A place for my raw, authentic, and poorly written thoughts and ideas. The content is in order of recent to oldest.
A kid comes up to me and says "if I start to seriously learn programming now, I will still always be behind you". I also hear things like, "you can do it because you're exceptional, not everyone is like you".
It's heartbreaking to see this mindset. I've been programming for 10 years, and because I'm decent at it, people assume they shouldn't even try since they'll always be "behind" me.
But imagine me thinking I'll always be behind geohot or karpathy or sutton or carmack. That would be equally absurd. These things aren't on a straight line as your school has you believing. There's no consistent comparison or competition going on.
The scope of knowledge and skills is vast, there's ample space for you to carve out your own path. Do what you like to do, try to help out where you can, take your time. You'll be "exceptional" after 10 years too, just in your own way.
This is perhaps a followup to the previous note. To achieve more things, I feel I now need leverage. Leverage is an important concept. Leverage (to me) means being able to do more with less effort (resources). For example, hiring someone to do something gives me leverage. I follow Naval Ravikant's ideas on this. There are 3 types of leverage: labour, capital, and code. Labour and capital are the traditional forms of leverage—hiring people and funding projects. Code is a new form of leverage thanks to the internet. Code means that machines can do things for you. It's highly scalable, permissionless, and reproducible. (There is also content he mentions.) To accomplish all the things I want to do, I have to acquire some form of leverage.
A couple of projects I want to do. One is to improve economic understanding. I want for myself and other people to be able to understand economic changes in the world. These could be through government policies, technological changes, natural events, etc. I don't know how to achieve this, but I was thinking of a public economic simulator/model. Reading the MoM of the state bank's meetings, I can tell they have a model using which they forecast things. The other one is to make it easier for people to consume news. This includes avoiding bias, avoiding slop, avoiding overload of info, etc. I don't know, it's actually super hard to consume news properly. I don't know what/who to believe, and I sure as hell can't investigate myself. I am not well versed on the sources such that I can identify their unique biases. What would a solution for this look like? An AI news aggregator, deep research-like tools applied to news sources, augmenting news with social media stuff, etc.
There are a couple things that I know/have done which would be useful for young computer scientists or engineers (I think). What is the best way to do this? The goal here is not to teach something new, but to bring stuff into the awareness of young engineers. One of them is better served as a tech blog post. The other one I feel is better as a lecture. Maybe I do both. Interesting that I want to pass on this stuff. A reason maybe that I may be the only person in the city that can "teach" certain stuff. For example, given that I built OpenVoiceChat from scratch, I maybe the only person that can convey the tricks and subtleties of building voice ai systems (probably not true). Same I feel is the case with data compression from an AI perspective. Having attempted (poorly) the hutter prize and extensively studied data compression and how intelligence is compression, I may be the only person that can effectively distill the stuff or direct young engineer properly. Well, they can of course go and learn all this stuff themselves. The goal here is to add this to their awareness such that there is a higher chance that they go and learn all this stuff themselves.
Researching and understanding things should be very accessible today, sifting through the internet and finding relevant information. As part of work, I've been doing feasibility and technical implementation research (figuring out if a thing can be done, if so, how) and it's been very rewarding. It took me back to the school days where I would write reports on things citing all the sources and giving a nice breakdown. At the time I felt I was actively adding to the knowledge base of people and myself. These weren't opinion pieces like the ones I write in my blog, these are more like a compiled set of facts to serve a purpose. This kind of writing is something I may be best suited for, idk. I feel, there is a real gain in knowledge and understanding this way. Maybe I start a technical blog as well.I like this guy's work.
We need to get real money back. Fake fiat money means it can be manipulated easily. I don't know how to solve this tho. Gold backed currency did fine. A currency pegged to human quality of life improvements would nice to have. I proposed it to be pegged to kilowatts, but someone pointed out that then everyone will be burning their carbon into the atmosphere. Does something like bitcoin work? Doesn't seem like it.
Sharing revenue instead of "stocks". I am not a fan of the "raise money from investors" game. I propose a new investment+employment model. Employees and investors of the organization share its revenue. The revenue is in proportion to the amount of money and value invested by the employee and investor. The revenue gets divided into percentages. Some of it can go back into the organization for reinvestment. Some of it would be spent on costs of production (in hindsight this makes it profit sharing). When you are hiring a new employee everyone's percentage decreases since the new person would have to be paid in some percentage of the revenue. This way the org would be incentivized to hire people they think will grow the revenue in the future. This also makes the flow of value very direct (if one measures it in revenue). You would want the org to try to increase revenue rather than shareholder value. This sort of model may not be the best for large long term projects.
Why does sadapay and the e-wallets exist? Wouldn't all the founders get told "what happens when the banks make an app of their own?". Banks have made apps of their own. Every bank has one. I still prefer sadapay over my banking app. My hypothesis is that banks don't really know how to create a good application with a good user experience, they also probably don't care. Maybe an indication that one should pursue startups that may seem like fragile because a big player could do it. If the big players don't care enough, it doesn't matter. Also if a big player hasn't already done it they probably don't care enough, otherwise they probably had the brains and bread to do it (not sure if this holds true always).
Not a fan of the marketing, investor buzz lingo in AI. What is an agent? Tim Berners Lee mentioned "software agents" in 2001. Did we make "agentic" software then? Was it the year of agents in 2002? The investor, business, marketing game is not great. There is a lot of not value creating things you can do that are considered "good". B2B space is not really efficient in this way. It's kinda like informal loose collusion.
Writing this makes me feel like this is so much better than talking to an LLM. I get to think, plan, introspect etc. I don't have to read long responses with "balanced" explanations. An LLM reading this can be helpful. If I mention things I need to do, adding them to my task backlog would be nice. Any unanswered questions can go to the "exploration" section for me to think about it more. Any long term goals or near future plans can be organized accordingly. This is much useful to me than the "thinking w/ me" or "answering or leading my thoughts" mode. Do the manual for me. I don't want to end up in LLM knowledge silo.
All my "product" ideas are things that would solve my problems. Here product equals things I should build. Are these "good" product ideas? Idk. They might be a good solution to my problem. But will they be something that people will pay for and use regularly? Idk. Should I build things that are "good" products or productize my ideas? Just read the paul graham essay again. I think the answer is "problems that I have myself". Also I am much more motivated to solve them than the other problems. I also have a good problem understanding. Building an AI time calendar/time-task management app. The AI voice note taking app was also something that solved my own problem. I give up on these ideas because I haven't been able to get sales for them. I should be much more patient with these. I should keep improving them consistently. OpenVoiceChat was a 1 year thing. Building good things takes time, I should be patient.
Code and content is leverage in the modern world - naval. I should be producing both. Code I am doing with the product I am trying to build. Content is missing for now. I should start streaming again. I should write more. Technical blogs might be my thing. I should do an experiment and write about it. I should look at the blogs of the people I follow. The simon wilson blog is pretty cool. Andrej karpathy is great. I love geohot's blog. I should def write more.
I feel like there is a fundamental difference in how I make decisions and how they are commonly made. I need to formalize it somehow. I dropped out because I felt I could learn everything on the internet. People go to university to get a degree. I feel like I live my life as if the world is as it should be. You shouldn't need a degree to get a job if you can demonstrate skill. However the world is mostly not how it should be. If I make my life's decisions based on how the world is or shouldn't be, my decisions are also not how they should be. For the sake of making the correct decision I have to consider how things should be.