naz

Implication, necessary and sufficient conditions are usually taught inadequately Here is how I would teach these concepts: blog.naz.ooo

@onlinedopamine financially, perhaps. but a lot of his takes are horrible, not a great role model generally I think

whenever you feel that you're too old to get started, remember that this dude only began building at age 27

i am doing the same. went over hammack's book of proof, then spivak's calculus, now going through probability & stats, differential equations, linear algebra and multivariable calculus wanted to build something for self-learners like us, that is focused on goat books, but licensing rights kill that idea quickly

how I’m relearning math from the basics after graduating. a thread

Here's the longer version of our Nature piece. Our argument is simple: statistical approximation is not the same thing as intelligence. Strong benchmark scores often say very little about how LLMs behave under novelty, uncertainty, or shifting goals. Even more importantly, similar behaviors can arise from fundamentally different processes. In another paper, we identified seven epistemological fault lines between humans and LLMs. For example, LLMs have no internal representation of what is true. They often generate confident contradictions, especially in longer interactions, because they do not track what is actually true. Another example. Yes, LLMs have solved some open mathematical problems, but these cases typically involve applying known methods to well-defined problems. LLMs cannot invent anything that is truly new and true at the same time, because they lack the epistemic machinery to determine what is true. None of this means LLMs are useless. Quite the opposite: they are extraordinarily useful. But we should be careful about what they are and what they are not. Producing plausible text is not the same as understanding. Statistical prediction is not the same as intelligence. So despite the hype from the usual suspects, AGI has not been achieved. * paper in the first reply Joint with @Walter4C and @GaryMarcus

@nico_jeannen maybe rather than starting with empty state, take a guess and pre-fill with some data anyway?

have been stagnating like crazy since then 💀 MRR is actually lower now, back under $900 (less if I only count Adkit) I spent 3 days investigating trying to figure out what's happening using Claude code connected to Posthog, Stripe and Heyo Chat to investigate (more on that later if anyone interested). First, I got way less traffic, but also every new subscription I got was negated by other people canceling = no growth. So it’s like less traffic + bad retention, so it’s guaranteed fail The Churn indicated by Stripe is tricky, it showed ~10% but it was closer to 40% actually 😬 Reason is because churn updates retroactively when ppl cancel, so if you just check the last 4 weeks in Stripe it will always be lower cause ppl won’t have churned yet (it’s been less than a month). But when I switched to 3 month view I saw it was much higher than expected. I thought the product was fine but it was not at all TL:DR: most people did not use the app that much and canceled after 2-3 month (either cause of bugs, forgetting about it, not knowing all the features or just because they only needed it for a short time) For ex in Adkit, to get the value of the product (ads competitor tracking), you can either follow advertisers already in the db, or add your own. But the empty state when you search and no advertiser is found makes it looks like if it’s not there then you can’’t do anything about it I thought it was obvious cause there is a big button “add advertiser” but apparently not (I saw session replays of users getting frustrated) In this case. I think the fix is just to add a better empty state with a CTA to make it very clear that people can track their own competitors Another issue is that I didn’t took the time to setup email sequences, so people who purchased kinda forgot about the app and also might not know all the features available. Then they cancel cause they don’t really use it. So now I’m making a full onboarding sequence to help users take habits of using the app, give them tips, etc. (@bento is the best for that). And then the other issue is getting started. From the data I got, people who have 5+ advertisers tracked are more likely to keep using the app and stay subscribed. So one thing I’m going to add is Starter Packs = users can follow a bunch of handpicked advertisers for a specific industry in one click That way they don’t have to browse the list of advertisers to get started => Less manual work => Less friction => (hopefully) more usage There is many more issues (lots of actual bugs), I fixed a good chunk of them already so hopefully this could improve things Big mistake is that I stopped talking to users. I did it at the beginning while launching the app but then I stopped. So now I’ll try to talk to one customer per week minimum. Either people who use the app a lot, started using or churned. Only way to truly understand users IMO. ====== Now that was ONLY for activation/retention (keeping existing users). This is only half the issue. Other half is acquisition (getting new users) I thought the app was growing well on its own, but turns out most traffic came from X (launch post + general traffic) The launch posts died out after while so traffic died eventually. Then I also stopped using X in January / Feb to detox mental health (and it did work!). but then it meant I lost most of my traffic source. I dint launch ads yet cause of all the issues I have to fix etc, but it will be time to start! Relying launch for long terms is not a strategy, esp when it comes from just one platform. I should probably try to post more about the app again But I need to get rid of doom and political posts in my timeline cause it makes me angry/depressed :D Started working on SEO a bit but never got any success with it so I don’t hope much. My friend @aurelien_dio is helping a bit so maybe this time it will work (he did SEO for big French corpo). === So, TL:DR, issue is that not only I don’t have enough water filling the bucket but also the bucket has a hole so any effort I put in is wasted. First fix the bucket, then pour more water. I know Adkit has potential, I just have to put the pieces of the puzzle together. I managed to get from 0 to 1 but now I need to figure out how to go from 1 to 2. Plan rn is to fix issues, then try new acquisition channels SaaS is fking hard, cause you dont just have acquisition, you also have activation and retention, that's like 3x more stuff to work on compared to one-time purchases (that’s why I recommend doing one-time when you start out btw, much less to manage). But it's ok, learning on the go and (hopefully) figuring things out over time I know I can make a great product, and I will ==== Some great resources that helped: longform.asmartbear.com/max-mrr / longform.asmartbear.com/customer-devel… by @asmartbear (insane this is free) Product-Led Growth by @wes_bush Deploy Empathy by @mjwhansen (best resource to talk with users IMO)

Marc Andreessen was wrong about software eating the world, and I see people making the same mistake about AI today. I wrote this almost three years ago and I wouldn't change a word if I were publishing it today.

@tibo_maker incredible! congrats

yes, I crossed $1,000,000 / month was trying to keep it quiet but it got leaked on a pod so... let's go quick backstory: I sold Tweet Hunter and Taplio for $8m in 2024 then got back to work and honestly, I was terrified when you set a standard for yourself, you have to live by it and that pressure is real every single thing I shipped felt scary - I felt like a fraud, "the lucky guy who got acquired once" every launch, I kept thinking: what if it flops? what if I go from "the guy who exited" to just... a nobody? so I forced myself to ship a ton a lot failed I was embarrassed by the bugs in my MVPs but I kept going Revid, Outrank, SuperX, PostSyncer, Feather -> five products thriving today they do more than anything I've ever built before grateful to my co-makers, to the team, and to everyone following along, giving feedback, using the products every day thank you, genuinely 🙏

if you go the route of people uploading their own books to the app, then you also want the ability to be able to actually write math in there. then your product becomes goodnotes essentially if, I am mistaken and people don't need the ability to write math in their uploaded files, and they just follow the book with pen and paper, then (ignoring licence rights for a moment) this app becomes a tool to chunk up a book into a knowledge graph and letting people progress through it independently and together - but then, this indicates that the core moat of the product is NOT "study math together", but more like "an app where you make it easier to learn any book"

the unreasonable effectiveness of math is simple actually. reality has structure, math is formalized structure, and we keep those formal structures which work. it's an evolutionary search process

there is a rhetoric in ai rn that vibing and half-assing is the future of technology. do not fall for this psyop. the future is deep understanding and mastery. always has been

The early founder years are one of the most stressful phases of life. Most are really not talking about it. - Some friends are climbing the corporate ladder and taking crazy vacations. - Someone just bought their first home - Someone already feels “settled” And you’re still trying to turn an idea into a company. At the same time: - Your bank account keeps shrinking - Investors keep saying “too early” - users never seem to care and retain on your product - People around you don’t fully get what you’re building - your mom ask when you will get a real job - Nights are full of overthinking - The runway clock keeps ticking - Time feels like it’s moving too fast yet progress is slow And the hardest part? - You smile to your friends and family - You say “things are going well.” - At 2 AM you wonder if this will ever work. But remember: - You’re not behind, you are on a different path - You’re doing something incredibly hard If you’re in this phase, be impatient for actions, patient with results and don’t give up. Building something from zero takes time. Anyone else feeling this lately? ✋

Anthropic studies how AI coding affects 52 professional developers: > the group who used AI felt “lazy” and noticed gaps in their understanding and the group which didn’t use AI felt the task was “fun” > AI significantly hurts skills formation of a new library by 17% > AI didn’t actually make people faster. the time saved on not writing code was spent interacting with AI > only people who fully delegated their work to AI were noticeably faster, but they learned the least there are three AI usage patterns that preserved learning and three that really hurt it. the first three patterns: 1. asking only conceptual questions 2. generating code, then asking follow-up questions 3. asking for explanations alongside writing code and the three patterns hurting their learning: 1. complete delegation 2. starting on their own then increasingly relying on AI 3. debugging where they asked AI to fix things without understanding why

ideally, we have all the books on our website, or people are able to upload their own ones for personal use (so that only you have access to your own uploads), but i need to do more research here, because this might prove to be challenging without breaching copyrights. so, initially, it will have to be based on the principle of "bring your own book", so the materials on the website would be guiding through the book, would have spaced repetition cards for all chapters, and study groups would have to focus on going through math concepts in each chapter. effectively, a social layer around specific books that are assumed to be owned by the users. this is just a preliminary idea that I have, it might evolve. I intend to use this website for my own learning / refreshing, so I would be improving it along with early users. do you want to jump into Discord that I just created for this website? You can find the invite link on here: math-circle.com

@AlgorithmicBot I’d try it out. is it based on your own uploads too or just the website’s library?

thinking of building a tool that I wish existed for math self studiers there are some golden standard books that massively level up your understanding building a social layer around studying them would drastically improve the chances of people completing them the biggest hurdle is obtaining the licence, which in some cases is likely impossible (Spivak, Calculus). and building this product out as a companion tool does not have a moat what the app would look like:

@Tri_Stanisaurus cooking up a community and platform for math self studiers: math-circle.com/roadmap

I want to know how a 30 y.o. can complete 6 years worth of math in a single year. I want to fully master topology with an unlockable skill tree.

my impression is that it's very easy to make money online so long as you're willing to be a boastful, ego-centric bullshitter and sell sham products to a growing fan base. you can easily be a millionaire this way. and perversely, this wealth only earns you more credibility

After people get a few months of seeing the impact of AI-induced slop in their codebases, and the resulting slowdown this incurs, I predict we're going to see a mass reversion back to "handwritten" code. TBH I'm feeling it myself some days. I'm actually faster with Cursor tab and just nailing what needs to happen. AI has a terrible habit of spending time on things that just don't matter. Might be more "work", but in terms of clock time, I'm getting things done more quickly without involving token inference. Maybe I'm weird. Maybe this is a terrible prediction (as things frequently are with things that change so quickly), but in this case I've been observing my behavior and others for months and I'm seeing a slow steady trickle back to the "old ways".

@dunik_7 @pangramlabs slop?

yesterday a friend messaged me: "i have 48 hours before the lawyers find out. you need to see this" he worked as a data engineer at a hedge fund in Zurich. he got fired on friday on monday, his vpn was still working he downloaded 3 jupyter notebooks before his access got cut one of them was called polymarket_edge_model_v4_FINAL.ipynb THE CORE IDEA: the fund doesn’t predict the outcome of the event. they predict the PANIC of other traders when the price moves sharply, 80% of players close at a loss. the fund takes the other side "liquidity vacuum trade" TRIGGER — 3 conditions at the same time: / price >2% in 90 seconds / volume >3x the average / 70%+ of orders on one side the bot trades AGAINST the crowd expected_price = EMA(200) × (1 + sentiment_offset) deviation >12% = entry stop at 25% take profit at EMA ±3% win rate: 73.2% across 1,847 trades. sharpe 2.7 i rewrote it in python over the weekend. 172 lines first week: +$412 from $1,500 the formulas are above. claude will handle the rest.

Some of our best hires were totally unqualified on paper. They always had the same qualities: entrepreneurial, high agency, smart, mission aligned, and they got shit done. If you’re hiring, especially in early stages, seek out & bet on these people. Don’t over-index on resumes.

How the fuck is this not obvious to people im going crazy

@HumanProgress how many dollars goes towards rent?

75 years ago, 1 out of every 5 dollars a US family earned went to food. Today that's closer to 1 in 10. A slow, steady, easy-to-miss kind of progress.