This feels so pretentious. People can keep it closed or open for whatever reason they want, and it has no correlation to how they solve problems or learn.
Personally, I like it open when I'm feeling social and in a good mood, and close it when it's noisy outside and/or I need to hunker down and focus for a bit without distractions. That doesn't say anything about understanding or solving problems, other than 'sometimes people need quiet to focus' which is not a very shocking revelation.
The entire page is underwhelming. For someone in the US, I walked away with basically no new information other than some stuff will enter public domain at new years.
> In our advent-style calendar below, find our top pick of what lies in store for 2026. Each day, as we move through December, we’ll open a new window to reveal our highlights! By public domain day on January 1st they will all be unveiled — look out for a special blogpost from us on that day. (And, of course, if you want to dive straight in and explore the vast swathe of new entrants for yourself, just visit the links above).
Good PC case fans are incredibly quiet, move a lot of air and last forever. That's the nice thing about having some many companies competing over optimizing them for so long.
I have a similar setup, just premade from CleanAirKits.
Allergies are gone, house has far less dust and cat hair floating around.
Can't do anything for wood smoke though, that would need carbon filters - which run out quite quickly.
For anyone interested in some gaming benchmarks, Gamer's Nexus (a reputable source with good methodology) has some numbers here: https://www.youtube.com/watch?v=ovOx4_8ajZ8
Based on their results, it sounds like there's still quite a way to go Linux gaming/Proton (ie: very inconsistent frametimes on Nvidia hardware), but it's definitely been taking steps in the right direction.
Do note that they make it very, very, very clear that their results are preliminary, and while they've put a whole lot of work into setting up benchmarking on Linux, they're not at all sure that they've got it all correct.
> ie: very inconsistent frametimes on Nvidia hardware
Yeah, Nvidia on Linux for non-"compute" use has always been a terrible, godawful shitshow. Given Nvidia's recent and fairly-clear disinterest in the "selling graphics cards for people to play video games with" market, [0] I can't imagine things will get consistently better anytime in the near future. [1]
[0] Why sell that silicon to video gamers when you can sell it to cryptominers and -these days- "AI" companies for a much, much, much higher profit?
[1] I mean, it took them how many Windows driver revisions for them to release somewhat-non-garbage drivers for their spanking-new fancy-ass 5000-series cards? And Windows is the only consumer OS that they care about! For video game use, the Linux drivers are gonna be starved for development resources, and -unlike ATi/AMD's drivers- noone in the world can work on them but Nvidia.
> Do note that they make it very, very, very clear that their results are preliminary,
Yeah, I'm not entirely convinced some of the results they're seeing aren't caused by their methodology. I don't think they are either.
I moved a gaming pc with a 4070Ti from Win10 to Cachy 3 weeks ago and have been purposefully testing out various games to see if it's workable; I'm about 50 hours and 15 games deep now and the only thing that doesn't work reliably is HDR. Outside of that I haven't run into any issues I haven't seen on Windows as well.
I've never run into any issues with my GTX 1070 and proprietary drivers. It's ran perfectly well for almost a decade. Like when people complain about nvidia drivers I don't even know what they mean. Do you run into artifacts, or crashes, or kernel panics, or what?
What’s with the Nvidia on Linux FUD like this is the mid-00s? I’ve run Nvidia cards on Linux for the better part of the last decade and it has been a pretty similar experience to when I was on Windows prior to that.
Nvidia’s Open Kernel Modules are good so far and the in-kernel Nova driver project also seems promising though some way off. I’m running a 5000 series card with Nvidia OKMs and so far it has been a really smooth experience.
I actually get higher FPS in graphically demanding games due to vulkan producing more optimized shaders, most of the lag actually comes from context switching and translation for directx, it's absolutely GREAT one dx12 due to how little translation there is.
I'm currently working on few very targeted optimizations for several hotspots I've found out from messing around so it will be interesting if I can solve those horrible stutter issues on cyberpunk since if I can fix it (in a ghetto and janky way), so can valve.
I can also achieve higher fps in games like overwatch (dx12) out of the box on nvidia on proton experimental which is surprising as 4 years ago the input latency was unbearable and I had drops as low as 30 fps, now I can achieve consistent 600 fps with minimums of 450 whereas on windows I get as low as 220fps and averages of 500ish. I do have anticheat related drops to <300fps due to the amount of translation happens when they decide to scan memory, registry and whatnot although it lasts <1s and doesn't happen during games it seems.
That just isn't true. While it is very good now, it is not faster and more stable than Windows. I have performance issues on Linux that just don't happen on Windows.
Well that is the issue. The experience varies quite a lot depending on a number of factors. Whereas on Windows it doesn't really vary.
I have an all AMD machine and almost all the games will run the same or better on Windows. I have friends that have tried gaming on Linux and all of them have found the experience worse.
I did run a win debloat script from and use a local account so I don't have the Windows Spyware running in the background so that may make a difference.
Just an aside. I've been using Linux for quite a while now (over-20 years) and the biggest issue is that the community constantly misleads new users about the experience of moving from Windows to Linux. The latest iteration of this has been gaming.
Perhaps it’s just evidence that anecdata isn’t some universal truth.
I’ve had the complete opposite experience for the vast majority of games, where in most cases performance for me has been better on Linux than it was on Windows (can’t compare like for like now as I no longer have a Windows install outside of a VM). Friends of mine experience weird mid-session crashes and hangs on Windows that I’ve never had on Linux. I’m running an Nvidia GPU which is supposedly some kind of Linux boogeyman, but have had only one issue with EDID of a specific monitor and that’s it. Just my experience YMMV.
> Perhaps it’s just evidence that anecdata isn’t some universal truth.
This isn't though. I have hard numbers. I've actually measured the performance. You get 5-20 FPS less and often more input latency and stutters (1%, 0.1% lows). If the machines doesn't well with Linux, it can be much worse.
Basically on HN whenever you express an opinion based on a significant amount of experience. You get someone basically saying "this is anecdote". There is a difference between "an anecdote" and "I've actually have a huge amount experience with this stuff.
Have you produced an exhaustive survey across a wide range of hardware and driver and display manager combinations? I’m happy to be an outlier here but my own experience doesn’t match with what you described hence my reply.
If I admit to anything less than doing a gamer nexus style benchmarking suite you will just claim it is an anecdote.
I have actually tested on a number of different distros and display managers and at least two different video card chipset manufacturers. No it isn't exhaustive, but it decent enough sample size to determine that the claim that Linux performs better than Windows isn't true. Even if it is the case,the results are so variable you are better just using Windows because things are more consistent.
I am saying this BTW as someone that first started using Linux in the early 2000s. I think gaming now is really good on Linux. Is it better than Windows? Well I don't have to run Windows now to play games and that is good enough for me.
And I get totally different results. It not just the distro. It the version of the Kernel, the version of proton, whether you are using X or Wayland etc. Etc. Etc.
The very point I am making is that it is so variable. So posting benchmarks pretending that it proves anything is asinine.
I won't even get into all the other issues with the mouse getting lost on some games, text being too small/to large. Having to fuck around with LD_PRELOAD flags and loads other gumpth that is never mentioned on a YouTube video.
Your claim was that you tested and "You get 5-20 FPS less and often more input latency and stutters (1%, 0.1% lows)", "it decent enough sample size to determine that the claim that Linux performs better than Windows isn't true". That you tested so well, that it can't be considered to be anecdata. You claimed that it is universal truth.
So I provided you with solid data from testers, who found many cases in which Linux was on par or faster than Windows.
There seems to be little evidence of that, at least from a reputable source.
For some games it can be. For some games Proton performs far worse than Windows. It's not steady across the board. And some have stability issues, bugs, major performance problems, or just flat out don't work.
I want Proton to be the future as well, but I think it's important not to oversell it as a drop-in replacement either.
EDIT: GN highly recommends against apples-to-oranges comparisons of the two, but even looking at their own data for AMD cards (with exact same CPU, RAM, and motherboard) it clearly shows Proton being behind on the order of 6-15%. Not a lot, but not ahead either. You can compare the numbers for the AMD cards against this video's here: https://www.youtube.com/watch?v=yP0axVHdP-U
EDIT 2: Instead of just down-voting because the result makes you unhappy, how about responding with well-sourced proof otherwise?
> There seems to be little evidence of that, at least from a reputable source.
I dunno. I remember a little while back some reviewers got a hold of both a Windows version and Linux version of a handheld gaming machine that had exactly the same hardware. The conclusion reached was that the Linux version was better in nearly every way.
As I remember it, a little while after this happened, some muckety-muck in the Gaming Division in Microsoft announced something like "We're making a new committment to consistent, high performance in Windows on handheld gaming devices! We're going to ensure all those little game-spoiling roadblocks are removed!". Which, like, good job making it NOT look like you're spasmodically reacting to bad press, guy.
> Handhelds and desktop PCs can perform differently.
The handhelds we're talking about are -essentially- a low-power [0] laptop in a tiny case. Again, we're talking about exactly the same hardware, that provides substantially worse performance when Windows is the OS than when Linux is the OS.
> I want Proton to be the future as well, but I think it's important not to oversell it as a drop-in replacement either.
Proton needs to be understood as a temporary solution to make the back catalogue on Linux comparable to Windows.
Eventually there is a tipping point and most games will have native Linux binaries. Once that happens all developers will gradually follow suit to avoid being left behind. Perhaps Valve's latest hardware efforts will finally bring this about.
Proton will still exist for older games and as hardware continues to become more powerful, loss of performance won't matter much.
From personal experience, all of the games that I personally play seem to work at least as well if not better on Linux than Windows. The only exception is FSR4, but that ought to be fixed soon.
FWIW, I wasn't the one who downvoted you. I can't even downvote here as I'm apparently "karmically broke". I have 468 karma, not sure how much I even need to downvote...
I think the TUV is mostly about safety equipment (light, breaks) that don't pass the two year check (they probably also test the alignment and other stuff), and your link might be about engine/transmission/power train defects.
Oh man, I'd love for a reduction in that. I sort-of deal with it.
As a kid I had nearly daily migraines - go into a very dark quiet room and be happy when I fell asleep, since I knew the migraine wouldn't be there when I woke up.
These days it's just headaches, 99% sure it's muscle tension in my neck. Kinda doubt GLP-1 could do anything for that, but I'd be pleasantly surprised....
I had very similar issues, both as a kid and an adult. I now take 500mg of Magnesium (Malate or Glycinate. NOT Citrate, as that is a laxative) nearly every night, and my headaches are almost completely gone now. I still get one every once in a while, but it's infrequent enough that I no longer worry about my use of Ibuprofen to manage it.
Get your magnesium level checked. Those that suffer from chronic migraines often are deficient in magnesium and will see a significant reduction in events after 4-6 weeks on a daily Magnesium supplement.
Agreed. Once it becomes commercially viable to start building things in space, it'll take off on its own. There will be constant pressure to build faster, safer, more capable craft. Whether that will lead to something like FTL isn't possible to know, but at the very least it's a step towards a space-faring civilization.
I think you misunderstood the person you're responding to. They did not say there was some higher force beyond the physical pieces.
What they're saying is that the brain is really really complicated and our understanding of biology is far too rudimentary right now to be saying "yes, absolutely, 100% sure that we know the nature of consciousness from this one measurement of one type of signal".
* Neurons are very complex and all have unique mutations from one another
* Hundreds of other types of cells in the brain interact with them and each other in ways we don't understand
* The various other parts of the body chemically interact with the brain in ways we don't understand yet, like the gut microbiome
Trying to flatten all of consciousness to one measurement is just not sufficient. It's like trying to simulate the entire planet as a perfect sphere of uniform density. That works OK for some things but falls apart for more complex questions.
I get that, but there's no need to complicate things unnecessarily.
I'll make an even stronger claim, that biological brains are not only computers, but that they operate in binary, as well. Active and inactive - the mechanisms that trigger activation are incredibly nuanced and sophisticated, but the transfer of information through the network of biological neurons is a matter of zeroes and ones. A signal happens, or doesn't. Intensity, from a qualia persepective, ends up being a matter of frequency and spread, as opposed to level of stimulation. That, in conjunction with all sorts of models of brain function, is allowing neuroscience to make steady, plodding progress in determining the function and behavior of different neurons, networks, clusters, and types in the context in which they are found.
All else being equal, at the rate neuroscience is proceeding, we should be able to precisely simulate a human brain, in functionally real-time, using real brain networks as models, by around 2040. We should have a handle on every facet of brain chemistry, networking, electrical signaling, and individual neuronal behavior based on a comprehensive and total taxonomy of feature types down to the molecular level.
Figure out the underlying algorithms and you can migrate those functional structures to purely code. If you can run a mind on code, then it doesn't matter whether you're executing a sequence of computations in a meat brain, in a silicon chip, or using a billion genetically engineered notebook monkeys to painstakingly and tediously do the computations and information transfer manually, passing sheets of paper between them. ( the monkeys, of course, could not operate in real time.)
There won't be another significant phase change, like we saw from hydraulics to computation equivalence. Computation is what it actually, physically is, at the level of electrical signals and molecular behaviors. It's just extremely complex and sophisticated and elegantly interwoven with the rest of the human organism.
Brain gut interactions aren't necessary for human subjective experience or cognition. You could remove your brain entirely from your skull, while maintaining a equivalent level of electrical and chemical signaling from an entirely artifical platform of some sort, and as long as the interface between the biological and synthetic maintains the same signaling frequency, chemistry, and connectivity, then it doesn't matter what's on the synthetic end.
There are independently intelligent aspects to things like the gut biome, and other complex biological systems. Those aren't necessary for brains to do what brains do, except in a supportive role. Decouple the nutrition and evolutionary drives from the mind, and you're left with a fairly small chunk of brain - something like 5B neocortical neurons is the bare minimum of what you'd need to get human level intelligence. Everything on top of that is nice to have, but not strictly necessary from a proof of concept perspective.
Isn't it strange that the most elusive thing in the universe - consciousness - so neatly fits the scientific framework of today? It doesn't mean that a pretty good imitation of mind can't be created in the machine world, unfortunately.
I don't think it's strange at all that the scientific framework of today is capable of investigating physical phenomena in the real world. It would be strange if it _couldn't_.
Agree completely. The brain is so incredibly complex that we've barely scratched the surface. It's not just neurons, which are very complex and vary wildly in genetics between them - it's hundreds of other helper cells all interacting with each other in sometimes bizarre ways.
To try to boil down it all to any simple signal is just never going to work. If we want to map consciousness it's going to be as complex as simulating it ourselves, creating something as dense and detailed as a real brain.
I don't think it's anything other than electric activity, but it's clearly not "some electrical signal". It's the totality of them. They are many, and complicated. And they seem to be required for consciousness. Doubt there's any proven conscious state in a human, lacking electrical activity in the brain.
Re just electrical activity, I think you can add empirical evidence it's chemical as well as beer and other substances can affect your perception of reality.
There was an AMA about conjoined twins on Reddit a couple of years ago, and one of the interesting parts was that they could each sense how the other twin is feeling in terms of emotions. This is due to a lot of emotional states being based on hormones that flow through their shared blood stream.
If we ever get to the point of having a tool that could do something that complex, we're well past the point of using human-written operating systems or using M-series processors.
Hmm I seem to recall Karpathy talk about the next era of software being like this: Essentially everything is made on demand exactly the way the user expects or something like that. it would definitely be a paradigm shift that requires rethinking how we approach software.
What do we really need? If you look at what Ai can solve today, smaller problems seem to be solved really well by AI. So just an AI with a large enough context, adequate speed to made the changes in near real time and a reasoning ability that can better handle problems vs whats available today might be adequate to realize this relality in some way.
What we call "AI" right now could never do something of the complexity we're discussing. It has no memory, no experiences, no consciousness, no capability to continually learn, no ability to abstractly understand ideas and apply them.
It's text probabilities in a box. Until something comes around that is NOT an LLM, it doesn't matter how much memory or training data you throw at it, the fundamental idea can't work.
Personally, I like it open when I'm feeling social and in a good mood, and close it when it's noisy outside and/or I need to hunker down and focus for a bit without distractions. That doesn't say anything about understanding or solving problems, other than 'sometimes people need quiet to focus' which is not a very shocking revelation.
reply