Today @poolsideai is releasing Laguna M.1 & Laguna XS.2, our latest generation models and first public models We started Poolside because we believed that to build truly capable coding agents, you need to own the full stack: data, training, reinforcement learning, inference. These models are the first result of that work, and we’re making them available to everyone

Likely most attention Quora has gotten in probably 7 years

I hold a few very un-popular but likely hard truth views on the world and life One of them is that most people will never achieve a single large goal they set out to do and it’s primarily because we as a society have taught them a very dangerous concept called ‘moderation’

I want to say something funny, but there is no chance I find something funnier than the idiocy of not shoring this up before you run the play. This is like corporate subterfuge 101, which a professional board would have known

Taxing illiquid unrealized gains is such an insane destructive idea you have to apply the old “don’t attribute to malice which can be equally attributed to incompetence” but that only leads one to then conclude the decision makers are so incapable of even bottom tier critical thinking everything is royally and truly fucked It’s the same as someone saying “why don’t you just fly to the sun at night”. You laugh along until you realize they’re serious Right now the world is going in two very different directions. So many (mostly in Europe, but I put Canada in this bucket too fwiw … everyone in leadership here needs to go and likely will get wiped out next election) are giving themselves over to degrowth (aka slow painful death) while a few are charging towards hyper growth (aka life) The problem is the degrowth folks are convinced they can fly to the sun at night and have dunning kruger confidence bc of their made up nyu degree in intersectional economics and it’s impact on indonesian palm oil farmers they feel qualified to make nuclear policy legislation

Chris. No. Please. I honestly can’t tell if this is true or not. This now became work for me to find out. Please end all future tweets with /s for satire or /r for ‘no seriously this is real’

This is, and I cannot overemphasize this, unhinged This is like listening to a college freshman talk about Stalin or Nicolae Ceausescu as ideals to aspire to. It’s just thoughtless chatter from the uninformed and dramatically, best case, naive or, perhaps more realistically, dim

This is a *perfect* example of something that happens in tech all the time…it’s not just Google but most visibly happening at Google and we are seeing it play out realtime. Google has amazing tech, it’s not the tech obviously Clear lack of vision and decisive leadership This results in no decision making, no direction setting, and pure maintenance mode glide path “work”. Entropy at its finest Easiest way to get a company who is experiencing this back on top? A small set of singularly area accountable people owning outcomes who say “no” to most things and “yes” only to a small handful of highly important things Most people can neither handle this pure leadership burden nor do the org change management to make this happen latent.space/p/adept

Let’s talk about one of the absolute dumbest things in tech, and I’m reminded of this bc of the NFL draft coming up and how every year nfl teams do the same dumb thing people do in tech: over analyze things that don’t matter

Imagine doing this with planes or trains or X-rays or penicillin or the lightbulb or indoor plumbing or the transistor or the microprocessor or … Progress is never inevitable. Many will fight it for various reasons. It doesn’t matter so long as just enough ppl have the will to keep going Humanity forgets these incidents with time and builds on the successes as new foundations. It’s how we have won and will continue to win

Wife gets a call from local pizza place Pizza: “did you order pizza? We have order & your number but no address” W: “probably my son who thinks he’s sneaking pizza while I’m out. Go ahead and deliver it & I’ll talk to him when I get home” P: “that makes sense. We got a note on the order that said ‘your website is terrible. Please consider making it better so it’s easier to order pizza’” Anyway, he’s available to review things for folks. Works for pizza

"It's impossible" or "can't be done" or "there's no way" are the epitome of mid thinking and if you ever find yourself or others doing this, just literally stop. It's a trap. Instead, flip it around like this: "If it could be done, how would it be done?" Always a way. Might require time, people, money, new things to be build but there is always a way...and once you find that way, that becomes the rough plan and the path. Just work the problems This has been amazing to use over the years with engineers. Most "impossible" is often just "I'm not sure how" or "it's very hard" or "there is lots of risk involved" in disguise But where I see this thinking most often is general twitter business stuff. Mid thinking infects twitter discourse on nearly all business related topics because, well, commentary is easier than doing the thing A sheer rock face is "impossible" to climb to the novice or non-climber. But a rock climber with any reasonable experience will see it differently and probably see several ways in which it'll work out, and some even that could be done free climbing. And if they had it slightly wrong and halfway up the rock face something is different, they don't panic, they reassess and figure out a new plan from that point, even if they have to backtrack slightly or scrap the original plan entirely and redo it. That's the point of experience and mindset. They find a way Good example: "What if Google builds this?" is the very definition of this thinking The answer to that question is: there is always a path through....for those good enough to see the pathway. There is always a way. Always. Or, again only for the non-mid thinking in the group, the answer *really* is "it doesn't fucking matter because we'll figure out a way as the obstacles arise" There's a natural reaction to want to have all the answers but the truth of life is that just isn't the way. The REAL trick is not having the answers, but being skilled enough to navigate the various landscapes and challenges as they present themselves. And more, the further out you can see the possible path the more you can anticipate and prepare for the possible obstacles, the better you are going to be. "Impossible is nothing" is akin to "I see a pathway to this and am prepared to do what is necassary. It's ok you don't see the pathway and that is why you won't achieve this" Never let someone else's mid thinking change your mindset. If you want to be great and do great things, shake off all notions of impossible or elephant and the rope constraints. There are no lines in the real world, there are no rules besides physics and even our notion of physics needs updating as we learn more

The simplest & most effective question I have ever found in years of selling large complex sales, putting together massive deals (m&a or otherwise), or putting partnerships together is just “how does the person across the table get recognized or promoted?” Once I know this know I know most things that are valuable in putting it all together And remember, you’ll meet sometimes dozens of people in these big deals. Ask the question about everyone you meet Incentives rule everything around us

The next decade is defined by 3 things and basically only 3 things for anything bits (atoms are different beast entirely and deserve their own thread) 1. Energy to power the 2. Chips for training/running the new emergent 3. AI intelligences/networks Most other things are rounding errors in the fullness of time. If you see this, your day to day behaviors are very different. And your investments of time, money, and people also look very very different

Human-level intelligence is going to happen, and we think it’ll happen first in software development. Today, we’re staking @poolsideai's position in the field. AGI’s promise is too significant for me and @eisokant to stay out of the mix

Satya was already all time great but he might have just gone goat And wow the 72 hour value destruction, cannot fathom The most obvious thing here is every AI startup in the world just got a gift but the other thing is every hyperscaler is having really tough conversation with their board this morning There is Satya/MSFT and there is a gap so wide we can’t see the other side and then there is every other megacorp leader at the moment See you at Reinvent lol

This is the most depressing and destructive graph I’ve ever seen for a first world country. I love Canada, but Canada won’t even be a thing if this continues Canada is the IBM of countries

Wife: “sometimes in life someone doesn’t have to be as smart because they’re very good looking. It’s just how it works sometimes but don’t…” 9yo: “omg, daddy must be a genius!” Wife: “no, it doesn’t work like tha….” 9yo: “because he’s hideous” Have kids folks, trust me.

Preview of what's coming in early '24 Dive in and sign up for early cabana access at poolside.ai

Have a long flight so time to do a thread on a question I get in some form at least half dozen times a week Where will value accrue in AI from here Think: foundation models, apps, middleware, full-stack blah blah blah etc

Got a DM asking why I’m picking on this one It’s rather simple. It’s in the category of so obviously bad that the mere suggestion of putting it forward or implementing it shakes ALL confidence in the person/group suggesting it So if someone is a politician or party and they think this is a good idea, one must do nothing other but conclude that the person/party is also as clueless or destructive in all other areas. It’s basically opposite of gell Mann amnesia. It’s so beyond the pale obviously bad that to even suggest it we have to write off EVERYTHING the person/party says. The idiot bit has to be flipped and it’s a permanent flip in the cases like this Imagine it’s 4th quarter, there’s one second on the game clock, you need a touchdown to win, you are at the 20 yard line, and your offensive play caller suggests you take a knee to surprise the other team. Do you think that person has a job on Monday? No. The idiot bit instantly gets flipped and very rightfully so. Heck, I’d call timeout and fire that offensive player on the spot no questions asked

Everyone seeing the names and then Marissa Mayer

16yo: “dad, can you help me fix my computer? My files won’t sync” Hands me laptop Me: “buddy, everything is in Latin” 16yo: “yeah I changed it so I could learn Latin easier” Me: “I can’t do anything with this. Let’s change it back” 16yo: “I can’t because it’s all in Latin”

Wife: “why did we just get $500 worth of big league chew delivered to the house? I swear sometimes it feels like you’re what would happen if a child got an unlimited budget” Me: “at this point we’ve been married for 21.5 years so if you think about it me being this way is kinda your fault too” Wife: (someone is typing bubble) Wish me luck!

Boy math is because you ate one cookie you might as well eat all the cookies and finish up with 3000 additional calories before you go to sleep since it’s still just ‘one day’

Me at my first Bridgewater board meeting

Hardest types of turnarounds are cultural, which is why you obsess over culture daily & never let it slip, even a little, ever It all starts with people & expectations Erosion recipe is clear: sloppy hiring, hands off leaders, no expectations, not firing, & no tough convos

I absolutely do not believe that if your goal is to do something even approaching very good let alone great, moderation is the way Moderation is for mediocre. Perhaps more accurately , moderation is for maintaining If your goal is to maintain your status quo, moderation

They will write entire books, that will become training data for sama's new thing, about these few days. Stanford profs are gonna teach corp gov classes rooted in what is going on here.

Between prior and this statement, this whole thing will go down as all time fumble and top 10 comms disaster in corporate history

This might be the ai code equivalent of the original rails blog post video that set off a dev tools golden period For the record, where we go from here with AI is gonna be wiiiiiiiiild

Thanks Google. Makes sense to me

Soooooooooooo My wife found my twitter googling me for something else and now I’m explaining old tweets to her over text So yeah, that’s new and interesting and “fun”

Everyone please stop texting me to see if we’re hiring a ceo k thx

Meanwhile …

I mean, c’mon lol … this board wow Sam has 100% of the leverage here and I’m not convinced the board gets that given their already blundered prior moves/comms

Scoop: over the weekend the OpenAI board had conversations with recording artist and self described stable person Kanye West about becoming interim ceo I am duty bound to note that this is a joke because at this point who can tell

I did a lot of random outside of work things in my 20s while my wife was in med school, residency, and fellowship Among the many things, got my masters in CS Got several personal trainer certifications and wrote a few fitness books Started a few sports teams (mostly basketball) in local rec leagues Taught programming for a few semesters at a local college at night Started a Rails user group and organized a local tech conference Built a little used web framework for Java (it sucked) Got super into niche Linux and Python oss (ended up working at Canonical bc of this later) Built a click fraud cloud by mistake when experimenting with early version of ec2 Built a super janky Erlang web framework (this was like … 2007 or 2008?) as an experiment (see click fraud cloud above) And … kinda random but I got my real estate license so I could buy/sell for myself/friends & family and get around the stupid commission fees. Getting real estate license is, and I cannot emphasize this enough, one of the easiest things one can ever do It’s 90 hours of classes (when I did it I had to physically go to a class but it’s online now), an exam, like $100 for fingerprints and that’s it. After that you have to find a broker (yet another racket) but there are these low-fee/fixed-fee brokers a realtor can use. They took attendance in class so I showed up and just did work work on the laptop. There are laws and regs you must know but you literally just read the book and take the exam and that’s it One thing that stood out to me when I was taking my exam (again, physically at some place). A decently large portion of the folks were on their 3rd or 4th or 7th attempt at passing the exam. I don’t recall if there was a limit to how many times one could take it. All in all I think I did about 7 or 8 homes for friends and family and my own. How many *total* hours did I spend on those? About 108 meaning 90 hours of courses, 1 hour exam, 1 hour fingerprints and 2 hour per home on buying/selling details, and paperwork with the broker. The broker did even less work I passed the fee savings on to the friends/family so it was a nice little extra for us as we were getting started in life but it wasn’t a lot of work and it was most definitely not worth 3% And this was all before Zillow/redfin which now makes all this just…the dumbest thing of all time. Whole thing needs to just die

Happy new year! Staying at this Airbnb for NYE & noticed the hosts’ cast iron was pretty dirty so wanted to do the nice thing & get it all clean as a surprise Pay it forward in 2024 people!

I think it does people an absolute disservice to tell them that if they have some large goal or dream anything other than the truth which is this: to do almost anything big or audacious means some extreme elements are needed And I mean extreme

Sentiment on Twitter two weeks ago: Google is fucked, Gemini is delayed, doa, fire Sundar Sentiment today: Gemini is amazing. That demo 😍, benchmarks look clean! Reminder to appropriately filter twitter sentiment….all ways and always

For a second let’s talk about what actually matters when hiring (or drafting) someone & it literally is just two things 1. Is the person capable, or capable of being capable 2. Can they take a punch to the face Nearly everything else doesn’t matter in the grand scheme of things

Kids had been asking to see The Marvels so we saw it for tonight’s family movie night My review: oh, oh my god … somehow this movie was actually made and released. No one, don’t care who, stopped this wreck at any point in production. Windows Me level bad

If you understand this post, you'll understand why BigCo needs to buy SmallerCo to do anything interesting after a certain point in time Past a certain stage, most BigCo are essentially PE holding companies bc they have lost the ability to build interesting things of their own When they get to this stage their treasury is their main asset and the people who embody the below are there to mostly eek out protectionist long-tails on the various revenue streams FWIW, this is actually *correct* career advice for normies to survive inside most BigCos and it's also why outliers never stay at BigCos unless they are in one something like 5-12 possible roles at those BigCos

Quora “competes” with OpenAI like I “compete” with Lebron for better basketball player. Lebron doesn’t think about me ever Still insane he was on that board, not because of competition but because quality

Maybe the real board governance was the friends we made along the way

Sure, someone can be absolutely fools gold lucky, but that’s not a realistic viable path for everyone let alone most. You can’t bank on luck even if it seems most want to What you can bank on is willpower, endurance, grit, tenacity, putting in the work, day after day improvement

Same energy

One of the biggest mind fucks of weightlifting is if you’re really really really good at it and somehow through discipline, diligence, & mind numbing consistency you’re able to maximize your natural potential, every workout after that one is doing everything you can to not lose it all AND STILL losing to time & age And this is your ‘god mode, I beat this whole thing’ scenario what a dumb hobby

I probably gonna end up dying on this hill, and I know I'm fighting the twitter zeitgeist tide, but whatever Downloadable weights is not open source AI. It's .EXE available from the '90s/'00s The dataset(s)/arch is the source, & the dataset is the more important of the two

There’s a reason the best seem obsessive There’s a reason hall of famers workout like fiends There’s a reason people who accomplish stuff are away switched on There’s a reason the greats never slow down

New King of Denmark. For those not in on these sorts of euro things, that's basically like director level at Microsoft or L5, maybe L6 at Google

Every few months this needs to be restated and because at every turn and every improvement it becomes more and more self evident. And we simply shouldn't be oblivious to the reality of the physics involved and ignore the reality that is plainly in front of us Here's everything in a nutshell: Application wrapper approaches to AI are making up for current generation model capability gaps. The surface area available to apps/agents/wrappers (whatever you want to call them) is essentially the surface area the models don't *currently* do that great at. As those models become more capable, the surface area available to apps becomes thinner, sparser, and much less distinguishable as "differentiated". Essentially apps all get pushed to the same surface area at the same time Meaning: value has and will accrue to the agi companies (as I talk about them in the QT below) and *only* becomes stronger as those models become more capable. This evolves 10x as those agi companies realize they have to own their own full stack and build vertical approaches (which we have finally seen evidence of from openAI) This is so painfully obvious it cannot be understated. Any other thinking here misses macro picture and attempts to maximize micro along local incentives Intellectually we know this as industry and we already have prior art on this from previous AI model generations (ie this current wave). Any (all?) apps/agents that wrapped GTP2 got wiped out/mostly become irrelevant with GP3, same for GPT3 wrappers with GPT4, same with....etc etc (and if we looked back at other platform shifts in history, we see the same thing play out in rhyming patterns) The surface area for things that wrap become smaller and weaker. Doesn't mean there isn't money to be made, doesn't mean there aren't great founder exits to be had, and doesn't mean there aren't good investments to be found. What it does mean though is one must *really* understand what is stone and what is sand. Most things are sand and will be washed away And here's the key, there are always gaps for companies to be built, even great ones. You simply have to navigate to white space that the agi model companies don't care about as much *and* find that surface area where the model getting 10x better can't do the work natively Here's a thought exercise. Imagine it's 2006/2007. You have the ability to put $1m into an independent company called AWS or Azure or GCP. Now, you also have perfect forward looking insight into the top 10 biggest venture backed startups built on AWS/Azure/GCP all the way to 2025. Where would you rather put your money, time, energy, and effort? One could Invest in one or all of those three mega platforms that power nearly every other company on the planet, or into those 10 biggest winners built on the new platforms. There is no right or wrong answer here fwiw, just inclination. Some people will be drawn to investing in apps for a variety of reasons, some people will be drawn to investing in the mega entities powering those apps. Each has pros and cons and both have their own version of upside, though obviously upside disproportionately skews to the mega platforms And the same goes for builders. Some want a shot at building something that could end up changing everything at once and become new bedrock for next layer, and some have see ways to use that new tech layer to solve more specific problems, perhaps with a discrete focus. No wrong answers, just viewpoints. We need both, though let's not confuse what is happening So, tl;dr - It is as it has always been. Which is also an inconvenient truth. Several super successful players chasing something immensely large, probably uncomfortably large, will end up being foundational tech for everything else in the world, thereby cementing a new platform layer upon which the next generation of interesting applications and narrower use-cases can be built. And hence, a new era is born History rhyming

It’s such a common failure mode. Every exec offsite should have a few straight days of just people making new accounts and trying to use all the damned things. I’m looking at you windows

Wife is away so tonight’s dinner conversation with the kids was introducing simulation theory and trying to drive the base reality concept home Youngest just yelled “neeeeeeeeeeerd” from the other room as she walked away so I got that going for me

“No please go on. I would love to hear about your tweet that almost went viral and how your LPs don’t understand why you went all in on vertical car wash saas companies and future of work bets. You are so misunderstood, all my friends think so too”

Mark Cuban getting his affairs in order just in case he gets asked to be OpenAI CEO at a moments notice

IMO the most regressive thing we have as humans is ideology, and particularly religious ones we imprint on kids all around the world at the most impressionable ages of their life. It’s tragic to see it this pronounced and this is a very typical example in the American south, but it’s also not unique nor particularly novel as it’s so ubiquities in other forms Like the political versions of this with the dsa, whatever ‘ism’ became the new goal post, or many current ‘cultural movements’ that masquerade as some form of acceptable in whatever packaging they come The amount of energy needed to find your way out of this as a child, young adult or even adult is tremendous (if one is able to), not to mention the wasted time (which cannot be regained) Of all the things that frustrate me with the world, this type of thing when taken in the totality of humanity is it. Entire lives are wasted, millions of people years worth of progress not achieved, and the overall drag on our collective outcome pulled down. Nothing pains me more than this

"It is difficult to get a man to understand something, when his salary depends on his not understanding it." -Upton Sinclair Always astounds me how much economic value is just lost to middleman systems but particularly real estate Hopefully Canada follows suit

16yo (6’4”) has started to refer to his height as ‘normal person size’ when he’s around me (6’0”) He knows exactly what he’s doing and honestly I’m impressed with his banter

Jesus people. The deadlift is overrated but the squat really is king. Sure you’ll have to adapt as you get older but there’s always belt squats or front squats or zercher etc let’s not overreact here

Yesterday today and tomorrow Most things being built *right now* are trying to codify and compete with yesterday in new ways The majority of the rest live at the current forefront and build for what they see there, which is building for today Very few build tomorrow

Next Canadian election is going to be a wipeout of epic proportions and it seems like Trudeau and his folks actually want it that as they triple down on just the dumbest possible things rather than do the needed hard things

lol man this is INSANE If only John Locke & Adam Smith could reach out from the great beyond & slap everyone involved with a talk on unintended consequences I'd say thankfully this will die, but our elected officials are minimally mentally enabled these days soooooo....

Who would have guessed that it would be Quora that beat OpenAI in the end?

I’ve been waiting for this moment my whole life. Let’s goooooooo

I aspire to not be very good at anything

The quickest way to lose faith in a government is to be forced to interact with any entity the government runs Every government official should be required by law to file their own taxes, schedule their own doctors appointments, register their cars, renew their licenses, go through regular airport security, file an insurance/public med claim etc etc. This should happen at every level (federal, state, town). It should be base requirement for holding office And every quarter we should hold public efficiency hearings for all government agencies with every head and deputy head being forced to stand before the people defending the department efficiency and ability to actually perform it's defined function *well* Nothing like using your own product to know when and where it's broken

Listen, I don’t care what you’re thinking you want. What I’m saying is if you really really really want something and you want to be great at it, you’re gonna have to change your mindset. I’m not even kidding a little

Just got this. Can’t remember what it was anymore. Anyone remember?

Travis Kelce showing Taylor Swift his NFL championship ring

“But Jason, that’s restrictive eating and unhealthy!” Yeah cool, be mediocre for the rest of your life “But Jason, working out that much means I have to give up something else!” Yeah cool, be an adult and honest with what you want then idk

New traveling salesperson problem just dropped

Me and the boys as we write up our end of year fantasy football performance reviews to send to our players, coaches, which totally matters and they will absolutely take seriously for next year and is not at all a waste of time for everyone involved

Oldest after trying 100% dark chocolate: “I now believe in hell”

Insane, whole thing. Feels like we are only a bit through the book and this is just the first major twist. Oceans 11 style pull it all together needed at this point

And more, truthfully, that’s hard mode! It’s actually easier to say ‘you want some outsized goal? You’re have to put in outsized effort, determination, and willpower’ and realize that pie & ice cream won’t be a thing for a while. Or yes, you’re gonna have to be at the gym more

Layer on Google’s particular lack of commercial and customer sensitivities and you have a recipe for disaster…and even so it is a $1.5T company…but one that should be worth way way way more, which is the real problem here given what they have under the hood to work with

This is that sr dev that only talks about clean code, constant refactoring, and needing to rewrite the entire code base in rust to shave 4ms boot time yet never ships anything though he could ‘build it in a weekend’

Community notes ftw Protein is the literal cornerstone of an optimal human diet and anyone who says otherwise is lying, uninformed, or selling you something

Airline A: “laptops must be stowed under seat for takeoff and landing” Airline B: “nah, that’s some arbitrary bullshit. Keep the laptop in the seat back pocket. It’s fine. A is a moron. But you must detach your iPad from the keyboard though, for safety.” Airline C: “what a hunch of losers A and B are. Those make no sense, go nuts but you must be unplugged from in seat power for now. No you can use a batter pack and have the wires everywhere SO LONG AS ITS NOT INSEAT POWER…for real, unplug the usb my guy”

Wife: “Why don’t you buy the cheaper dumbbells, those hex ones?" Me: “Because I already have the round ones and I can’t mix up the consistency" Her: “But they’re cheaper and weight is weight" Me: “still…" Her: “….you’re an idiot..." Me: “Ah yes, but I’m *your* idiot"

I will keep saying it AI doomerism is the new video games, rap music, Y2K, killer bees, bird flu, mayan apocalypse, 5G, etc pick your moral/hysteria panic where some people make an entire career out of scaring the world for mostly personal gain

The “give them the benefit of the doubt” is they have no idea what would happen. The “they know what they’re walking about”’version involves killing so so so many people it’s insane

People, as a general rule, are really bad at dealing with exponentials. When we discuss AI and its future impact, most can discuss exclusively what they see today and some mild iteration of that today data point. It’s exceedingly rare to be able to see the curve and hence, the future and ramifications Which leads to really really messy time period where we end up in the midwit meme for nearly all things (investing, education, govt, geo politics, personal etc) with the vast majority focusing in things that might matter for a moment but ultimately won’t matter I have nothing to offer other than we are in that exponential and whether you see it or not doesn’t matter, but for sure work to understand it and that will determine how well prepared you are for that future Figure out what matters and once you can determine that, can start to filter out the noise

Sam has a chance to do the funniest thing

gotta find a group of people that absolutely give a shit and get after it with the heat of a thousand suns