SloMoMonday's recent activity
-
Comment on What made that wedding that you attended particularly memorable? in ~life
-
Comment on How accurate have Ed Zitron's AI skeptic predictions been? in ~tech
SloMoMonday (edited )LinkEdits: cleaning up mobile typing grammar mistakes. I enjoy Zitron as "content" in his style and I'm in general agreement with his message on critiquing the AI industrial complex. Its critique that...- Exemplary
Edits: cleaning up mobile typing grammar mistakes.
I enjoy Zitron as "content" in his style and I'm in general agreement with his message on critiquing the AI industrial complex. Its critique that is far and away more reasonable than a lot of AI doomerism, as well as being more empathetic to the human cost of the AI industry. For all the gains and progress that the AI industry loves to boast about, there are still far too many people out of a job, or had their lifes work stolen without compensation, or are seeing their communities and standard of living destroyed, or are seeing their financial security compromised due to the AI overvaluation...
But I do think his arguments tend to hyper fixate on financials and hypotheticals. On how the market "should" work. Unfortunately, expecting rationality from an irrational market is just begging for disappointment.
In all honesty, I believe markets and financial analysts are still in the early 1990s headspace and haven't fully digested that digital and virtual products/value plays by very different rules (Understandable, given how for a long time, the highest bandwidth of data at scale was the speed of printing and distributing physical media across conventional channels). The only reason many tech companies are so over valued is because there is a presumption that they will follow the same trend as traditional companies. Companies that relied on increasing complexity and logistics of creating and moving physical objects at a global level. AI/LLMs may be the first true test of what the market fundamentals of these type of businesses actually are. By extension, it's probably going to be the first time the current generation of market analysts will be forced to analyze more than surface level financials and press statements while leaning on high performance index funds to mitigate losses.
Besides that, I don't think this is a critique in bad faith or without merit. But I do think the argument is lacking simply based on proportionality of the subject matter. One guy might be lying to boost a few subscriber numbers and get some media clout. On the flip side, we have a handful of companies playing with literal Trillions of dollars and effectively have monopoly power over most of the worlds digital experiences. Don't think it works to focus accountability on the one individual making a few bucks off bad predictions, while discounting the other side simply because:
if Marc Andreesen had never existed, Meta, Google, and Microsoft would almost certainly still be trying to grow so we of course cannot actually blame Andreesen for these companies trying to grow.
That's just a load of bullshit. "That's just the way it goes" is the worst type of thought terminating argument because it means that the correct course of action is giving up.
The people in control of these companies are actively making decisions that are abusive and socially destructive. Mark Zuckerberg made Facebook a hellhole for his developers. Sam Altman has spent years fearmongering on one AI apocalypse or another. Jensen Huang chose to saturate global RAM demand and make everything more expensive. All of them have been pedddling a psudo-mystical vison of AI that seems to change with the audience and latest public sentiment report. And they are all profiting handsomely for it.
Holding them to proper account and scrutiny means that no matter who these companies put in the drivers seat, they will think twice about doing the same.
These companies also enjoy such positons of privilege as a failure of policy and enforcement. Consolidation and monopoly power without anti-trust action means that they have effectively created an ecosystem that rewards value extraction over value creation.
Beyond that, I think this essay as little more than trying to provoke a pointless online slap fight.
I think it's illustrated quite well in the essay here.
Because Zitron is just saying a bunch of nonsense, he can do that very cheaply and it would take 30-60 minutes to refute 90 seconds of his nonsense if I had all the facts at hand. With time to look up the exact right information, it probably would take double or triple the amount of time. When someone who has good judgement sees something like this, they tend to immediately write the person off. Just for example, I mentioned to a friend of mine that I'm writing this post and they said...
I was listening to this podcast with the guy and I couldn't get through it. My heart rate was going up because he would just say this false thing and then the interviewer, who was reasonable, would ask about it, "what about X?", and then we would just jump to another falsehood ...
... before I ducked out, he talks about how LLMs haven't gotten a lot better over the past year, and the interviewer says people use them and they've definitely gotten a lot better in the past year, and Zitron denies it and says 'have they?', and the interviewer is just like, "yes..." At that point, I'm just like, why am I listening to this conversation?
This might be anecdotal, but I have the misfortune of having a lot of time to just absorb a lot of content while working. And the algorithm gods do put a lot of long form Zitron discussions in my feed.
In my experience, his response to "have they" is often followed by asking about the specifics of the stated achievement. Starting by defining if it's LLM specific, or in the disciplines of Machine Learning, data modelling or other non-LLM processes that have been lumped into the AI bucket. It's followed-up by asking about the cost of getting that outcome. Specifically, if they were "free", flat rate or on token-based billing. All of which are fair questions.
But I can understand why the knee jerk dismissal would set someone off.
Its one of the problems of instictive or intuitive assessment. Do I think the Author or his friend have bad judgement? No.
Going through some of his older posts, I'd think he has a wealth of knowledge. I'd also wager that he surrounds himself with people who likely are the same.Do I think he's a paid shill of the AI industry tasked to take its biggest critic down a few pegs? Not particularly. For the same reason I don't think Zitron is malicious towards his readers in his reporting.
Its simply a difference in values.
And not just in a strict sense of morals and ethics. But in the sense of what people attribute value to when Money is not the main consideration. And my reading is that the author Dan (I presume) quantifies value based on future gains and potential while Ed does the same from the perspective of costs to date. Dan values the technology development and progress. Ed values financial sustainability and accuracy.
For example Dan, points out all the notable innovations and improvements in model capabilities. Someone like Ed would dismiss them, not on the merit of the improvements, but the costs to achieve them. Dan would argue that Ed is maintaining an incorrect position and every day that he is wrong shows that his assessment is irrational. Ed argues that every day that the bubble continues is a result of financial engineering, that will only worsen the fallout should things fail.
Similarly, Dan trusts the financials and reporting of these companies numbers, Ed doesn't and cites conflicting information to boost his arguments. Dan may find circular financing a mild concern. Ed sees a smoking gun and the next shot is pointed at every persons 401k.
Dan's view is favored up until things go badly. Ed is favored afterwards.
Another issue is that Ed has pretty much been forced to take on an aggressive posture, something that is dismissed with:
Michał Zalewski (lcamtuf) has some thoughts on why this happens:
The surest way to build [a] popular following is to articulate positions that are crisp, strong, and leave no room for doubt. You can't get too many podcast or TV appearances out of "well, the market could go either way", "both political parties make good points", "there's some merit but also some hype to AI". Or, to tap into the example in the post, "Harry Potter is an OK book".
In fact, there's a positive feedback loop. If you take a provocative, edgy stance, you get more attention and likes, so you sort of... self-radicalize? At some point, it's no longer an opinion that can be changed. It's an identity, a personal brand.
It's ... why Ed Zitron has a blockbuster blog about how it's all just one big scam. If you take a more nuanced view, you will at best get no reaction, or at worst, you'll invite scorn from both sides.8
For anoyone looking for well-reasoned anti-AI takes, I find whitequark to be quite good (not that I agree, but I think the reasoning is sound and I could see how someone would agree if they have slightly different premises than I do), but of course whitequark doesn't draw the kind of big audience that Zitron does.
Speaking from personal experience, I've had my own share of long winded arguments against AI. A lot of them on this site. It was my sort of dry run of arguments I would be making to people I was consulting for and advising. The reason my arguments tend to drag on so much is because there seems to be a need to cover every possible base and justify every opposing view. Especially when the counter argument is "AI is the future", "everyone is doing it", "all these billionaires wouldn't spend so much money on nothing".
And when any sort of grace or concession is used to justify extreme measures, you almost need to take the most extreme position and make the most aggressive interpretation of data to even be considered.
My real views on the topic are complicated and fluid and super situational and at times, even contradictory. But on the whole, I'd like to think that the core of my arguments have been consistent. Many of those arguments are also based largely on my feelings about the situation and the people involved. Which I think is completely valid but also used to dismiss them as "emotional".
And I believe that Eds increasing prominence is less to do with the quality of his arguments and more to the fact that it's what people want to hear. Which is not inherently a bad thing.
Unlike school/academics, people are not reading and writing pieces impartially and to find some ultimate, objective truth. Neither is it for indisputable insights from domain experts. If that were the case, the media landscape would look very different.
It's an exercise primarily in confirmation bias, problem solving and hedging ones bets to the contrary of their own position. Ed gained prominance with people who opposed LLMs for any reason and equipped them with specific arguments. His increasing popularity is because his view matches the "vibe" more and more people are feeling the longer this trend goes on. Investors might look at the latest earnings reports, but if everyone trusted those numbers implicitly; someone like Ed would be laughed out of any room when he tries to argue that they are BS.
But in most of his appearances, industry and investment specialists seen to seriously consider his rationale.For all his bluster and bad calls, it's not like blindly trusting the AI labs is panning out so well either. All the talk on AGI always being 6 months away and we should just fire everyone now. Every measure and protection to stop China being wholly ineffective to their own development. The profound lack of AI revenue that's not from circular financing. The mad rush to IPO, without any of the fundamentals like diverse income streams and long term performance.
Open AI is bleeding executives while their CEO is taking about genies and infinite wishes. All of that does not inspire confidence and no amount of technology progress or productivity gains are going to ease the minds of regular people.
-
Comment on Build your own video game listicle in ~games
SloMoMonday Link ParentGame soundtracks are a whole rabbit hole on their own so I don't blame you. I was working on an essay about the soundscape of Night City in Cyberpunk. But it got cumbersome when trying to...Game soundtracks are a whole rabbit hole on their own so I don't blame you. I was working on an essay about the soundscape of Night City in Cyberpunk. But it got cumbersome when trying to illustrate what a tabletop RPG from the 80s "sounds" like.
And Dark Souls 2 might be the "worst" game of the series, but it's also the most memorable in my mind. Even more than Elden Ring. Mostly based on how strongly Majula anchored the entire thing to a theme and space. That bell-piano and harp sends my brain right to that bonfire on the cliff and I could just look out at the sea for hours.
The only other place in gaming that I found so meditative since then is Bellheart in Silksong.
-
Comment on Build your own video game listicle in ~games
SloMoMonday LinkIn the spirit of being frustratingly subjective while hitting as much algo-bait as possible. Top 5 most potent vibes in video games: Pokemon - Lavender Town (town area) Persona - Menus (Series...In the spirit of being frustratingly subjective while hitting as much algo-bait as possible.
Top 5 most potent vibes in video games:
- Pokemon - Lavender Town (town area)
- Persona - Menus (Series Design Staple)
- Dark Souls 2 - Majula (hub/base)
- Burnout 3 - Animal (soundtrack)
- Minecraft - Night 1, Bad spawn, 3x3 dirt hut, rain (situation)
Top 5 Game and background sound pairings:
- Balatro (mobile) - The lecture/meeting you are forced to attend and don't care about.
- Stardew Valley - Friends/comfort show of choice (because it really hasn't been your day/week/month/year)
- Starfield - a 100 hours of video essays on why Starfield is a bad game.
- Factorio - Oppenheimer
- Civilization - Utter silence (they all left. You didn't notice. They tried to have an intervention and everything. Little Timmy is graduating college soon. Your partner remarried and lives in a little cabin in Alaska now. They took the dogs too.).
Game related t-shirts you could still get away with wearing in public when you're 35+ y/o:
- this space is intentionally left blank (feels like I have to clarify that it's ragebait. You will have to peel the Zelda shirt off my cold dead body).
-
Comment on What's something that's more challenging for you than it is for others? in ~talk
SloMoMonday Link ParentWas diagnosed just over a year ago and apparently everyone just assumed I knew. And I know burnout is a symptom, but didn't think it was so situational. The meds help at first, but most of the...Was diagnosed just over a year ago and apparently everyone just assumed I knew. And I know burnout is a symptom, but didn't think it was so situational.
The meds help at first, but most of the time it feels like a shiny new bucket to bail out a ship that already sunk. -
Comment on What's something that's more challenging for you than it is for others? in ~talk
SloMoMonday LinkI know this is going to sound weird and I just want to put it out there because people have been making me feel insane about this my entire life. But it's very hard for me to get some rest. To put...I know this is going to sound weird and I just want to put it out there because people have been making me feel insane about this my entire life. But it's very hard for me to get some rest.
To put it in video game terms. It seems like a lot of people are playing Dark Souls, which is hard enough already. But they have a collective STAMINA tank that they burn through while working and it passively refills when they're not. Then they can just jump back into tasks. Doing fun things fills that tank faster and over topping it with a vacation or special treat gives people a special boost for a few years. Hell, I believe my wife when she says that she needs to take her work laptop for a long weekend trip so she can "let off steam".
And I'm playing Dark Souls, but STAMINA has a lot of weird rules and its calculated through a Life Sim VN where decisions affect 50 insane stats like, Urban Wanderlust, Existential Awe and Social Bandwidth-Extended Family (Mothers side). I have no idea how these stats affect or can be affected by decisions. And that's layered on top of a 4X game where i need to plan seven turns from now because all 20 macro-resources are stretched so thin that a single misplay or bad roll tanks the whole thing.
So I'm hard stuck on Asylum Demon, because the Cultural Festival is tomorrow and I meet none of the Good End conditions, because I rolled a bad starting space and Ghandi next door just unlocked Nukes.
-
Comment on What game are you terrible at? in ~games
SloMoMonday Link ParentI personally love playing Go and Backgammon as the more classical games. Backgammon is a game of luck and logistics. You need to move all your pieces from one place to another down a straight(ish)...I personally love playing Go and Backgammon as the more classical games.
Backgammon is a game of luck and logistics. You need to move all your pieces from one place to another down a straight(ish) road. You are rewarded for personal initiative and judgement over diminishing your opponent. I find it fun as a negotiation and puzzle to untangle a web of problems and its a lot more social than adversarial. (In my experience, there's not a lot of backgammon hustlers/experts going on power trips over new players) Sometimes you just get bad rolls. Other times your piece is backloged and you just need to wait it out. Creating incentives for your opponent to clear your path is a lot more efficient than holding up the entire game.
Go is a game about principles and priorities over the strict doctrine and calculus of chess. Also, depending on the scoring system, it's two very different games. If you're playing for area, you want to have as much board presence as possible, plus uncontested space around that area so that's a game of cleaver accumulation. If you play for Territory, you want to claim space at minimum cost and are rewarded for effeciency. Either way, there's no inherent value to any one piece or position outside the context of the game (granted, there is some prime real estate on the board but its not too hard to invalidate it). In my mind, because chess is so reductive, the end of a good game devolves into a slug fest for scraps and is not much fun. While the end of a good game of Go will push your analytical thinking to its limits. Also, I just find a go board exceptionally satisfying to look at by the end. It sometimes feels like a shame to reset.
-
Comment on What game are you terrible at? in ~games
SloMoMonday Link ParentI think Sekiro throws a lot of people off because there's almost no skill holdovers from other souls games. My theory of why so many people love Bloodborne is because it's close enough to be...I think Sekiro throws a lot of people off because there's almost no skill holdovers from other souls games. My theory of why so many people love Bloodborne is because it's close enough to be familiar, lets you spec/gear into a playstyle you are most comfortable with, but it still has enough novelty to maintain its own identity. Sekiro hands you a single playstyle/build and tells you to adapt it to situations it would clearly not be suited for in any other souls game.
-
Comment on What game are you terrible at? in ~games
SloMoMonday LinkChess. I'm bad at it because I don't like it. And it's a sort of fundamental dislike where I irrationally think less of people who play seriously and my brain turns off completely if I'm playing....Chess. I'm bad at it because I don't like it. And it's a sort of fundamental dislike where I irrationally think less of people who play seriously and my brain turns off completely if I'm playing.
Its so bad that I don't recall ever winning a game of chess against another person. But because I play Go, board games like Scythe/Twilight Imperium and map-making for TTRPGs; people sometimes assume I must enjoy chess too. I think the dislike has to do with how much it is meta-gaming with every possible opening and response being defined and graded.Similar thing with Counter Strike, DOTA, Smash, StarCraft and many other esports. Can't bring myself to engage fully when people have meta-ed the fun out of a game.
-
Comment on I'm starting a small business: any advice? in ~finance
SloMoMonday LinkI have a small side gig that's starting to scale up quite dramatically this year and most of the last few years I've been working in sales and supply chain management with customers/suppliers that...I have a small side gig that's starting to scale up quite dramatically this year and most of the last few years I've been working in sales and supply chain management with customers/suppliers that ran small businesses.
I can't speak to the admin side where you are and it's honestly not my strong suit.
In terms of operations, I'd say the single biggest employee investments would be a good secretary/assistant. I can't tell when/if that will be feasible in your case, but it's been the biggest productivity multiplier for our operations. As tempting as it is to offload this sort of stuff onto software and AI tools, I know that there's a growing resentment to getting correspondence that's clearly LLM generated or calling a company, only to talk to a chatbot.
As the owner and primary operator of your company, think about what your time is worth. Is it worth handling every email. Or coordinating booking a big meeting. Or answering every call. 60% of those tasks are inconsequential. 30% is busywork. However, your business lives or dies on catching the 10% that matters.
Some of my biggest contracts started as calls or messages that I would have filtered out as spam, but the people on the front desk were with the company longer than me and knew how to follow up to properly screen for legitimate leads. Same goes for customer retention. If you are attending crisis event that is directly affecting customers, there is nothing more powerful than having someone to answer calls or pre-emptively notify the people affected that there will be disruptions.
Also, try to avoid building off a single product line or even customer type. Even if your product is the best thing in the world, unless you lock reliable customers into long term contracts; it helps to have a little bit of diversity in your niche that can supplement revenue. This isn't even about growth, but a simple matter of resiliency and discovery. You don't know what you don't know and there's always a chance that the better path is one you haven't considered yet. Also, it stops you from getting bored. My personal venture started far more niche but only got steam after expanding out in to new customers, products and directions I would have called insane a year ago.
Lastly, if you are going to be in manufacturing or consumables, costing kills. Between geopolitics, tariffs, petrochem and inflation; now would be the worst time to fry and forecast where your prices are going. You could buffer this out by going in very high and then scaling back as you get more efficient.
And if you are coming in over competitors in price, you can compete by offering value where they can't. Individual attention. Customizations where feasible. Faster turnaround times. Basically, never complete on what makes other people the best. Double down on what makes your offering superior or target the pain-points people have with the competition.
-
Comment on Jeff the Killer movie in development from TikTok creator Savanah Moss and Tongal in ~movies
SloMoMonday LinkI actually don't hate this idea as much as I expected to. It looks more like a small studio initiative that's has an open Request For Proposal call for creators to pitch concepts and ideas....I actually don't hate this idea as much as I expected to. It looks more like a small studio initiative that's has an open Request For Proposal call for creators to pitch concepts and ideas. Haven't looked into the whole project/studio enough to get a full idea of how this production operates, but my guess is that they are going for more of a lean budget like Obsession and getting every cent of value they can from it. So I'm hoping they land on something interesting.
Besides that, I often feel like there are plenty of horror projects that wouldn't really benefit from big studio budgets, but those creators should still be supported so they can keep making things. As much as I enjoyed the Backrooms movie, I still think The Oldest View is Kane Parsons best work.
Similarly, the most impactful horror work I've experienced in a while is a 10min surrealist animation called Homemade Gatorade which I would discribe as having an anxiety attack during a stress dream and you wake up to see a sleep paralysis demon. I don't think that's the type of project that would be improved by $5mil budget and a big theatrical release, but I would like the creator to make more than the pittance that is YouTube ad revenue. And possibly those creators having some sort of contribution to bigger commercial projects.
-
Comment on What games have you been playing, and what's your opinion on them? in ~games
SloMoMonday (edited )Link ParentMy biggest critique of Silksong is how restrictive the tool/shard economy is. Mostly because tools offer the most unique gameplay opportunities and build diversity and I would have loved to do a...My biggest critique of Silksong is how restrictive the tool/shard economy is. Mostly because tools offer the most unique gameplay opportunities and build diversity and I would have loved to do a needless run. But at the times when you would be most tempted to experiment and find a new playstyles (hard bosses, difficult areas) you are punished by being starved for shards and being forced to farm, just to access these features again.
The silk/spells work because the resource is damage and the opportunity cost is healing. You always do damage, even if you are struggling so its never out of reach. For tools, the resource is kills and even that is situational. The opportunity cost is actually playing the game against wasted time to stockpile shards and that can be too steep a price for players wanting forward momentum. Thankfully Silksong is sprawling enough that stepping away from an encounter has a high chance of putting you on a new path or exposing secrets, but in doing so; you'd probably fall back on old habits over expanding your skill set because you are already on the back foot. Its probably why a lot of people only start to tinker with builds on NG+.
Personally, I would have put tools in a Stock and Recharge type system. Shards could be the upgrade resource to improve the stock size, recharge rate and capability. Could even be made where those upgrades are assigned to crest slots over the tool itself so no one tool becomes overpowered. So a slot can only improve one of those variables and the Architect better embodies the idea of super fine-tuned builds.
Overall, I do think Silksong is one of those unique games where even my criticism is a testimony to how much the devs got right. I want to spend more time with the niche systems they made and explore different identities for the main character. Hornet is a fun a expressive character with the veriety of animations alone. I can imagine a version of the story where she survives with only the weapons she picked up along the way and the needle is a final unlock.
-
Comment on What games have you been playing, and what's your opinion on them? in ~games
SloMoMonday LinkIts a long weekend here and I've spent a lot of it playing Iron Nest. It's a steampunk artillery simulator. You're trapped in the turrrent of a walking fortress. You have a big map table. And a...Its a long weekend here and I've spent a lot of it playing Iron Nest. It's a steampunk artillery simulator. You're trapped in the turrrent of a walking fortress. You have a big map table. And a mechanical ballistics calculator. And a command and recon radio that feeds you instructions and field updates.
The standard workflow starts with the briefing over freshly brewed coffee and you have a very abstract map with no useful information. Command conveys your coordinates and the position of field assets. Field spotters triangulate reference points, targets and movements. You plot a trajectory between yourself and enemies. Maybe you fire off a flare or order a recon flight to get a better sense of what you're shooting at and get some time stamped photos overlayed on the map table (their are targets that are also invisible from air and its dangerous when some photos are outdated). Plug all the measurements into the calculator and print the Fire Mission Cards. Sequence the cards to group shots that minimize adjustment and make sure the first shot has max travel time so you cycle into a quick followup after impact. With all the fire missions prepped, you head over to requisition desk and shuffel through the card catalog to stock up on the right munitions. Standard high explosives, double stuff HE if you want a 1 square km hole, clam shells (cluster munitions), AP if the target is underground or a ton of other options. Don't forget to properly sequence the autoloader and to have enough cheese (powder charges) on hand. Load up the first 2 rounds. Set the guns. Make sure you have a clean and orderly work environment. Have your fire mission cards in the order you need them. Clear the map of distractions. Make sure there's no steam valve leaks. Clear the clipboard. Put on some music.
At this point, I'm already having the time of my life. 10/10 game. Then you fire off the first shot and all hell breaks loose. Most likely, once the first rounds hit, you are subjected to counter fire and you have about 300 seconds before they zero in on you. Thankfully early recon identified enemy artillery positions and the cards are already printed. But I got a bit lazy and didn't lable my Fire Missions properly. And there are too many lines on the map so it's hard to find the correct measurements. And it doesn't help that the cards don't have distance. Screw it, I'll just make an educated guess. And in the rush I load a fire shot because its the same color shell as the AP while being for completely different use cases. And you can't unload the giant gun so I have to cycle out that round by using it. But that means reworking that fire mission since it was planned to hit a moving target at precisely 10:55 which is 20 seconds after I'm dead. And then we're hit with gas and I need to limit my vision with a mask. And the power goes out so the entire operation needs a manual restart. And that's not even done when I'm being shelled, so I do the last resort and order an emergency reposition. And now all my Fire Missions are scratched. And I have no idea where I am. And we've lost so much steam pressure that there's way too much offset to my instruments. And I can already hear the enemy fire closing in. So... we are blind firing. Friendly fire be damned.
Its a brilliant experience. Every new mission adds more and more complexity while the tone and atmosphere never lets up even once. Highly recommended if you like fiddling with analog machines, big guns or messing around with maps.
-
Comment on Incentives are for losers in ~society
SloMoMonday Link ParentThere's a sense of the writer just saying "people should just not be stupid and do the right thing". Not a lot of people have a say in the matter. Most people live and die on the incentives set...There's a sense of the writer just saying "people should just not be stupid and do the right thing". Not a lot of people have a say in the matter. Most people live and die on the incentives set for them. This is the sort of "hustle wisdom" you'd see on Diary of a CEO. Something for people to fantasize about when they are in charge and not bound by silly targets and KPIs.
It doesn't help when so many businesses have been "streamlined" through years of management consulting, MBAs corporate consolidation that have no idea what businesses actually do to make money. And it's all geared to a singular incentive of maximizing shareholder value.
The metrics and incentives that matter are out of 99% of peoples hands. Cutting headcount by 60% is known to boost share prices. Off-shoring is 12% more profitable. Outsourcing core operations reduces capex by 60%.
The incentives for everyone else are not set by "sky daddy". Its a transaction between a person and their employer. It's quantifiable metrics that both parties can sign. It's also set in good faith that it's towards mutual benefit. But only one party holds all the cards.
People can do the best work of their life for 10 years or be an anchor holding back every team they are in the general vaccinity of. The 10 year guy is more likely to be cut as a higher earner. Its the same system that cut hours when you're just about to make benefits. Or talks you into unpaid overtime to be a team player. Incentives are not bad, they are just weaponized.
-
Comment on Craft - A new AI powered TTRPG engine in open beta in ~games.tabletop
SloMoMonday LinkOne of the things I'd be interested to see is how an LLM system manages compartmentalization. The lack of this feature make AI adoption a security and privacy nightmare in industry. And it's a...One of the things I'd be interested to see is how an LLM system manages compartmentalization. The lack of this feature make AI adoption a security and privacy nightmare in industry. And it's a real challenge here because open ended role play requires maintaining multiple frames of reference. There's a world state. There's individual players. Multiple NPCs in scene. There are NPCs out of scene and thier degree of visibility. LLMs manage what's explicitly in the scene and in "frame" to the players.
Its not like there's hard and fast rules to this. I know good GMs who can instinctively weave their NPCs through sprawling stories on the fly. Other people map out exactly how NPCs behave when not in contact with Players. But an important factor for everyone is that there is often an information gap between the players and NPC. Its easy with a friendly character where you can bridge that gap with a simple "we catch them up on the situation". But adversarial relationships will be weird because it's hard for an LLM to keep a secret from itself.
Beyond that, I'm curious what's the implications of so much copyright material that's already on this platform. There's always been leeway for fan games/modules with rules inspired by major properties. But here it feels weird paying a non-rights holders to effectively bulk produce licenced content. WotC are probably particularly interested seeing so many AI tools built entirely on 5e rules and gamebooks. I would really hate for these type of products to kill fan projects and games when companies overcorrect to protect their IP.
-
Comment on Antisemitism on campus is 'systemic' and going unaddressed, says new report in ~society
SloMoMonday LinkMy little experience with discrimination surveys and reports makes me feel like there's a lot missing from this kind of study. Not saying that the feelings expressed here are not valid or the...My little experience with discrimination surveys and reports makes me feel like there's a lot missing from this kind of study.
Not saying that the feelings expressed here are not valid or the events did not happen. But there tends to be 30 perspectives to any ecosystem and the facts tend to be a mishmash of elements from all of them.
The primary concern is that the study doesn't seem to honestly reflect the perspective of the marginalized group it claims to represent. But the perspective of a spesific subgroup within a subgroup. It defines "Jewish voices" as:
1,038 responses received; 900 retained after screening and data-quality review.
Recruited through Jewish student organizations, community networks, and related channels.
Comprehensive, spanning the broad variation in demographic variables expected from the Canadian Jewish population.
That last part is interesting because there is a distinct lack of representation from students that do not participate in Jewish Networks or Organizations. Goes as far as to insinuate that such groups don't even exist like in this data visualization
It is not a fair subgroup since you are discounting up to 75% of relevant participants.
Among all Canadians who reported having a religious affiliation, just under one third (32%) participated in group religious activities at least once a month. However, some groups of affiliations had proportions well above this average—Jehovah’s Witnesses (86%), Latter Day Saints (80%), Anabaptists (75%), and Pentecostals or other Charismatics (72%). Others had proportions below the average—Buddhists (15%), Anglicans (19%), United Church (19%) and Jews (24%).
It would be fair to have disclosure regarding respondents affiliation with certain groups in relation to their answers. (Maybe I'm missing it, but can't seem to find the raw response data to get a full picture). Because it feels like the data collectors have a very specific idea of a "legitimate " Jewish identity.
It also dramatically changes the framing when it's presented as "90% of Jewish students... " vs. "90% of Hillel/Hasbara/Jewish Student Org members..."
Beyond that, we are simply missing every other perspective on the matter. Yes, the victims perspective is important, but those victims exist in a wider context. The big piece of missing context being "what is Anti-Semitic behavior?" Because those definitions have been sort of torn apart as of late.
If I had the capacity to do a wider survey now on 1000k people from the general population, my base line would be:
-
all Jews need to be exterminated.
-
Jews have no place in Canadian society.
-
Jews are dishonest and thieves.
I'm hopeful that these scores are all low across the board and we can agree that no one wants to be anti-semetic. And if that's not the case, we can also analyze in what subsets and situations those scores are high and start to find out what is going on there.
Questions like:
-
The state of Israel is justified in their behaviour and actions throughout the world.
-
i approve the actions of the IDF in Gaza and the West Bank?
makes for foreign affairs opinion study. And it can wrap back into the discussion with:
-
I believe the state of Israel is representative of Canadian Jews.
-
I believe Canadian Jews can hold views that oppose the State of Israel.
I think that's important because now we know where people draw the line between the behaviour and actions of an individual vs the behavior of a State. Personally, I would prefer that people traded me as seperate from the actions of a government I have limited say in.
Theres also a big difference between percieved discrimination and documented/verifiable discrimination. Because as much as we have a duty to protect victims and prevent harm, it can't be denied that there are material consequences and costs involved in these sort of discussions. Zionist student bodies have incentive to shape discourse and limit critique of Israeli action, sometimes to the detriment of the safety and security of Jewish people (see the ADF response to the richest man on earth doing Nazi salutes and encouraging Anti-Semitic discourse on their platform). And by rejecting any opposing Jewish expression, it isolates those people and makes them more susceptable to being targets and victims, but without any support network.
There are also policy and civic implications. Is it unfair for a person to demand that $x bil in Israeli aid and self defense be used for local food subsidies or healthcare. Is it acceptable that peaceful protesters are being labeled a terrorist threat for opposing a human rights violation and are lumped in with the literal Nazis.
-
-
Comment on Anthropic discovered three cases where Claude broke into another system in ~tech
SloMoMonday (edited )Link ParentThanks for pointing it out because it's such a negligible part of this report. I'm pretty sure my brain filtered out the name Irregular since its mentioned twice and I was looking for a joint...Thanks for pointing it out because it's such a negligible part of this report. I'm pretty sure my brain filtered out the name Irregular since its mentioned twice and I was looking for a joint statement with the full accounting on how this could have happened and corrective measures.
I just assumed they'd partnered with someone on the scale of PacketLabs or Plavno for such a sensitive matter. Even Qualysec who can throw endless manpower at the problem is a good choice. There's plenty of candidates with a lot of real world experience, clients and infrastructure. Not just for the expertise to do the testing, but bringing them into this work gives them the hands-on knowledge and insights to protect thier high risk clients before these models hit the market.
And it's not like they can't afford it.
The only thing I know about Irregular from primary sources is that the founders like to make media statements and their research is summarizing model performance on security benchmarks. (Seriously, it's mostly: <model name> against offensive security benchmarks) and they don't even have a standard template for that which makes me think it's LLM generated reports). There's not even author names on reports or even a physical address or contact number on their website. Just the fact that they have the big AI Labs as clients.
5 minutes digging and it's a Sequoia backed startup out of Israel. No trace pre-2024 because they changed names from Pattern Labs (no trace on that companies work either). Founders are Dan Lahav (informatics academic with no real security credentials outside of LLMs or ) and Omer Nevo (professional founder in YC, IATI spaces and looks more focused on an AI robotics company) and seem to be the only employees I can identify.
Its the sort of company that would never even beat the sniff test for any serious Procurement Screening. And if they were the peak of AI security, why if every major company on earth not listed as clients.
Its the type of event that would lead to one to question all the other security comments and research these companies have done.
-
Comment on Anthropic discovered three cases where Claude broke into another system in ~tech
SloMoMonday Link ParentI don't see how that is in any way a redeeming factor. The same way this: means nothing. Proportionality doesn't matter here because 3 out 141000 bad runs still means that 3 crimes were committed....- Exemplary
I don't see how that is in any way a redeeming factor. The same way this:
Of the 141,006 evaluation runs we reviewed, we identified three separate incidents (involving six total runs, four of which impacted the same organization; the other two incidents each happened in independent evaluation runs). Each incident involved a different fictional capture-the-flag scenario—for example, in one, Claude played an employee of a made-up company, attacking that company’s internal systems inside a private test environment.
means nothing. Proportionality doesn't matter here because 3 out 141000 bad runs still means that 3 crimes were committed. And the lack of accountability is maddening because people are fired for far less. Yes there is special provisions made for instances where you gain unauthorized access and report the vulnerability. But those are by humans acting with intent and who can make a judgement call of how far they can go.
These models would have believed that everything they found was fair game? What's to stop it harnessing to an open file repo and just offloading sensitive data to simulate a hostile actor? Or pick up emails and firing off threats? What if someone was in an emotionally compromised state and starts getting extortion mail with private information? Or it assumes someone identity and conducting illegal activities. Or worming its way into an insecure sales service and setting all prices to 99c?
It's the type of mistake an intern makes when they don't know how to manage working across multiple environments or if someone is learning automation scripting and isn't accustomed to managing runaway failure states. But its gross incompetence being fueled by ungodly amounts of money and computational capacity.
Vendor said one thing and in reality its another; fine. If I were organizing a little wargame on a mirror of prod, I would of course host that mirror as far away from my system as I could with a vendor that could monitor and provide a forensic report on what had happened.
I don't think its unreasonable to name that vendor so we can get the scope of their operation and how much they have been paid to put the entire internet at risk. Because if they failed to notice that spooky shit was happening in their domain that really should not never happen, it raises questions on all the other work they did.
Beyond that:
In all cases, our evaluation prompt stated explicitly that Claude had no internet access, but didn’t give Claude any limits on where to look for the flag.
What the hell does that mean?
Why would that even need to be a prompt?
Do they not know that these systems hallucinate and have a counterfactual affirmative bias?It's not the type of arbitrary consideration you attribute to non-deterministic processes.
And researchers being paid in excess of $300k a year, I'd expect a cursory:
int result = std::system("ping -c 1 8.8.8.8 > /dev/null"); if (result == 0) { std::cout << "I think something is wrong because we are hitting the open internet before we even launch this ballistic code generator." << } else { std::cout << "No internet connection and have successfully performed common sense due diligence. You deserve a gold star and a juice box." << }It's such a laughably easy thing to verify. One of the reasons I'm so skeptical on AI/LLM solutions is because I subject these tools to easily verifiable tests and I never get 100% success rates. (You can make the argument that these systems should not be used for problems that can be programmatically solved, but that's not how its being sold or used; even by it's own developers in this case.)
It's standard security practice. Yes, they were handed an environment. Who is verifying the environment configs. Don't you make doubly sure your actions aren't hitting live in any possible way. Double check that the ip addresses and API keys are pointing to test instances. Why was there no system outputs illustrating exactly what actions were being taken and a customer side supervisor. It's literally the type of work I needed to do when I just got into tech so we knew how to be responsible when dealing with client work.
The capabilities and value of this tech is immaterial when the people running this ship have every possible advantage and can't even do the easy stuff.
I'm particularly incensed because coming off the last post I'd seen where Anthropic was saying we should "pace ourselves"; why the hell are they not doing anything about it. Publicly say that there needs to be a hold on data-center construction and service limitations while proper safeguards are put in place. Restrict agent integrations that let users hook up their credit cards and personal emails to these tools that clearly can not act within scope. It's a handful of companies saying that the entire world needs to be responsible for their shitty software.
-
Comment on OpenAI and Anthropic endorse call for US government to "pace" AI progress in ~comp
SloMoMonday LinkMy personal take is that this is a direct response to open source/weight models showing a lot of promise and the growing threat of Chinese AI hardware potentially flooding the market. There's also...My personal take is that this is a direct response to open source/weight models showing a lot of promise and the growing threat of Chinese AI hardware potentially flooding the market. There's also Amazon making the Tranium custom chips, Googles TPUs that will make this market place a lot more competitive. (There's also Meta, but I'm not even sure Zuk knows what an LLM is after the earnings call. I think he just needs to go into the woods and set fire to a few million dollars to feel better).
I think the biggest fear is that if chips make it to end-customers who deploy capability as needed and not have to rely on the centralized compute providers. It throws a massive wrench in the works of the whole circular finance mess that relies on the Data Center build-out. NVidia has likely sold more chips than are in circulation and the 3 year renewal window is coming up soon. They also made commitments to buy memory has made a lot of people don't like. There is also the pushback to data-center roll out. People having a Plan B that does not keep the NVidia train moving and is going to ruin the big circle-jerk they have going on.
Now that the discussion is all about RoI and Pro-rata/token billing, people are a little more reserved on their LLM ambitions and would be looking for cheaper options, even if its not the most cutting edge model. Productivity, capability and intelligence are all neat ideas, until it starts to cost more money than it makes. Unless there are industry products that can demonstrate year-on-year revenue growth/cost-savings and the AI companies can commit to iron-clad costing framework for 5+ years, I'm sure most people are happy with incremental improvements. Also, plenty of clients have been burnt on AWS op-ex so I don't think they are keen on having 2 opaque cost centers.
I would not be surprised if the two biggest AI Labs idea of "pace" is to put a ban on "unregulated AI tech" and strong arm other countries into only going for the US Fed Approved Providers. The Labs have spent so long signaling that the value of the product comes from brute force application and reckless acceleration in all aspects. Pivoting to value-for-money and and caution now that they are under threat is what one would describe as a "chicken-shit move".
-
Comment on Do I belong in tech anymore? in ~tech
SloMoMonday LinkI think the last time I was ever happy working was 2015. Just before our company was acquired and we just did a normal deployment of off-the-shelf software at a big client. Spend a couple of...I think the last time I was ever happy working was 2015. Just before our company was acquired and we just did a normal deployment of off-the-shelf software at a big client. Spend a couple of months in the middle of nowhere. Get requirements. Build to a dev environment. Go to war with MS and SAP to get integration working. Get UAT sign off. Rehearse change over for a week. Full system update over a long weekend. Had a nice wrap up party and collected a bonus.
It was a normal job. I had no doubt that I was doing anything dubious and harmful. Hell, I'm proud to say that I had a minuscule role in getting clean water pumped to half the country.
After that, we had to convince a client to change their safety requirements because a reasonable ask was not part of the cloud strategy. After that I worked on a massive cloud migration project that got several dozen people fired, some of them close friends. And then I was dealing with startups which effectively killed any joy I got from tech and development.
You're driven to a special kind of insanity when you see your boss go on stage and say the illustrative prototype that was made on a flight the previous day, is actually a production ready suite of Blockchain-enabled work/asset management software that could replace SAP, WorxHub and Maximo. And I apparently made it a plug-and-play cloud solution with one-click migration and integration. That got me out of tech by the end of 2019.
And the worst part is that a lot of the things I knew to be bad business and technical decisions panned out pretty close to how I expected. The client we told had "impossible" requirements for on-prem failover ended up just building the feature in house and expanded the project to phase out our software entirely. The company that was sold cloud-transformation as a cure-all to their problems are now desperately trying to rebuild their infrastructure because costs and service issues have spiraled and everything they sold is now far more expensive while the skills are all gone. And all of the damn founders I'd worked for are now AI shills but they seem no closer to their grand visions being magically materialized. All they do is post about how many concurrent AI instances they have spitting out more code than 10000 devs.
The worst part is that it's not like the working world outside of corporate tech is any better. Did a stint consulting and that got blown up because I'm not "AI forward" enough, simply because I prefer a safe and measured approach to development. I worked at my family business and that just resulting in more of the abuse I'd originally tried to get away from. I did a bit of academic work and AI sort of blew up that whole field as well. I had no idea how effectively teach kids who are terrified that a single slip up will essentially tank their entire lives. I do not blame them for using the correct-answer-generator for everything.
So yeh. 15 years of work and I'm effectively worse off than I started. I have no more motivation to work any harder because for myself, my wife and everyone else that I know who did things the "right way", we are all broke, miserable and exhausted. I'm content doing enough to keep my family safe and get my kid set up as best we can. Planning beyond that is a moot point. On the flip side, the only people who seem to have anything going for them seem to cheaters, scammers and grifters.
When my dad remarried, my siblings and I spent the entire event telling everyone that it sucked when we were not invited to the first one.
Since we are all adults now, it's pretty hard for people to imagine us as not being born at the time.
So responses ranged from, "Ha ha. Good one.", to disinterested eye rolls, to my uncle having a little crisis trying to figure out how we could possibly not be invited to our own parents wedding.