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)
via mastodon
image description
a screenshot of a bluesky post from Tim Dawson:
lot of negativity towards Al lately, but consider :
are these tools ethical or environmentally sustainable? No.
but do they enable great things that people want? Also no.
but are they being made by well meaning people for good reasons? Once again, no.
maybe youāre not being negative enough
I was reading David and Amyās stuff on energy costs of AI and ended up skimming the MSFT āenvironment sustainability reportā andā¦ god
Like I donāt even know how to satirise this. Help me sneer pros, for I am too depressed to make fun of this.
Additionally, we are exploring how technology, like savory vapes and cocaine, can help me kick my meth habit.
Google also said something similar in one of their reports. Something along the lines of sure AI wrecked their sustainability report this year, but just you wait until it optimizes the data centers! As if the robots could find holes in thermodynamics or something.
Anyway itās not that great but hereās my attempt at the sneer you asked for:
āAdditionally, we are exploring how attaching flame-throwers to the bottom of private jets and flying over the tree-tops of forests can further increase the accountability and traceability for our Scope 3 carbon emissions.ā
Appreciate, but flamethrowers on jets still sounds somehow less idiotic than tracking CO2 emissions with BLOCKCHAIN
Like a century of science: yeah weāre pretty sure where carbon emissions come from. Everyone needs to slow the fuck down. Thereās no need to pontificate about the specifics, especially if that somehow produces even more emissions. That would be catastrophic, you see.
MSFT: hold my beer
When [muskās new] supercomputer gets to full capacity, the local utility says itās going to need a million gallons of water per day and 150 megawatts of electricity ā enough to power 100,000 homes per year.
The locals might get ornery about this, and Iād be willing to make the trip to help.
There are some steam turbine power plants (like coal-fired) (on smaller side) with power output like that
Continuing on from this nugget that Lex Fucking Fridman will be āanalyzingā the Roman Empire, some nutter in the xhitter thread hoped the real reason the Empire fell would be āinflationā
https://awful.systems/comment/4649129
Looking forward to some chuds referencing the coming 1,000 hour podcast as proof the Roman Empire fell because woke
Itās remarkable to me how far and how rapidly this guy swerved outside of his initial lane, all while having absolutely terrible voice and diction for being a long-form interviewer. Heās worked on that, but itās clear that his initial success was based off of targeting high-level professionals who otherwise wouldnāt very often be sought out for the type of interviews Lex does. Iām thinking of guys like Jim Keller and Chris Lattner, who would probably only make such public appearances in the form of keynotes at conferences for their specific niches.
But you canāt convince me that youāre really the worldās best technical interviewer if youāre also uncritically sitting down with Donald fucking Trump, or deciding that youāre suddenly enough of a historian to take on Gibbon with your fucking podcast. Whoās financing this guy, anyway? Is MIT actually kicking him cash, or is it just an RMS scenario where they give him space because theyāre concerned about where he might end up otherwise?
The only thing Iāve seen from Lex Friedman was his interview of Brian Kernighan. For most of it I just thought it was very kind of BWK to patiently indulge this kid, who was clearly still new and unaccustomed to public speaking or researching his interview subjects, despite the weirdly professional gear setup and production.
Note that he uses the same strategy as Joe Rogan: invite a smart person on, ask them introductory questions about their research, and then just kind of sit there with a dumb look and fail to understand what theyāre saying. I gather that itās easy to empathize with and doesnāt require listeners to actually learn much since theyāre essentially sitting in a 101 course with a professor who is reading the curriculum aloud. What puzzles me is why MIT funds this shit.
I donāt think itās very surprising. The various CS departments are extremely happy to ride the wave of easy funding and spend a lot of time boosting AI, just like how a few years ago all the cryptographers were getting into blockchains. For instance they added an entire new āAIā major, while eliminating the electrical engineering major on the grounds that ācomputationā is more important than electrical engineering.
Lex hasnāt optimized the skill of technical interviewing; he has optimized the skill of simultaneously stroking the intervieweeās and the (implicitly) listenerās ego.
some nutter in the xhitter thread hoped the real reason the Empire fell would be āinflationā
Someoneās been watching too much Tuttle Twins (Warning: link to fascist propaganda youtube channel).
The links between hard money/goldbugs and (US) hard right goes back a long way, at least to the 30s I believe.
Interestingly, an almost pathological fear of inflation is also part of the foundational myth of the BRD, but if you look at the actual history, Weimar-era hyperinflation wasnāt really the root cause of Nazism, the Depression arguably was a bigger contributing factor.
It makes a certain amount of sense with the conspiracy theories that are at the heart of fascist understanding of politics, though. Goldbuggery treats inflation like itās a very simple question of monetary policy rather than a complex emergent part of an economic environment centered around constant growth. This means itās a perfect tool for (((Them))) to be using from their secret position of power to invert the obvious natural order and keep Us (and more importantly from a propaganda perspective, You) away from the luxury and power that We deserve. The fascist conspiracy theories also answer the obvious problem with the goldbug narrative: if itās so easy to fix inflation and would have no negative consequences, why donāt the people we keep electing to fix it justā¦ do that?
Well put. Another example I like to play in my head (never debated a goldbug for real in my life, not starting now) is that if the gold standard is so great, how come a small-ish country like Switzerland or Singapore hasnāt started using it and outcompeting everyone?
Thereās only 2 answers to that:
- the gold standard doesnāt work in the modern economy, the one that has lifted millions out of poverty and created untold wealth (to great environmental damage, sure)
- the gold standard is being kept from them by (((they)))
Answer 2 is obvious if youāre a fascist.
@gerikson @YourNetworkIsHaunted That and even hyper-cautious countries like Switzerland have been selling off their gold reserves to at least some extent, because they listen to sane economists rather than nut jobs.
Not gold, but some countries do work to an officially restricted money supply! Those that have officially dollarised, e.g. El Salvador and Ecuador.
Iām familiar with .sv. The government is horribly constricted - because they canāt print money and the populace doesnāt trust them to print money - so every year itās more sovereign bonds. Then a fuckwit like Bukele comes along and thinks that bitcoins will make anything better and not worse.
So yeah, turns out past 1930 that not being able to do monetary policy fucking sucks.
Apparently it implements chain-of-thought, which either means they changed the RHFL dataset to force it to explain its āreasoningā when answering or to do self questioning loops, or that it reprompts itsefl multiple times behind the scenes according to some heuristic until it synthesize a best result, itās not really clear.
Canāt wait to waste five pools of drinkable water to be told to use C# features that donāt exist, but at least it got like 25.2452323760909304593095% better at solving math olympiads as long as you allow it a few tens of tries for each question.
Some of my favorite reactions to this paradigm shift in machine intelligence we are witnessing:
bless you Melanie.
Mine olde friend, the log scale, still as beautiful the day I met you
Weird, the AI that has read every chess book in existence and been trained on more synthetic games than any one human has seen in a lifetime still doesnāt understand the rules of chess
^(just an interesting data point from Ernie, + he upvotes pictures of my dogs on FB so I gotta include him)
Dog tax
Would there ever be a way to tell that they didnāt just feed the answers into the training data?
One to keep an eye onā¦ you might all know this already, but apparently Mozilla has an āadd ai chatbot to sidebarā in Firefox labs (https://blog.nightly.mozilla.org/2024/06/24/experimenting-with-ai-services-in-nightly/ and available in at least v130). You can currently choose from a selection of public llm providers, similar to the search provider choice.
Clearly, Mozilla has its share of AI boosters, given that they forced āai helpā onto MDN against a significant amount of protest (see https://github.com/mdn/yari/issues/9230 from last July for example) so I expect this stuff to proceed apace.
This is fine, because Mozilla clearly has time and money to spare with nothing else useful they could be doing, alternative browsers are readily available and there has never been any anti-ai backlash to adding this sort of stuff to any other project.
I have thought about this guy every day since I saw this post
I immediatelly knew who and what you were talking about without even clicking.
May the fact that he also lives inside my head rent-free be some solace to you.
If only all my snark could elicit such absurd perfection.
Every day? I dont think I have read that post at all. (This is both a joke and not a joke, as I had not read it, I did now and I was amused, so thanks).
I told one of my college professors Iād been having issues with some software I had to learn to use for another class, and he said ācan I give you a tip? try using chat-gpt to explain how to use itā and without thinking I said āwhy would I use chat-gpt? Itās rubbishā and his face fell. Sorry, Prof, I know you were trying to help.
This was after heād said to the class that he knew we would all be using chat-gpt for assignments.
new idea: get the morewrongers to work themselves up about āontologically, is āsuperhuman predictionā the same class as superintelligence?ā
why? oh, yāknow, just things:
Iāve clowned on Dan before for personal reasons, but my god, this is the dumbest post so far. If you had a superhuman forecasting model, you wouldnāt just hand it out like a fucking snake oil salesman. Youād prove you had superhuman forecasting by repeatably beating every other hedge fund in the world betting on stock options. The fact that Dan is not a trillionaire is proof in itself that this is hogwash. Iām fucking embarrassed for him and frankly seething at what a shitty, slimily little grifter he is. And he gets to write legislation? You, you have to stop him!
You have to let it predict things that will happen before 2019, duh.
BRB training an LLM to be a super-goalpost-mover
Percentages are cheating, especially percentages below 50.
Iāll predict a 49% chance Mont Blanc erupts tomorrow, covering half of Europe in chocolate.
As zero cannot exist as a chance, there is a chance this will happen.
Thatās why I vote 3rd party in presidential elections.
how to spot the US angst:
I got mad at this last time I saw Nate Bronze assign percentages to the 2024 US election.
Like, my dude, what the fuck does that mean? Is the election result a random variable? What is its PDF? What maths could you have possibly done to arrive at a crisp [0, 1] probability value?
How did you go from āpredict the futureā, an obviously wildly fuzzy and inaccurate vibes check, to a concrete real number??
obvies theyāre just 30% better at predicting the future than you
Iāll cut you with a Y axis I swear on me mum
azimu(s)th
āThis tool can do X at a superhuman levelā is often quite an embarrassing thing to believe. (in before somebody says computers can do calculations at superhuman levels). Saying your tool can do that is also pretty cringe.
Ok this might be a bit petty of me but, yes this HN comment right here officer.
A group pwns an entire TLD with a fair amount of creativity, and this person is like (paraphrasing) āif you think thatās bad news just wait until you hear AIs can find trivial XSS and SQL injections š±ā.
Aside: have I ever mentioned here that you should really stick with .com / .net / .org / certain country domains? Because this sort of stuff is exactly why. Awful.systems can get a pass since the domain name is just that good.
quoted because this is fucking gold and paraphrasing isnāt doing it:
Do you have any references/examples of this?
tons
rapid7 for example use LLMs to analyze code and identify vulnerabilities such as SQL injection, XSS, and buffer overflows.
Can you point me to a blog or feature of them that does this? I used to work at R7 up until last year and there was none of this functionality in their products at the time and nothing on the roadmap related to this.
mustāve been another company then which i got confused with the name
Good thing you have tons of examples.
Right?
e: youāll never guess what a bunch of DEI Steveās other posts are about
I wonder if he got standard ML and LLM confused. (I did hear there was some usage of LLM/ML to help with some documentation stuff I think on a riskybusiness podcast, but I would have to relisten for the details. It could also just have been promotional stuff, while they are not actually using it).
Poor DeiSteve, it always sucks when you have a decades old username which suddenly takes up political meaning.
ācreated: 51 days agoā
Oh no.
Standard ML the programming language or standard as in conventional and ML as in machine learning?
Machine learning.
E: Turns out the DEI guy might have been right btw. https://www.rapid7.com/about/press-releases/rapid7s-ai-engine-supercharges-security-operations-with-generative-ai/ if by generative AI they mean a LLM.
And from their blog: https://www.rapid7.com/blog/post/2023/11/29/rapid7-takes-next-step-in-ai-innovation-with-new-ai-powered-threat-detections/ so there could be a chance the āI used to work thereā person didnāt know everything going on in Rapid7. āRapid7 Has Been an AI Innovator for Decadesā
Awful.systems can get a pass since the domain name is just that good.
a new source of anxiety has formed
in all seriousness, a backup domain name might not be the worst idea one day. I donāt think Lemmyās federation particularly likes being ripped out of one FQDN and migrated to another, but itās probably preferable to shutting down cause the owners of our TLD thoroughly shit the bed
awfulās probably okay, .systems is run by Donuts and theyāre one of the bigger operations around
pro-tip: do not learn things about how TLDs work (and I mean the bit beyond dns architecture), it is cursed knowledge you canāt unlearn
and with that warning delivered, yāall may freely run to hyperfocus on this, and realize too late itās a gateway drug
regarding backup domain: yeah always handy to have something, but nfi how to port it. APās identity design there really leaves something to be desired :/
(e: good lord I was out of it when I wrote this post)
the original post is fucking amazing
here in the future, nothing fucking works
.mobi? They became the admins of the file format? And they paid for it? Good luck with that. ;)
E: me after reading the article. Ow god nothing fucking works indeed, that is dire. I actually checked the date to see if this wasnāt some old post. Nope 9/11 2024. Buffer overflow + lapsed domains.
Fuck it, weāre going back to bang paths. ficix!hetzner!awful!self please add support for this.
you have no idea how much Iāve been tempted to do UUCP
I liked this comment on the HN post:
Our computer security analogies are modeled around securing a home from burglars, but the actual threat model is the ocean surging 30 feet onto our beachfront community. The ocean will find the holes, no matter how small. We are not prepared for this.
Phase 3: Reality Distortion (2041-2050)
2041-2043: Financial Multiverse Modeling
- Quantum computers simulate multiple financial realities simultaneously
- Development of āSchrƶdingerās Ledgerā prototype, allowing superposition of financial states
https://www.linkedin.com/pulse/roadmap-quantum-accounting-milestones-evolution-garrett-irmsc
2026-2027: Blockchain Revolution
- Widespread adoption of blockchain for secure, transparent financial transactions
- Development of industry-specific blockchain solutions
- Smart contracts automate complex financial agreements
ah good, weāre still on schedule for that I guess
I have to admit that I wasnāt expecting LinkedIn to become a wretched hive of āquantumā bullshit, but hey, here we are.
Tangentially: Schrƶdinger is a one-man argument for not naming ideas after people.
Somebody played cookie clicker.
you left the best one:
First successful experiments in ātemporal arbitrageā using quantum prediction models
not only they know about ātemporal arbitrage experimentsā but they also already know that these were successful. that kinda defeats purpose of experiment
Frankly, it just didnāt stand out. The entire post is such a hoot. Someone must be microdosing meth again.
found a quality sneer in the wild https://mander.xyz/post/17910623
From the reactions:
āWith enough garbage the model will become sentientā
āI mean thats how humans are raised thoā
AAAAAAAAAAAA
(There is a tendency among promtfondlers to, in their attempt to hype up their objects of affection, diminish humans and humanity).
As my little 2 year old said, after listening to a white noise generator for days. āHoly shit, I think therefore I am!ā
diminish humans and humanity
AKA tell on themselves
Turns out purging your ranks of wokes and furries is more fun than, you know, actually developing a working Linux distro: Nix 2.24+ is vulnerable to (remote) privilege escalation.
(linking to the lobste.rs discussion because I feel it does a decent job curating related links around disclosure timelines etc.)
howās the nix drama going these days? I need more spilled tea to sip, anywhere I can read a recap? did everyone just gave up on not being sponsored by border surveillance drones?
we have some dedicated Nix threads on FreeAsm, our community for open source stuff that might give you a good recap of the latest stuff. where Iām sitting as a former Nix contributor and advocate is ā nothing is fixed, all this obvious fash shit worked so well itās being tried in other open source communities, and I really would like a nice alternative to NixOS so I can finally release a lot of the Nix stuff Iāve been sitting on (and also because nixpkgs keeps breaking ā so many maintainers left that some of the further corners of the package set I use are already showing a lot of rot)
Lix fixed it last month of course
this isnāt surprising, but now itās confirmed: in addition to the environmental damage generative AI does by operating, and in spite of all attempts to greenwash it and present it as somehow a solution to climate change, of course Microsoftās been pushing very hard for the oil and gas industry to use generative AI to maximize resource exploitation and production (via Timnit Gebru)
tbh i donāt see a single sane way that genai could be used for anything like they say it can be, if it works itās gotta be something more or less custom. but ms doesnāt care, because theyāre selling shovels so it doesnāt matter if their shit doesnāt work as long as someoneās buying. it sorta starts looking like cryptobros in 2020-ish trying to insert themselves as middlemen everywhere where thereās already some money
My heuristic is basically: until we can easily control the environment of earth, terraforming will not be on the horizon. Given that we are hell-bent on driving off of as many cliffs as we can, Iām pretty happy to continue thinking that weāll never get off this rock proper within the olā lifespan
this is a fun read
another I ran into earlier (via an outfit from these parts) is this piece from The Conversation. I wish I could get as irked about linear extrapolation as I usually do, but in lieu of guardrails arresting (and clawing back!) all that shit from all these ghoulsā¦
You could build a whole library of articles like this: hereās one from early last year
(Edit: I based parts of this post on misremembered/misinterpreted tweet from musk, so reader beware, Iām being wrong about Musks stated plans here, see my later reply in the thread. Not sure if I should leave the post up, edit it (eurgh effort), or just delete)
Made a decision for the donāt waste time common good.
@Soyweiser Youāre assuming the first Starship to Mars has a cargo of canned primates. Rest assured, it canāt and it wonāt. (Also, I doubt Starship will be ready for Marsāextended duration in space, rememberāin time for the 2026 launch window. Although a robot probe as a payload launched atop Starship is entirely possible in that time frame: you donāt need reusability, just a bloody big payload bay and the ability to reach orbit once, which Starship achieved as of OFT-3.)
deleted by creator
