Need to let loose a primal scream without collecting footnotes first? Have a sneer percolating in your system but not enough time/energy to make a whole post about it? Go forth and be mid: Welcome to the Stubsack, your first port of call for learning fresh Awful youāll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cutānāpaste it into its own post ā thereās no quota for posting and the bar really isnāt that high.
The post Xitter web has spawned soo many āesotericā right wing freaks, but thereās no appropriate sneer-space for them. Iām talking redscare-ish, reality challenged āculture criticsā who write about everything but understand nothing. Iām talking about reply-guys who make the same 6 tweets about the same 3 subjects. Theyāre inescapable at this point, yet I donāt see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldnāt be surgeons because they didnāt believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I canāt escape them, I would love to sneer at them.
(Semi-obligatory thanks to @dgerard for starting this)
Today in āPromptfondler fucks around and finds out.ā
So Iām guessing what happened here is that the statistically average terminal session doesnāt end after opening an SSH connection, and the LLM doesnāt actually understand what itās doing or when to stop, especially when itās being promoted with the output of whatever it last commanded.
Shlegeris said he uses his AI agent all the time for basic system administration tasks that he doesnāt remember how to do on his own, such as installing certain bits of software and configuring security settings.
Emphasis added.
āI only had this problem because I was very reckless,ā he continued, "partially because I think itās interesting to explore the potential downsides of this type of automation. If I had given better instructions to my agent, e.g. telling it āwhen youāve finished the task you were assigned, stop taking actions,ā I wouldnāt have had this problem.
just instruct it ābe sentientā and youāre good, why donāt these tech CEOs undersand the full potential of this limitless technology?
so I snipped the prompt from the log, and:
āÆ pbpaste| wc -c 2063
wow, so efficient! Iām so glad that we have this wonderful new technology where you can write 2kb of text to send to an api to spend massive amounts of compute to get back an operation for doing the irredeemably difficult systems task of initiating an ssh connection
these fucking people
Assistant: I apologize for the confusion. It seems that the 192.168.1.0/24 subnet is not the correct one for your network. Letās try to determine your network configuration. We can do this by checking your IP address and subnet mask:
there are multiple really bad and dumb things in that log, but this really made me lol (the IPs in question are definitely in that subnet)
if it were me, Iād be fucking embarrassed to publish something like this as anything but a talk in the spirit of wat. but the promptfondlers donāt seem to have that awareness
But playing spicy mad-libs with your personal computers for lols is critical AI safety research! This advances the state of the art of copy pasting terminal commands without understanding them!
I also appreciated The Register throwing shade at their linux sysadmin skills:
Yes, we recommend focusing on fixing the Grub bootloader configuration rather than a reinstall.
wat
Thanks for sharing this lol
itās a classic
similarly, Mickens talks. if you havenāt ever seen āem, thatās your next todo
so, Iāve always thought that blindās āweāll verify your presence by sending you shit on your corp mailā (which, yāknow, mail logs etcā¦) is kinda a fucking awful idea. but!
this is remarkably fucking unhinged:
A medium nationās worth of electricity!
from the (current?) prick-in-chief at YC in this post:
and everyone in our industry owes a debt to open source builders
nice of you to admit it. now maybe pay down some of that debt by using sending of your piles of money to those projects
oh, whatās that, you only want to continue taking from it and then charging other people service rent, without ever contributing back? oh okay then
a debt in the special non-financial sense
the miracle of VC is learning to write like ChatGPT
ah, the special non-financial sense, you mean blood debts? I think many of us will be happy to hold tan et co to account for blood debts
Theyāre getting paid in āexposureā
Folks, I need some expert advice. Thanks in advance!
Our NSF grant reviews came in (on Saturday), and two of the four reviews (an Excellent AND a Fair, lol) have confabulations and [insert text here brackets like this] that indicate that they are LLM generated by lazy people. Just absolutely gutted. Itās like an alien reviewed a version of our grant application from an parallel dimension.
Who do I need to contact to get eyes on the situation, other than the program director? We get to simmer all day today since it was released on the weekend, so at least I have an excuse to slow down and be thoughtful.
I havenāt had to report malfeasance like that, but if that happened to me, I would be livid. Iād start by contacting the program officer; Iād also contact the division director above them and the NSF Office of Inspector General. I mean, that level of laziness canāt just have affected one review! And, for good measure, Iād send a tip to 404media, as they have covered this sort of thing. That might well go nowhere, but it canāt hurt to be in their contact list.
Total amateur here, but from quickly reviewing the process it looks like the program officer would be your primary point of contact within NSF to address this kind of thing? But then I would assume they read the reviews themselves before passing them back to you so I would hope they would notice? The bit of my brain thatās watched too much TV would like to see them answer some questions from an AI skeptic journalist, but thatās not exactly a great avenue for addressing your specific problem.
Mostly commenting to make it easier to keep track of the thread tbh. Thats some kinda nonsense youāre dealing with here.
(Another post so soon? Another post so soon.)
āGen AI competes with its training data, exhibit 1,764ā:
Also got a quick sidenote, which spawned from seeing this:
This is pure gut feeling, but I suspect that āAI trainingā has become synonymous with āart theft/copyright infringementā in the public consciousness.
Between AI bros publicly scraping against peopleās wishes (Exhibit A, Exhibit B, Exhibit C), the large-scale theft of data which went to produce these LLMsā datasets, and the general perception that working in AI means you support theft (Exhibit A, Exhibit B), I wouldnāt blame Joe Public for treating AI as inherently infringing.
I researched cool topics using ChatGPT, Claude, Google
Thatās not what research means, you embossed carbuncle.
I linked NotebookLM to the Wikipedia entry of each topic and generated the podcast audio
Itās fucking James Somerton with extra steps!
https://xcancel.com/karpathy/status/1841848120897912967#m
jfc
Edit: replaced screenshot of tweet + description with link to tweet
jesus christ, as if āpodcastsā consisting of random idiots regurgitating wikipedia werenāt bad enough as it is
aaaaaaaah
aaaaaaaaaaaaaaaaaaaaaa
AAAAAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH
ācuratedā
it shouldnāt surprise me that the dipshit who was massively involved in āsolving languageā doesnāt understand the meaning of words, but grrrrr
(and I say that as an armchair linguist who understands that language is as people use it (fuck prescriptivism))
dipshit who was massively involved in āsolving languageā
āIn the what now?ā, he said, voice trembling with a mixture of horror and excitement
2015 - I was a research scientist and a founding member at OpenAI.
proudly displayed on his blog timeline
Wait, did OpenAI claim somewhere they will āsolve languageā?
no, thatās a personal extrapolation/framing characterising some of the shit Iāve seen from these morons
(they engaged with very few linguists in the making of their beloved Large Language Models, instead believing they can just data-bruteforce it; this plan gone as well as has been observed)
Ex-headliners Evergreen Terrace: āEven after they offered to pull Kyle from the event, we discovered several associated entities that we simply do not agree withā
the new headliner will be uh a Slipknot covers band
organisers: āWe have been silent. But we are prepping. The liberal mob attempted to destroy Shell Shock. But we will not allow it. This is now about more than a concert. This is a war of ideology.ā yeah you have a great show guys
I wish them a merry cable grounding fault and happy earth buzz to all
By āliberal mobā he means āpeople who asked for their money back and arenāt coming anymoreā
So the ongoing discourse about AI energy requirements and their impact on the world reminded me about the situation in Texas. It set me thinking about what happens when the bubble pops. In the telecom bubble of the 90s or the British rail bubble of the 1840s, there was a lot of actual physical infrastructure created that outlived the unprofitable and unsustainable companies that had built them. After the bubble this surplus infrastructure helped make the associated goods and services cheaper and more accessible as the market corrected. Investors (and there were a lot of investors) lost their shirts, but ultimately there was some actual value created once we were out of the bezzle.
Obviously the crypto bubble will have no such benefits. Itās not like energy demand was particularly constrained outside of crypto, so any surplus electrical infrastructure will probably be shut back down (and good riddance to dirty energy). The mining hardware itself is all purpose-built ASICs that canāt actually do anything apart from mining, so itās basically turning directly into scrap as far as I can tell.
But the high-performance GPUs that these AI operations rely on are more general-purpose even if theyāre optimized for AI workloads. The bubble is still active enough that there doesnāt appear to be much talk about it, but what kind of use might we see some of these chips and datacenters put to as the bubble burns down?
But the high-performance GPUs that these AI operations rely on are more general-purpose even if theyāre optimized for AI workloads. The bubble is still active enough that there doesnāt appear to be much talk about it, but what kind of use might we see some of these chips and datacenters put to as the bubble burns down?
If those GPUs end up being used for Glaze and Nightshade, Iād laugh like a hyena.
App developers think thatās a bogus argument. Mr. Bier told me that data he had seen from start-ups he advised suggested that contact sharing had dropped significantly since the iOS 18 changes went into effect, and that for some apps, the number of users sharing 10 or fewer contacts had increased as much as 25 percent.
aww, does the widdle appās business model collapse completely once it canāt harvest data? how sad
this reinforces a suspicion that Iāve had for a while: the only reason most people put up with any of this shit is because itās an all or nothing choice and they donāt know the full impact (because itās intentionally obscured). the moment you give them an overt choice that makes them think about it, turns out most are actually not fine with the state of affairs
@froztbyte @jwz Not the biggest Apple fan, but you got to give them credit: with privacy changes in their OSs, they regularly expose all the predatory practices lots of social media companies are running on.
As previously mentioned, the āBehind the Bastardsā podcast is tackling Curtis Yarvin. Iām just past the first ad intermission (why are all podcast ads just ads for other podcasts? Itās like podcast incest), and according to the host, Yarvin models his ideal society on Usenet pre-Eternal September.
This is something Iāve noticed too (I got on the internet just before). Thereās a nostalgia for the āoldā internet, which was supposed to be purer and less ad-infested than the current fallen age. Usenet is often mentioned. And Iāve always thought thatās dumb because the old internet was really really exclusionary. You had to be someone in academia or internet business, so you were Anglophone, white, and male. The dream of the old pure internet is a dream of an internet without women or people of color, people who might be more expressive in media other than 7 bit ASCII.
This was a reminder that the nostalgia can be coded fascist, too.
I have a lot of time for nostalgia about older versions of the web, but it really ticks me off when people who actively participated in making the web worse start to indulge in nostalgia about the web. Doesnāt Yarvin get a lot of money from Peter Thiel?
There were women and people of colour on the old web, and feminists and radical anti-racists too - they were just outnumbered and outgunned. One of the earliest projects listed on the cyberfeminism index are VNS Matrix, who were ācorrupting the discourseā way back in 1991.
Following up from this truth bomb: https://awful.systems/comment/4877052
@Soyweiser: Sorry AGIbros, not even the Dutch believe AGI is near.
For your delectation, here are the HN comments
Iām in the other camp: I remember when we thought an AI capable of solving Go was astronomically impossible and yet here we are. This article reads just like the skeptic essays back then.
Ah yes my coworkers communicate exclusively in Go games and they are always winning because they are AI and I am on the street, poor.
Thereās not that much else to sneer at though, plenty of reasonable people.
Hereās the lobste.rs disucssion: https://lobste.rs/s/4xzxqk
oh i dunno, there was
Honestly - Computer Science has given us more clues about how the human mind might work than cognitive science ever did.
This remark is actually part of a long fight between CS and CS people. And it is really frustrating in various ways, as CS always thinks they did better than CS while being blind of the actual accomplishments of CS they donāt know and just how complex the subject matter is. It is an annoying failure to communicate between both disciplines. (A lot of people donāt fall victim to this btw, but it can be really annoying to encounter a āOur CS is good, and theirs is bad because strawmanā, who often donāt even realize that various words have different meanings in the different fields).
For the record, I think the Counter-Strike people are correct on this one, mainly because heuristically Confederate States advocates are wrong by default.
Exactly the problem im talking about. What about all the good things the confederate state ā¦ no wait.
Hopefully this doesnāt break the rules. But where can I find some educational podcasts that arenāt overly capitalist, reactionary, rationalist, or otherwise right-leaning or authoritarian in nature.
I want to specifically avoid content like Lex Friedman, Huberman, Joe Rogan, Sam Harris. That sounds good on the surface but goes down a rabbit hole of affirming reactionary bias.
Iām not amazing with words, so I hope what Iām saying makes sense. Thanks.
If you want interesting historical deep dives, I always enjoy Dig - the history podcast. Well researched by actual scholars, which goes hand in hand with the episodes not dropping that often.
Behind the Bastards is very easy to listen to and usually focuses on documenting the bad shit that various reactionary and fascist figures have done (in a humorous manner ā the host was a writer for Cracked during its peak). a couple of the most recent episodes have covered some of the same topics we talk about in SneerClub and TechTakes, and theyāre well worth a listen even if you know the subject matter well. I havenāt checked it out yet, but I think It Could Happen Here is a spin-off with the same main host thatās also broadly anti-fascist.
e: also, and I had to look this up cause I keep switching podcast apps: I Donāt Speak German is also good, and my co-admin David was on it (episode 82? I swear it was more recent than thatā¦ David were you on more than once?)
i was! ep 82 and 85
IDSG has slowed down a lot cos Danielās got shit going on right now, but they try to do one when they can
I tend to like āCool People Who Did Cool Stuffā more than āBehind the Bastardsā. Need some nugget of hope in these dark days. A lot of the cool people have been downright inspiring.
My daily podcast is āIt Could Happen Hereā, but some other mainstays in the educational side include:
- Live Like the World is Dying
- Strangers in a Tangled Wilderness
- Itās Going Down
- Final Straw Radio
- Reaction (especially liked her dives on the Pinkertons and āThe Business Plotā)
- Srsly Wrong [unrelated to the similarly named thing]
- The Iron Dice
- Bad Hasbara
- Frontline Herbalism if you like plants
Most everything from Cool Zone Media is going to be pretty decent. Havenāt listened to the whole catalogue, but Ed Zitron of Better Offline is an established nonmember (as far as I know) friend of the sneer and Behind the Bastards is truly excellent.
Maintenance Phase is an excellent examination of diet and health grifters, and Mikeās others (Youāre Wrong About and If Books Could Kill) are also pretty excellent.
I also want to spotlight Wittenburg to Westphalia, a history podcast ostensibly about the wars of the reformation and the social and economic chamges of the early modern period. But in order to really give a sense of how dramatic those changes are, he has so far provided only an incredibly thorough examination of medieval European society from the politics to economics and social structures. He has an episode about unfree labor that I found particularly interesting.
Second on Maintenance Phase! I marathonned it on a road trip a couple of days ago, and not only is it a well-researched and a fun listen, youāll discover that so much of the stuff Aubrey and Michael discuss is directly congruent to our typical subjects. Canāt recommend it enough.
Theyāve inspired me to work on an effort post for MoreWrite, tentatively titled, āA Unified Theory of Bullshitter-Driven Social Diseases.ā
Which isnāt going to be as pompous as it sounds, I promise!
Sounds great! Looking forward to reading it
as seen via jwz, the tail wagging the dog continues (archive) at mozilla
āif you canāt beat 'em, join 'emā but the wrong way around. I guess they got tired of begging google for money?
And, for the foreseeable future at least, advertising is a key commercial engine of the internet
this tracks analogously to something Iāve been saying for a while as well, but with some differences. one of the most notable is the misrepresentation here of āthe internetā, in the stead of āall the entities playing the online advertising game to extract from everyone elseā
Thanks I hate it.
[Advertising is] the most efficient way to ensure the majority of content remains free and accessible to as many people as possible.
Content is a scarce resource yāknow. Heaven forbid the content farms go out of business; or we might end up having to read Sherlock Holmes isekai fanfiction rather than a content farmās two paragraphs and three screen-fulls of ads surrounding the tweet du jour. That would be
terribleactually quite nice.We know that not everyone in our community will embrace our entrance into this market. But taking on controversial topics because we believe they make the internet better for all of us is a key feature of Mozillaās history
WTF. How is it possible for a company to be this self-congratulatory about entering the advertising space?! Someone needs to fork Firefox.
the predominant fork (afaik) is called LibreWolf
some indications that it is not atm entirely resilient to the upstream bullshit, but Iām curiously eyeballing whether that changes
personally, Iām hoping on servo
this aged like fine milk (granted itās from 2019)
whatās the over/under on the spruce pine thing causing promptfondlers and their ilk to suddenly not be able to get chips, and then hit a(n even more concrete) ceiling?
(I know there may be some of the stuff in stockpiles awaiting fabrication, but still, canāt be enough to withstand that shock)
If weāre lucky, itāll cut off promptfondlersā supply of silicon and help bring this entire bubble crashing down.
Itāll probably also cause major shockwaves for the tech industry at large, but by this point I hold nothing but unfiltered hate for everyone and everything in Silicon Valley, so fuck them.
Building 5-7 5GW facilities full of GPUs is going to be an extremely large amount of silicon. Not to mention the 25-35 nuclear power plants they apparently want to build to power them.
So on the list of things not happeningā¦that would be 25-35 reactors, as long as cooling is available you can just put them in one place. 5GW is around size of largest european nuclear powerplants (Zaporozhian, 5.7GW; Gravelines, 5.4GW; six blocks each) or around energy consumption of decently-sized euro country like Ireland, Hungary or Bulgaria. 25GW is electricity consumption of Poland, 30GW UK, 35GW Spain
this is not happening hardest because by the time theyād get permits for NPP theyāll get bankrupt because bubble will be over
Gonna build enough clean energy to power a medium sized industrialized nation just to waste it all on videos of babies dressed up as shrimp jesus.
CEO of cloudflare says heāll donate the bandwidth for Wordpress dot org to shut mullenweg up https://xcancel.com/eastdakota/status/1841154152006627663
Cloudflare is such a weird company in various ways. Saying loudly that they canāt judge groups when people ask them not to support the neo-nazis, harassers and worse (they have moved on this under pressure, but it takes a lot of pressure). But then they do this.
Wasnāt the first time he shut down 8chan (or was it Kiwifarms? Something along the lines), he immediately came out to say āItās really bad that I have the power to take down a website of shitheads.ā Just seemed like everything about his ideology is confused.
Nobody told him to make his CDN so dominant in the market. He kinda chose to have this power. If you think Cloudflareās services are some kind of a universal inalienable right, make them free, you pussy. Be a philantropist, share that sweet bandwidth with the poor. Gimme a /24, coward. Why are you taking money for a basic necessity, you monster?
Yāall can thank me for taking down 8ch and KF, because I also do not run hosting infrastructure for either of them. Same goes to all of you who do not run a CDN that welcomes nazi websites. Thank you for your service.
Thereās nothing confused about his ideology. Anything to do with porn or sex work will get kicked to the curb without a second thought, but fascists and bigots will be defended up to the point where they try to restart sectarian violence in Ireland (which was the final act of kiwi farms, from recollection. Presumably even the ceo couldnāt defend them at that point, though I bet he tried).
Dude doesnāt give a shit about free speech, or lament the power heās ended up with. He just really loves fascism.