to milk it more and feed less.

A mere fashion industry would be fine, because you could just ignore that and stay old-fasioned. This fashion industry is coupled with enforced corporate dress code.

Fewer internal barriers.

An interesting study, will try to find it, thanks. I've been thinking about "naming things" and it seems the common idea of "good naming" is pretty superficial. I myself would point to MOV in assembly, which is actually COPY, but this never was an issue.

What actually matters is that:

- Names express not only differences, but also similarities.

- Names cannot be totally random; they must have certain pre-baked relations vaguely similar to their relations in the model. But they do not need to be too good. Invented or repurposed names in moderations are perfectly fine.

- The most important thing in naming is the overall consistency. But when the system changes you need more names and the consistency suffers. After a few rounds it gets too bad and you need to redo the whole system. This us why naming is hard.

- Certain physical properties matter, such as equal length.

- Names are very similar to pictograms or hieroglyphs.

As one young philosopher put it, wisdom is a moral choice. It is a decision to only speak about what you truly know; and that is very little. E. g. most of us do not truly know the Earth is a sphere; we only "know" this from a hearsay and that is not it.

A more useful definition of an em is the nominal font height.

The underlying mechanism of that is still search. We do have similar mechanism; we can notice it in dreaming, delirium, "word salad" disorder and so on. It will generate formally "new" things and they will look convincingly normal, especially if our defences are down, as in sleep.

I am not sure what the current state of models is, but as far as I know lifting a limit on the model output length tends to noticeably degrade the quality of the result so that it starts to show the same delirious tones.

Fundamentally it is recalling of past experience, a reaction. In many ways it is an opposite of acquiring true new knowledge. We keep that mechanism in check using other human qualities models do not have.

Working with LLM is basically like hiring a madman. It can do useful work, but you cannot trust it. To be economically viable the verification cycle has to be cheap or the cost of an error must be small. Searching a library or doing a home project are fine. But these aren't the lucrative markets investors are dreaming of.