> In the last decade, keeping my OS on a different volume from my data in `/home` has saved my work at least 5-10 times.
Mine is of course just another anecdote, but despite seeing some crashes or hangs over the years, I can't remember the last time I've even had to manually fsck my ext4 filesystems, and I'm honestly not sure I've ever seen any unrecoverable corruption on ext4 or ext3. (I'm not saying it doesn't happen, as it certainly can, but I wouldn't think it's that common unless you're operating on a fairly large scale. And before journaling filesystems, crashes were of course a whole different deal.)
I used to keep /home separate on my personal devices (and maybe some other partitions as well because it was "a good practice"), but for more than a decade now my personal devices have had everything on a single large root fs, save for /boot and swap.
I've stuck with rather conservative fs choices and haven't used btrfs much, so I don't know if the story is different there.
What you gain with a single large root partition is of course that you don't need to know in advance how much software you'll want to install under /usr or how much space you'll need for /home or /var.
Multiple partitions can certainly be a good idea from a system management point of view, for cases where people either want to actively manage the system, or when their needs are demanding enough that it's warranted. If I were deploying a server for any kind of real use, yeah, the database or the server logs would certainly not be on the root volume.
But in case of desktops or laptops, most people, even engineers, don't really want to be meticulously managing their personal computers. Nor are most people's personal computing needs that demanding. The convenience of not running out of space on a single partition because you guessed the your future needs wrong wins over.
> Swap on a different physical drive is faster.
> Spread the load across multiple media for performance. That is how server OSes do it, because it is faster.
Of course it is, but in contrast, laptops don't really tend to have separate physical drives on which to put swap. I doubt most people would care enough about swap performance to install a separate high-performance drive for it even in desktops.
With that said, I also do keep a swap partition even on personal computers, and hibernation can be useful. (I think there are some issues with hibernation and encrypted swap in Linux, though, so I haven't bothered with it in a while).
I think the first release of Windows 95 did include TCP/IP networking support. I remember using it. But I think it wasn't installed by default, so you had to go and install it separately via the control panel.
The sandboxed application would be something you explicitly chose to run. When the browser is the platform, every web page you visit has access to the general-purpose runtime. (Of course permission checks e.g. for microphones exist.)
Why do you still need an explicit install step if everything is correctly sandboxed? If your only answer is you don't trust the sandbox then you shouldn't trust the one provided by the OS, either. Your browser's sandbox is tested way more than anything else.
It was actually reasonably common around the mid 90's for game CDs to have their music as audio tracks on the CD.
I guess there was a point when CD-ROM was the target format anyway but game content hadn't necessarily grown to fill it yet. Having the CD drive play the audio directly (it could be directly connected to the sound card) with no decompression or software processing required probably saved some CPU cycles as well.
I think at least half a dozen games I had around the time did that. Of course most of the soundtracks wouldn't be interesting enough on their own.
Did the early proprietary CD-ROM interface cards that shipped in "multimedia upgrade kits" and on sound cards even support DMA?
I know both DMA and PIO-only SCSI HBAs existed at the time, but I'm not sure about the proprietary interfaces that were more common for consumer CD-ROM drives.
Panasonic (MKE) was pure pooled PIO and didnt even support Interrupts. Sony and Mitsumi do have DREQ/DACK pins so most likely did ISA 8bit DMA. ISA DMA is pretty terrible and slow (4.7MHz clock, steals CPU cycles for the duration of the byte transfer), barely enough for 2x CDROM. IDE PIO is better on 286 and up than ISA DMA thanks to nice dedicated REP INSW. Taking into account ISA limits transferring one CD sector should take no less than ~700-1000us.
While ISA SCSI and VLB controllers could do DMA, they rarely did with ISA 16MB memory limit and VLB compatibility worries. The one VLB IDE controller dlaiming to do DMA, Promise PDC20630, does MWDMA2 to the drive but with CPU manually pooling IO pins on the other side, total waste https://www.vogons.org/viewtopic.php?p=1323789
TLDR: Talk of enabling DMA for your HDD and maybe even CDROM became common only somewhere around 1996, a good year into Intel shipping busmastering IDE controller in PIIX on Pentium boards (430FX and up). You had to put devices on separate IDE channels and hope everything is compatible and works.
It was routine for preteen and teenage me to put any game CD or game magazine coverdisc in the Hi-Fi CD player at some point, and see if the track number indicator would show anything higher than 01. It was definitely quite common. Some that I remember are Grand Theft Auto, Carmageddon, Alone in the Dark 2, Rayman, Little Big Adventure 2, Tunnel B1, 7th Guest, The Dark Eye...
The music was made interesting by the sheer novelty of being able to listen to it that way, and get a bit of the game to enjoy even after the parents decided computer time is over =D
I remember at least Civilization II, Descent II, Settlers II, Outcast, and I think some rallying game I had as using Redbook audio. The industrial soundtrack of Descent II was actually pretty good.
The interwebs also mention G-Police, which I remember playing, but I have no recollection about the soundtrack.
That has been the trend for decades in lots of other developed countries as well. It doesn't have to be in any way because of increased means of surveillance.
I honestly don't think a language model is enough for self-awareness, regardless of the exact model of awareness.
A language model (or an image model or whatever) cannot even be sentient, and I think sentience is a prerequisite for awareness.
Even if we express a lot of our subjective experience with words, the language is just a symbolic representation of those experiences. The qualia themselves, even those that are quite abstract, are rooted in our physical presence and evolution.
You can't have an understanding of what hunger or physical pain feel like if you have no need for food or a sensory capacity for feeling pain. You can't understand what loneliness or pride at an achievement mean if you don't have a neural network wired to value social connection or status. We value connection because we're social animals that have needed each other for survival.
Even the more abstract of our subjective experiences are in some way rooted in our physical evolution.
I see no reason to believe that a neural network built entirely based on the symbolic level of language could have the features needed for the subjective experience itself.
AI awareness might actually be more believable if that awareness manifested itself in an entirely different way than in humans. But if we assume awareness because outputs resemble what we consider meaningful as humans, yet the neural network has had no inputs or evolution that could form the actual basis of human-like experience, I think we're seeing something that isn't actually there.
> The qualia themselves, even those that are quite abstract, are rooted in our physical presence and evolution.
There is no objective evidence of qualia. All evidence of qualia are vocal or other expressions of belief in qualia. Perceptions clearly exist and are observable, subjective experience and qualia, not so much.
> I see no reason to believe that a neural network built entirely based on the symbolic level of language could have the features needed for the subjective experience itself.
If your objection is to models based on "symbolic level of language" which you think lack semantic understanding of, say, trees, you should ask yourself how our brain, based on physics which also lacks any semantic category for trees, can somehow develop a semantic understanding of trees. All of these appeals to differences with the brain never seem to acknowledge that fundamentally, the brain has the same explanatory gap with physics.
> But if we assume awareness because outputs resemble what we consider meaningful as humans, yet the neural network has had no inputs or evolution that could form the actual basis of human-like experience
This assumes a lot. It seems very possible to me that intelligence inherently develops a map of natural categories (natural kinds), and language naturally develops around such categorical understanding. Semantics are then fundamentally the network of associations between categories, eg. there is no fundamental difference between symbols and semantics, and the latter cam be inferred from the former, and that's exactly what LLMs do, and why the semantic maps between different languages are so similar and how they can translate between languages.
You can argue there's no objective evidence for qualia in others than oneself, but I'm sure you know you have subjective experience just as I know I do.
> All of these appeals to differences with the brain never seem to acknowledge that fundamentally, the brain has the same explanatory gap with physics.
I don't claim that ANNs or other computational models are fundamentally different from brains in computational capacity (although AFAIK biological neurons are a lot more complex than those in ANNs, so drawing a parallel just because both are "neural networks" would be erroneous as well.) I'm claiming that language alone cannot represent qualia, so an ANN trained on the level of language alone cannot have subjective experience.
> This assumes a lot.
Perhaps, but so does the idea that "intelligence inherently develops a map of natural categories", including the rather elusive concept of intelligence itself. And that today's AI models could possess intelligence in whatever form that sentence requires.
If forming natural categories alone were enough for sentience, clustering algorithms could be sentient.
It has traditionally been assumed that only humans are sentient. Present understanding leans more and more towards many animals having some kind of sentience as well. Some of them are probably even self-aware. Language in animals is a lot more controversial, though.
Nobody knows how (or even why) sentience emerges. But since it seems to exist in other animals as well, including those that probably don't have language or abstract concepts, it would seem to be a product of something else than language. It might even be some kind of a spectrum, with different animals having different levels of sentience and, ultimately, even awareness.
Making definitive claims about whether LLMs do or do not have specific properties absolutely does require precise definitions of those properties that can be used to evaluate those questions. Merely hand waving that LLMs didn't undergo the same evolutionary process is not a definitive argument.
For example, the Turing machines and the lambda calculus don't look anything alike, but they are fundamentally interconvertible, and so in a real sense they are fundamentally equivalent. Without a model, all of your arguments are completely unconvincing for exactly the same reasons, eg. that there may exist many paths to fundamentally equivalent ends.
I just don't think linguistic (or other symbolic) representations alone can contain the information, in any sense of the word, of what e.g. human subjective experiences actually are like. The concepts we express with language get their meaning from our physical reality, even if quite indirectly in case of some abstract concepts.
Hunger as a concept doesn't mean anything without the physical need. Politeness or bluntness, even in writing, don't mean anything without social dynamics. And we have social dynamics (and neural structures that directly process social cues and associated feelings) because we've evolved into social animals for whose survival that was important.
I see no reason to believe that a model trained only with symbolic representations, with no connection to the physical world phenomena that those symbols represent, could contain the subjective experience itself.
Neural network models may be able to derive novel (or at least novel-looking) output rather than just an obvious rehash of their input, but I don't think any set of bytes can fundamentally contain information that was never entered into it. (Even if e.g. a model produces previously unknown mathematical results, those results can in principle be derived from the information that they were trained with.)
I'm not saying that artificial neural networks couldn't, in principle, be aware. ANNs and biological neural nets may be equivalent in the sense that any information and processing structures represented by a biological one could in principle be represented by an artificial one. If that's the case, and awareness is purely a product of our neural systems as materialism would imply, it should be possible for an ANN to be aware, too.
But when the model has been trained with only language, and IMO the subjective experience can't be derived from the symbolic representation alone, I can't see how the model could include the actual subjective human experience.
An AI model could of course have an awareness and subjective experiences that are totally different than our human experience. But then the fact that it happens to produce output resembling what humans find meaningful shouldn't be considered indicative of such awareness.
This is of course more of a philosophical argument than a technical one, and I'm happy to hear counterarguments, but not on the level of off-hand dismissal.
> I can't see how the model could include the actual subjective human experience.
People who say LLMs have subjective experience aren't saying they have human-type subjective experience. Nobody who sees an LLM express hunger when role playing as a hungry person thinks that the LLM is actually hungry.
I too can role play as a hungry person despite not being hungry, so there is no reason in either case to conclude that the words produced reflect genuine internal subjective states. The point is that such internal states may still exist.
Sure, but those internal states could then exist for just about any program, or at the very least for any ANN. Nobody in their right minds would have suggested that those might have been aware before the rise of generative models.
When people get the idea that a LLM (or some other generative model) might be aware, surely that impression is because of human-like output. I don't think there's any objective basis for that thinking.
16 GiB was good 11 years ago but laptops were also sold around that time that maxed out at 8 GiB due to limited slots or soldered memory. And lots of laptops are still being sold today with 16 GiB of memory soldered in, with no upgrade possible.
(Some are still being sold with only 8 GiB, too, but I think those may be becoming outliers, or purposefully intended for only light use such as the MacBook Neo.)
Of course it also depends on what you use the computer for. For browsing and streaming 16 GiB is absolutely fine. For other uses 16 GiB can easily become constrained, though, even for tasks that a laptop can otherwise handle fine. And if the RAM in a recent laptop is non-upgradable, will it still be good 5 or 10 years from now?
Exactly, I also want to add that 16 GB at that time would probably correspond to 128 GB today.
My 13 year old laptop, that I still use from time to time (mostly because it is the only windows machine I have) has 8 GB of ram (which truly is barely usable today), it is on the slimmer side and my requirement to have at least 8 GB of ram at that time really limited my options (and there was no option to get more than that in a slim form factor). The norm (not even entry level) was 4 GB and there were tons of options with even less than that.
Police do need authority and respect in order to do their jobs, of course. But at the same time their position of power and their role as maintainers of public peace mean they should have a thicker skin rather than thinner. They should be able to keep their professionalism and composure in the face of adversity better than the average person, and someone being verbally disrespectful towards them is at the lower end of that.
Police officers who think their pride needs to be put first in the name of authority may be common around the world but that doesn't mean reactive cops are a good thing.
> a tiny minority of people care about [...] You know, unlike the minor issues of war and peace and hunger and poverty and economics and minority rights et cetera :)
The complexity and realism of some particular regulation aside, I honestly don't understand dismissing other people's opinions on the grounds that not a lot of people care about them. Should people not have an opinion because lots of other people don't share it? Or should those opinions just not be expressed?
Everybody's got their own personal interests and values. It's natural to express and even vouch for them. If they aren't shared by enough other people, they probably won't make it to the top in a democracy. That's how it works and it's fine. But I don't get the idea that someone shouldn't express an opinion about X because lots of other people don't care about X.
AFAIK mirtazapine shouldn't cause habituation the way actual "sleeping pills" or benzodiazepines do. That's one of the reasons it may be preferable as a sleeping aid, especially in the longer term.
Anecdotally, when I took mirtazapine for sleeping problems, it did sometimes seem to have a stronger effect the first time I took it after not using it for a while. After that the effect stayed stable. Overall it shouldn't cause habituation, and my doctor said as much.
Of course trust your doctor and not strangers on the internet, though.
> After that the effect stayed stable. Overall it shouldn't cause habituation, and my doctor said as much.
Yea so this is where it gets murky for me. I experience some habituation actually. But my actual doctor went like "wtf is this?" and she didn't really mentioned what she knows about it. So on this particular pill my friend is my doctor. Not an ideal situation. I mean, he is an actual doctor but for him to be my doctor in this is a bit fucked up. He knows a lot more about mirtazapine than my GP though since he read up on it.
I suppose it can be quite different for different people. I stopped using it because it often (not always) made me still feel tired and unfocused in the morning, something that apparently also doesn't happen to everyone.
> But my actual doctor went like "wtf is this?"
Different country, but where I live, prescribing low-dose mirtazapine for insomnia appears to be fairly common practice even though it's off-label. I've had it suggested or mentioned by three or four different doctors, including GPs. I also know several other people who have been prescribed it.
The doctors here seem to prefer low-dose mirtazapine as safer over typical CNS depressants such as benzodiazepines for insomnia nowadays, at least if the problem may be longer-term.
So it's not really something particularly weird. Of course different countries also have different medical cultures so I guess it's not surprising if it's not that common in other places.
Mine is of course just another anecdote, but despite seeing some crashes or hangs over the years, I can't remember the last time I've even had to manually fsck my ext4 filesystems, and I'm honestly not sure I've ever seen any unrecoverable corruption on ext4 or ext3. (I'm not saying it doesn't happen, as it certainly can, but I wouldn't think it's that common unless you're operating on a fairly large scale. And before journaling filesystems, crashes were of course a whole different deal.)
I used to keep /home separate on my personal devices (and maybe some other partitions as well because it was "a good practice"), but for more than a decade now my personal devices have had everything on a single large root fs, save for /boot and swap.
I've stuck with rather conservative fs choices and haven't used btrfs much, so I don't know if the story is different there.
What you gain with a single large root partition is of course that you don't need to know in advance how much software you'll want to install under /usr or how much space you'll need for /home or /var.
Multiple partitions can certainly be a good idea from a system management point of view, for cases where people either want to actively manage the system, or when their needs are demanding enough that it's warranted. If I were deploying a server for any kind of real use, yeah, the database or the server logs would certainly not be on the root volume.
But in case of desktops or laptops, most people, even engineers, don't really want to be meticulously managing their personal computers. Nor are most people's personal computing needs that demanding. The convenience of not running out of space on a single partition because you guessed the your future needs wrong wins over.
> Swap on a different physical drive is faster.
> Spread the load across multiple media for performance. That is how server OSes do it, because it is faster.
Of course it is, but in contrast, laptops don't really tend to have separate physical drives on which to put swap. I doubt most people would care enough about swap performance to install a separate high-performance drive for it even in desktops.
With that said, I also do keep a swap partition even on personal computers, and hibernation can be useful. (I think there are some issues with hibernation and encrypted swap in Linux, though, so I haven't bothered with it in a while).
reply