It doesn't matter how amazing your air purifier is if it
sounds like a tornado. If you only run it for a few minutes
before getting annoyed, the purifier is not doing its job.
Top speed is about 53dB. The other speeds range from "barely noticeable" to "is this thing on?"
Combine MERV13/14 with off-the-shelf PC case fans (which
have been optimized for decades to be as quiet and long-lasting
as possible) and the result is an easily-maintainable, quiet,
highly-effective purifier.
More nonsense. My oldest Coways's been in service for ~10 years including ~5 years of services running 24/7/365 near my cat's litterbox.
For a given CFM, with all other things being equal, larger fans (like the ones in Coways) are going to be easier than a bunch of smaller fans. Though, if using an array of Noctuas or similar, I'm sure it could be done.
Many purifiers try to use sensors to determine when to
run the filter, but this fundamentally a flawed approach
and just means you have low quality air 99% of the time.
If you put the Coway on "Auto" and fart next to it, it will spin up to a higher speed for a few minutes. That seems like pretty good detection to me. You do seem to get false positives though/// sometimes it spins up for apparently no reason, so I usually just leave it on low.
The solution to these problem is a less-fine filter. For dust and
pollen, something like MERV13/14 is absolutely fine.
This is certainly true. Though a lot of people buy these to help with pet allergies and want/need HEPA. My understanding is that a lot of the harmful stuff in wildfire smoke also requires HEPA-level filtration, though I might be wrong.
off-the-shelf furnace air filters that are VERY cheap compared to
custom-fit HEPAs which may stop being made the moment that purifier
product line inevitably ceases to exist.
Once again, you're making this stuff up. Now, it's a good point in general -- nobody should buy a rando air purifier without looking into this aspect. Admittedly, standard furnace air filters do have the edge here of course. Corsi-Rosenthal style boxes are going to give you guaranteed filter availability from now until the collapse of civilization. And if you're filtering air in a large or dirty space, then yes, Corsi-Rosenthal boxes are the winners in CADR-per-dollar. I'm building one for my garage workshop before pollen season.
But have you actually taken 30 seconds to look on Amazon and noticed the dozens of third-party filters made to fit the Coways? It's like buying Bose headphones. Yeah, you'll need to replace the earpads eventually, but they've been using the same earpads for like 15 years and there's an ecosystem of cheap 3rd-party versions... not 100% ideal, but also not a problem. As you note, maybe the situation's different outside of the US.
He got honeypotted by a rival superpower and appears to have actively helped to circumvent national security concerns and export controls to the tune of tens of billions of dollars.
My god, how much more spectacular could his mistakes be? Does he have to eat a baby live on TikTok while proclaiming "transistors are dead, we're going back to vacuum tubes" or something?
There are very few human beings in history who could have taken nVidia from zero to megatrillions like he did, but the idea of him continuing to run the company after this is absolute absurdity.
if you think getting powned by chinese intelligence is 'stupid mistakes' you are extremely missinformed about the lengths they will go through and the resources they will deploy.
you really dont understand or know how such a placement works. They can very easily do it to you regardless of how paranoid you are.
everyone can be fished.
the simple fact China had to deploy an operative to get yo him means he and Nvidia are well secured. its an expensive and risky option, not the first by far and intelligence agency would take to get to someone.
He produced a one-in-a-zillion success, and he certainly did it the hard way -- he was apparently unusually hands on. The word "genius" gets thrown around a lot, but I think it's apt.
But.
He put together a string of frankly insane mistakes that are now endangering that 5 trillion dollar company from both the inside and outside. There's no chance in hell that they're going to go from $5T back to 0, but you are not understanding the trouble they're in.
Internally, boinking your direct reports (who also happened to have a string of recent promotions) is incredibly ruinous to a company's internal culture, destroying notions of internal meritocracy. How would you feel if all your hard work was apparently not as important as putting out for the CEO? Do you think that would make you eager to do your best work for that CEO? And that's even before thinking about the insane legal liability.
Externally, he got honeypotted by a foreign government and managed to simultaneously mess with national security and, it seems, circumvent export controls to the tune of tens of billions of dollars. nVidia is in some extremely serious shit, and will have the government breathing down their necks between now and the end of time.
How are you puzzled that people point out the radical hypocrisy?
Musk is 'below the bar'.
Jensen is 'just at the bar' for this maybe slightly above.
I'm not condoning, I'm contextualizing.
He didn't do anything wrong or illegal (that we know of), other than breach of trust with his wife.
This media around this is crazy and slanderous.
All of the press is screaming about 'affair with spy and possibly secrets' , which is very unfair.
She's not a 'spy' - she's a hustler looking to make money by reselling GPUs, and she's 'talking to the Chinese government' as literally they are all compelled to do.
Are Jensen and Elon 'spies' because they talk to the State Dept. and US Gov all the time about goings on in China?
In an ideal world - we would have to contend with Jensen's actions, but I'm doubtful he'd be nerfed for personal foibles, which are not material or public concern.
Assuming that he did not hand over key secrets or information, which I highly doubt and which there are no specific allegations, this is a media / meme wave, completely out of context.
The only saving grace in this, is that the Nvidia board may be friendly to him, in which case, maybe they 'know something' that is not public, and this is the 'nice way out' but even then I'm doubtful.
That's true but only slightly true - and the evidence is that rappers can get away with absolutely horrific behaviour that other musicians cannot 'because the culture'. Kaney just set records for his concerts after releasing 'Hail Hitler' album for gosh sakes.
Elon has normalized his bad behaviour and he controls the board - there's no 'smoke that can make a fire' around him as long as the companies are doing ok. He also aggressively threatens people and very few will speak out against him.
There might be a tinge of racism with the 'honey pot' stuff, and 'chinese spy' has hints of orientalism as well. The headlines are really leaning in to the 'Asian Sex Spy' stuff which is terrible. And I've see posts with rich guys and their Asian wives - not nice.
Fair point, but I don't think it's only partially the case.
Next to the Nvidia chief was a woman named Huang Le, or Alice Huang, an executive of a Singapore-based data center company called Megaspeed, which was poised to buy $2 billion of Nvidia A.I. technology over the next year.
Though Ms. Huang and Megaspeed are little-known players in the A.I. industry, their association with Nvidia and its chief executive has recently become a preoccupation in Washington.
Commerce Department officials have been investigating whether Megaspeed, which has close ties to Chinese tech firms, is helping companies in China sidestep American export restrictions, according to more than a half dozen current and former officials and other people familiar with the companies, who spoke on condition of anonymity to discuss an examination that is not public.
The inquiry, which is active, calls into question how closely Nvidia is tracking where its A.I. chips end up and highlights the possibility of American export laws easily being sidestepped. Megaspeed is also facing scrutiny from Singaporean police, who told The New York Times in a statement that they are investigating the company for breaching local laws, without elaborating further.
It's absurd to suggest that Nvidia would need to secretly ship a handful of chips to some sketchy actor when Nvidia chips have a a mile a line long for sales.
Jensen can sell his inventory 10x over. You have to have buying power just to get a hold of them.
Anything is possible but it would be extra stupid for one of the smartest people in the world to risk everything for literally no reason at all.
> It's absurd to suggest that Nvidia would need to secretly ship a handful of chips to some sketchy actor when Nvidia chips have a a mile a line long for sales.
Consider the opposite possibility - that they were doing this and this this might prove that the "mile long line" for sales is in fact not as long as it looks from the outside.
Fair enough - but there are zero specific allegations against him in terms of national security.
His mistress was talking to the Chines government - literally as all people over there are compelled to do - and - was hustling Nvidia chips - as thousands of people are doing.
The idea that Nvidia would knowingly help people do something bad would require some material evidence before it could be a headline.
> The Chinese spy is bizarre because these guys have corporate security and cooperation of the US government. How was this allowed to happen?
Yeah, this is the most interesting part to me.
It points to the US government knowing about this, and holding onto it as compromat until the time was right to kick him out. But I don't even have a guess as to why this was viewed as the right time...
I agree with your sentiment in general: we should not make assumptions about the parameters of others' relationships.
But in this case, Huang himself explicitly mentioned the hurt that he caused his family. I don't think it gets much clearer than this.
I owe my family and all of you an apology.
I was unfaithful to my wife. My actions hurt her and
our children. There is no excuse for what I did. I am
deeply sorry.
Throughout my life, my family has given me love and
support. They deserve my honesty and loyalty. Instead,
I let them down and damaged the life we built together.
I take full responsibility for what I did and the hurt I
caused, and I accept the criticism that comes with it.
To my wife and family: I love you. I know an apology
cannot undo what I have done. I hope our family will heal
and that, over time, I can earn back your trust.
But you see how you're hypothesizing complex secret dynamics to rationalize what he admits is simple infidelity? When they get a divorce, do you intend to explain how that too is 4d chess and secretly means they really love each other?
overestimating (based on one datapoint?) the number of
people who want to use their devices from within an LLM
chat app,
I don't know about that. "Apps" were a step forward in many ways, but when spending time with older and/or non-technical users they are pretty overwhelmed by the sheer number of apps.
I think they would really like to have one unified, plain-language interface.
I agree with you. That might be why people only use a few apps even if they download 100. It's just really hard to learn the user interface of every single app.
Non-technical users also get REALLY confused/upset about routine app updates that change the UX in any way whatsoever.
They learn these apps differently than "we" do.
They (again, particularly older users who didn't grow up with GUIs) don't understand GUI paradigms, so they compensate by memorizing... even a small change to their memorized flow is highly disruptive because they cannot reason about how to get back onto the happy path.
(It's like the old, pre-GPS-navigation days when you went on a road trip with a stack of printed out directions from MapQuest. It was great, until you missed a turn and had no idea how to get back on track...)
Ideally, yeah, non-technical users could just invest a little time learning common UX paradigms. But a lot of people just won't. And really, GUIs were just a means to an end anyway. Maybe we just need to meet them where they're at with a more unified, naturalistic interface. The word "maybe" is of course doing a lot of work.
Tangent about music recommendations specifically, not the overall discussion...
I've played around a lot with using my own AI to generate music recommendations and build playlists.
> I don't want to use Spotify's AI. I want to use my own AI, who knows me more
It's pretty tough to beat what Spotify (or Apple, or whatever your platform is) knows about you w.r.t. listening habits and preferences. My successful experiments have involved exporting Spotify / Apple Music data into sqlite and a skill that tells my agent how to query this data.
It works pretty well! Still, I'd say it's only roughly on parity with the AI stuff Spotify is working on. The main benefit to rolling my own is being sure (at least, for now) that the recommendations aren't being corrupted by sponsored placement which I'm sure is Spotify's end goal.
I have no real point to this point, just blabbering. Have you cooked up any good solutions via rolling your own?
I think we mostly DID offload basic math, for better or worse.
For software engineering I don't know the right balance.
I've used the "we mostly stopped caring about machine code generated by a compiler; maybe we should stop caring about LLM-generated code as well?" analogy myself.
But even I have to admit there's two big holes in that analogy: if I use a compiler to turn C into machine code... (a) compiler bugs are orders of magnitude rarer than LLM-generated bugs (b) my C code remains the source of truth, or at least source of intent. Whereas with LLM-generated vibecode, there's no source of truth, just an ever-evolving mudball. I (kinda?) like the ideas of "documentation as source of truth" and/or "tests as source of truth" ideas, but I don't think we're nearly at the point of viability.
TL;DR yeah I think actual understanding of what LLMs produce is gonna be important for the forseeable future
Like humans, LLMs also have to expand more effort when dealing with large piles of fragile spaghetti code. Spaghetti code isn't just aesthetically displeasing; there's a objective complexity increase and for large enough apps the increase is exponential. It's hard to imagine this ground truth ever changing.
Like humans, LLMs get into a place where fixing one thing breaks three others. They compensate partially with insanely big test suites, but now fixing one thing breaks three others, and 57 different specs, half of which also need to be updated.
I think roughly three outcomes are possible.
1. LLMs get fast/good/cheap enough that this just doesn't matter. Your 50M LOC monolith is a mess? Yeah, it still takes more effort for the LLM -- but this just means it costs 5 additional seconds and $0.03 instead of $0.02 to ship a new feature.
2. Stepping away from the "code as source of truth" paradigm. What if specs/docs were the vetted source of truth, and we just regenerate the whole 50M LOC monolith every time somebody makes a change, just like we rebuild binaries from C today whenever there's a change? (This probably requires some of the LLM advances from #1. Also, at this point, perhaps we should just skip the intermediate language and generate ASM -- what purpose is human readable source code serving at that point?)
3. Status quo as I see it -- human engineers overseeing projects, both guiding AI and being accelerated by it.
I think #2 is probably the... most ideal future? Though, of course, if we switch to that paradigm, the source-of-truth docs will have to be very thorough and structured, and will probably resemble a programming language of sorts themselves, except describing intents instead of operations.
Well the NSA is probably decent at building software.
Ultimately I'm sure they are but my secondhand understanding from friends is that it's difficult for the US gov't to attract and retain software engineering talent.
It's not just total compensation. Working on classified stuff can be a bit of a career killer, since you can't talk about your past work when looking for a new job.
Plus there's the antipathy toward the current Trump administration, and the general moral unwillingness of many to do defense-related work in general.
Also, last I heard, cannabis use was still a disqualifier for DoD work and/or various security clearances. There are periodic ongoing polygraph tests, not sure about ongoing piss tests. The government is big; I'm sure the exact policies differ from department to department. But this turns a lot of people off.
Of course, every single thing I've said above might also be viewed as a positive by some engineers. Maybe they attract enough "true believers" to more than make up for the other factors.
I would argue that US intelligence, especially on the technical side of things has put emphasis on privatization of these sectors.
US intelligence knows they can't go to policy makers and get funding for billion dollar domestic spy programs. So they stand up private companies and give them smallish grants at the beginning to get them on their feet, i.e Flock, Palantir, Signal Foundation, EFF, cloudflare, oracle. I could go on and on about orgs that their start off the back of US intelligence, whether through direct funding or through their alumni network.
The US Department of Defense has a long tradition of private public partnerships, but the privatization of intelligence has been accelerated in the last decade.
So while the NSA might not be able to attract talented SWEs for gov salaries to work on defense, their vendors that owe them their existence can.
Parent poster is exactly the kind of person who argues that using tax dollars for school is wrong because they, personally, do not have kids. Because they can't realize that they indirectly benefit from it.
John Green: "So let me explain why I like to pay taxes for schools, even though I don't personally have a kid in school: It's because I don't like living in a country with a bunch of stupid people."
We (the public) have already been forced to accept that students' education is directly linked to the property taxes. In other words, students only deserve the education linked to an average the the worth of the community. Its some of the grossest discrimination against student, based purely on family and locality worth.
Poor community? Sorry Timmy, your schools SUCK.
Rural community? Get to farming Timmy. You dont deserve a GOOD education either.
Living in a rich suburb of a big city? Congrats, your schools are well funded and low student:teacher count. Your education will actually be pretty damned good.
"For a purifier of its capabilities, the Mighty is notably quiet" - https://www.nytimes.com/wirecutter/reviews/best-air-purifier...
Alternatively, just take a glance at the ~30K reviews on Amazon. - https://www.amazon.com/dp/B01728NLRG
Top speed is about 53dB. The other speeds range from "barely noticeable" to "is this thing on?" More nonsense. My oldest Coways's been in service for ~10 years including ~5 years of services running 24/7/365 near my cat's litterbox.For a given CFM, with all other things being equal, larger fans (like the ones in Coways) are going to be easier than a bunch of smaller fans. Though, if using an array of Noctuas or similar, I'm sure it could be done.
If you put the Coway on "Auto" and fart next to it, it will spin up to a higher speed for a few minutes. That seems like pretty good detection to me. You do seem to get false positives though/// sometimes it spins up for apparently no reason, so I usually just leave it on low. This is certainly true. Though a lot of people buy these to help with pet allergies and want/need HEPA. My understanding is that a lot of the harmful stuff in wildfire smoke also requires HEPA-level filtration, though I might be wrong. Once again, you're making this stuff up. Now, it's a good point in general -- nobody should buy a rando air purifier without looking into this aspect. Admittedly, standard furnace air filters do have the edge here of course. Corsi-Rosenthal style boxes are going to give you guaranteed filter availability from now until the collapse of civilization. And if you're filtering air in a large or dirty space, then yes, Corsi-Rosenthal boxes are the winners in CADR-per-dollar. I'm building one for my garage workshop before pollen season.But have you actually taken 30 seconds to look on Amazon and noticed the dozens of third-party filters made to fit the Coways? It's like buying Bose headphones. Yeah, you'll need to replace the earpads eventually, but they've been using the same earpads for like 15 years and there's an ecosystem of cheap 3rd-party versions... not 100% ideal, but also not a problem. As you note, maybe the situation's different outside of the US.
reply