> people publishing articles that contain these kinds of LLM-ass LLMisms don't mind and don't notice them
That certainly seems to be the case, as demonstrated by the fact that they post them. It is also safe to assume that those who fairly directly use LLM output themselves are not going to be overly bothered by the style being present in posts by others.
> but there are also always clearly real people in the comments who just don't realize that they're responding to a bot
Or perhaps many think they might be responding to someone who has just used an LLM to reword the post. Or translate it from their first language if that is not the common language of the forum in question.
TBH I don't bother (if I don't care enough to make the effort of writing something myself, then I don't care enough to have it written at all) but I try to have a little understanding for those who have problems writing (particularly those not writing in a language they are fluent in).
> Or translate it from their first language if that is not the common language of the forum in question.
While LLM-based translations might have their own specific and recognizable style (I'm not sure), it's distinct from the typical output you get when you just have an LLM write text from scratch. I'm often using LLM translations, and I've never seen it introduce patterns like "it's not x, it's y" when that wasn't in the source.
The “paging space needs to be X*RAM” and “paging space needs to be RAM+Y” predate hibernate being a common thing (even a thing at all), with hibernate being an extra use for that paging space not the reason it is there in the first place. Some OSs have hibernate space allocated separately from paging/swap space.
> There’s a common rule of thumb that says you should have swap space equal to some multiple of your RAM.
That rule came about when RAM was measured in a couple of MB rather than GB, and hasn't made sense for a long time in most circumstances (if you are paging our a few GB of stuff on spinning drives your system is likely to be stalling so hard due to disk thrashing that you hit the power switch, and on SSDs you are not-so-slowly killing them due to the excess writing).
That doesn't mean it isn't still a good idea to have a little allocated just-in-case. And as RAM prices soar while IO throughput & latency are low, we may see larger Swap/RAM ratios being useful again as RAM sizes are constrained by working-sets aren't getting any smaller.
In a theoretical ideal computer, which the actual designs we have are leaky-abstraction laden implementations of, things are the other way around: all the online storage is your active memory and RAM is just the first level of cache. That ideal hasn't historically ended up being what we have because the disparities in speed & latency between other online storage and RAM have been so high (several orders of magnitude), fast RAM has been volatile, and hardware & software designs or not stable & correct enough such that regular complete state resets are necessary.
> Why? At that point, I already have the same total memory as those with 8 GB of RAM and 8 GB of swap combined.
Because your need for fast immediate storage has increased, so 8-quick-8-slow is no longer sufficient. You are right in that this doesn't mean you need 16-quick-16-slow is sensible, and 128-quick-128-slow would be ridiculous. But no swap at all doesn't make sense either: on your machine imbued with silly amounts of RAM are you really going to miss a few GB of space allocated just-in-case? When it could be the difference between slower operation for a short while and some thing(s) getting OOM-killed?
Swap is not a replacement for RAM. It is not just slow. It is very-very-very slow. Even SSDs are 10^3 slower at random access with small 4K blocks. Swap is for allocated but unused memory. If the system tries to use swap as active memory, it is going to become unresponsive very quickly - 0.1% memory excess causes a 2x degradation, 1% - 10x degradation, 10% - 100x degradation.
What is allocated but unused memory? That sounds like memory that will be used in the near future and we are scheduling in an annoying disk load when it is needed
You are of course highlighting the problem that virtual addressing was intended to over abstract memory resource usage, but it provides poor facilities for power users to finely prioritize memory usage.
The example of this is game consoles, which didn't have this layer. Game writers had to reserve parts of ram fur specific uses.
You can't do this easily in Linux afaik, because it is forcing the model upon you.
Unused or Inactive memory is memory that hasn't been accessed recently. The kernel maintains LRU (least recently used) lists for most of its memory pages. The kernel memory management works on the assumption that the least recently used pages are least likely to be accessed soon. Under memory pressure, when the kernel needs to free some memory pages, it swaps out pages at the tail of the inactive anonymous LRU.
Cgroup limits and OOM scores allow to prioritize memory usage on a per-process and per-process group basis. madvise(2) syscall allows to prioritize memory usage within a process.
> There is 264KB of space left for your newly created files.
This could be increased noticeably by using one of the common extended floppy formats. The 21-sectors-per-track format used by MS¹ for Windows 95's floppy distribution was widely supported enough by drives (and found to be reliable enough on standard disks) that they considered it safe for mass use, and gave 1680KB instead of the 1440Kb offered by the standard 18-sector layout. The standard floppy formatting tools for Linux support creating such layouts.
--------
[1] There was some suggestion² that MS invented the extended floppy format, they were sometimes called “windows format”, but it³ had been used elsewhere for some time before MS used them for Windows and Office.
[2] I'm not sure if this came from MS themselves, or was invented by the tech press.
[3] and even further extended formats, including 1720KByte by squeezing in two extra tracks as well as more data per track which IIRC was used for OS/2 install floppies.
IIRC rsync uses your default SSH options, so turning off compression is only needed if your default config explicitly turns it on (generally or just for that host). If sending compressible content using rsync's compression instead of SSH's is more effective when updating files because even if not sending everything it can use it to form the compression dictionary window for what does get sent (though for sending whoe files, SSH's compression may be preferable as rsync is single threaded and using SSH's compression moves that chunk of work to the SSH process).
> Are there really people who "spend weeks planning the perfect architecture" to build some automation tools for themselves?
Probably. I've been known to spend weeks planning something that I then forget and leave completely unstarted because other things took my attention!
> Commenter's history is full of 'red flags'
I wonder how much these red flags are starting to change how people write without LLMs, to avoid being accused of being a bot. A number of text checking tools suggested replacing ASCII hyphens with m-dashes in the pre-LLM-boom days¹ and I started listening to them, though I no longer do. That doesn't affect the overall sentence structure, but a lot of people jump on m-/n- dashes anywhere in text as a sign, not just in “it isn't <x> - it is <y>” like patterns.
It is certainly changing what people write about, with many threads like this one being diverted into discussing LLM output and how to spot it!
--------
[1] This is probably why there are many of them in the training data, so they are seen as significant by tokenisation steps, so they come out of the resulting models often.
The best marketing is indistinguishable from non–marketing, like the label on the side of my Contoso® Widget-like Electrical Machine™ — it feels like a list of ingredients and system requirements but every brand name there was sponsored.
> … the internet is already full of LLM writing, but where it's not quite invisible yet. It's just a matter of time …
I don't think it will become significantly less visible⁰ in the near future. The models are going to hit the problem of being trained on LLM generated content which will cause the growth in their effectiveness quite a bit. It is already a concern that people are trying to develop mitigations for, and I expect it to hit hard soon unless some new revolutionary technique pops up¹².
> those Dead Internet Theory guys score another point
I'm betting that us Habsburg Internet predictors will have our little we-told-you-so moment first!
--------
[0] Though it is already hard to tell when you don't have your thinking head properly on sometimes. I bet it is much harder for non-native speakers, even relatively fluent ones, of the target language. I'm attempting to learn Spanish and there is no way I'd see the difference at my level in the language (A1, low A2 on a good day) given it often isn't immediately obvious in my native language. It might be interesting to study how LLM generated content affects people at different levels (primary language, fluent second, fluent but in a localised creole, etc.).
[1] and that revolution will likely be in detecting generated content, which will make generated content easier to flag for other purposes too, starting an arms race rather than solving the problem overall
[2] such a revolution will pop up, it is inevitable, but I think (hope?) the chance of it happening soon is low
I'll still program. I'm barely touching current "AI" other than certain bits of code completion that are on by default in VS and are the right mix of occasionally useful (saving a few tens of keystrokes) and easy to ignore when not. When it gets to the point where I can't compete without using LLMs heavily, I'll have to do something else. Perhaps wait tables somewhere with Spanish customers as I'm trying to learn the language, it won't be much or a pay cut because by the point where someone of my years can't compete at all, almost anyone will be able to do a shite job with LLMs so programming will be heading towards being a minimum-wage job anyway.
> Everything else is being destroyed by AI: art, music, books, personal websites.
Yours doesn't have to be. I started writing bits to go online again after many years not bothering. It isn't intended for public consumption, I'm deliberately doing nothing that would be called SEO in fact I'm going anti-SE in some ways, though it isn't hidden should I chose to pass a link to someone or if they decide to pass it on further. It is intended for me and a select few, if someone else finds it somehow and likes it then good the them, and much of the point of making it is the joy of making something. Everyone else using LLMs isn't going to take that from me.
> Why read a blog post, when Google AI Summary can just give you the summary?
Because the summary may be wrong, or technically correct but with entirely the wrong tone or missing key details a summary should have. Or because you want to read the post - you don't have to use the summary.
I don't read the summary generated as part of most searches these days, reading them is optional. I don't make effort to stop them being made though (using porn mode, adding “fuck”/“fucking” to search terms, etc.) - hopefully making them waste resources on generating things that are never used will et noticed and they'll ramp that down a bit at least for accounts like mine.
> Why read a book, when you can just get AI summary of it?
Before AI, did you just skip to the last few pages to see what happened? If not, then just keep reading like you did before.
Avoiding the slop books that are flooding the market ATM might be more of a concern, but there was plenty written before the LLM bubble started to keep you in things to read for a lifetime.
> The only thing you are left to do is to eat and take a sht throughout the day.*
Other hobbies. I like trail running (or if in a slower mood, country walking), AI isn't going to stop me doing that. HEMA and other activities with friends too.
> How should people make money? No idea, as in the "prosperous future", everything is replaced by AI.
That is a completely different question to the rest where you are asking about things that you would do for enjoyment, and is a bigger societal and philosophical concern than I have time to chew on right now…
> What's the point anymore?
Same as it was before: do what you need to do to survive, then do what you need to do to enjoy any remaining time after that. Perhaps I'm helped here by being back on the up after a couple of years recovering from proper burnout: I know the way things are going won't make me feel worse than that, and I survived that. Current politics and other social issues away from tech on the other hand…
> Why keep going?
When all hope feels truly lost, at least keep going out of spite.
> When you explain a technical problem to someone who isn't intimately familiar with it you're forced to think through the individual steps in quite a bit of detail.
The point of Rubber Ducking (or talking/praying to the Wooden Indian, to use an older phrase that is steeped in somewhat racist undertones so no longer generally used) is that it is an inanimate object that doesn't talk back. You still talk to it as if you were explaining to another person, so are forcing yourself to get your thoughts in order in a way that would make that possible, but actually talking to another person who is actively listening and actually asking questions is the next level.
I guess I can see where others are coming from (the LLM is different than a literal rubber duck) but I feel like the "can't reply" part was never more than an incidental consequence. To me the "why" of it was always that I need to solve my problem and I don't want to disturb my colleagues (or am unable to contact anyone in the first place for some reason).
So where others see "rubber ducking" as explaining to an object that is incapable of response, I've always seen it as explaining something without turning to others who are steeped in the problem. For example I would consider explaining something to a nontechnical friend to qualify as rubber ducking. The "WTF" interjections definitely make it more effective (the rubber duck consistently fails to notify me if I leave out key details).
That certainly seems to be the case, as demonstrated by the fact that they post them. It is also safe to assume that those who fairly directly use LLM output themselves are not going to be overly bothered by the style being present in posts by others.
> but there are also always clearly real people in the comments who just don't realize that they're responding to a bot
Or perhaps many think they might be responding to someone who has just used an LLM to reword the post. Or translate it from their first language if that is not the common language of the forum in question.
TBH I don't bother (if I don't care enough to make the effort of writing something myself, then I don't care enough to have it written at all) but I try to have a little understanding for those who have problems writing (particularly those not writing in a language they are fluent in).
reply