formalsystem

The writing process for this article was meme first, content second

It seems like the article was a bit polarizing, some of the comments made me realize I made a few imprecise statements and I'll fix those. Other comments didn't approve of my tone and that's a bit harder to fix since I tried writing this article in the same way I usually speak. I've written a lot of more dry content for my robotics ebook and I was purposeful in trying something different for this article.

Some of the comments actually generalized my observations to management and software more generally and it's always nice to see people taking my ideas further than I thought they could go.

One major thing I'd like to point out is I don't think "ML is dumb" is the right conclusion, there are lots of incentives making it so more of ML is stagnating but this is certainly not true for the field at large. Interesting ideas need to involve a certain amount of risk. The latter half of my article showcases a few ML adjacent projects which I think are absolutely fascinating.

And if you're interested in reading more stuff by me my robotics and machine learning ebook is very easy to read http://robotoverlordmanual.com/ and will teach you all you need to start building robots at home

Sure they can invest in something more productive but the whole point of having reserves is to be more robust to negative changes e.g: COVID or jump on new opportunities (some company goes up for sale).

I spent the last 2 years working on a research heavy startup that I felt would change game development forever, it was hard from an engineering, research, marketing and sales standpoint.

Late into the second year I started to feel burnt out and decided instead to write a robotics textbook which got way more attention to both myself and my project in about 2 months worth of work than 2 years did. This then got me to take Twitter a bit more seriously in terms of making friends, peers that I can easily see turn into collaborators or customers in the future. Sort of an audience first approach as opposed to product first approach. Tools like Twitter, Medium and Substack make this approach a lot more feasible today.

I even remember having a sort of disdain for freelancing because I thought it didn't scale whereas if I did pursue it more seriously I'd imagine I'd have had more direct solo enterprise sales experience, met more clients, figured out what their problems were and solving them instead of imagining an ideal where game developers would solve their problems like I wanted them to. Seeing my savings decrease every month took a mental toll on me, I figured there would be a light at the end of the tunnel and I essentially feel like I miscalibrated my goals. Instead of trying to build a billion dollar startup, I should have spent more time figuring out how to never need a job ever again.

The benefits of the multiple small bets approach is that when you're wrong it doesn't cost you 2 years of your life, you're less fragile to sudden decreases in income from one small bet, you don't have a boss, pandemics don't affect you, you choose your peers and hours etc.. Making your first few dollars online (not from a paycheck) changes you, makes a you a lot better at pursuing opportunities when they happen.

After COVID hit I started looking for a job again and my interviewing experience was very interesting to say the least. I described very honestly my failures and startups loved my story and appreciated my openness - bigger companies on the other hand essentially told me that well your startup didn't work out so you must be lazy and dumb. The outcome is what determines if I'm a good founder or not but it was really painful being lectured by people who had never taken a risk in their lives.

I see so many founders dreaming of huge companies going towards IPO when I can see solo founder companies being a lot happier if they pursued multiple small bets instead. You can't hedge your life over multiple billion dollar startups with a 1% chance of working out but you can certainly hedge it over multiple small bets each with a 10-30% chance of success. What's good for the aggregate or society isn't necessarily what is best for you.

Anyone can query at any time the total amount of Bitcoin that have been issued without relying on anyone. After downloading the Bitcoin client just run gettxoutsetinfo

You can try to guess how many dollars have been printed or how many bars of gold exist.

If you view Bitcoin as a payment network like VISA then obviously there are many things the network needs to improve but if you view it as a savings technology then it's the best one that's ever been created. I never understood how "But Can I buy Starbucks with Bitcoin" became a meme, I don't expect to be able to buy Starbucks with Apple stock.

Claims that Bitcoin burns useless energy or doesn't have feature X completely miss the point. Avoiding capital controls like in Lebanon, protecting yourself from hyperinflation due to corrupt governments with a printing press like Venezuela are Bitcoin's killer features.

It's common to view hyperinflation as something unique to corrupt countries but that's only because CPI is a dishonest way to aggregate inflation data. Inflation is a vector NOT a real number, any asset that people care about has become generally unaffordable i.e: housing, education, healthcare.

BS crypto vendors are looking for the next big application of blockchain technology when inflation protection and avoiding capital controls are arguably the most impactful innovation since the internet. Bitcoin and Time are the only 2 scarce assets that exist in this world.

I got this, it's cheap and great https://www.amazon.com/GAOMON-M10K2018-Graphic-Pressure-Batt...

I use it to annotate pdfs and also with the MS whiteboard, it's not necessarily the best to collaborate with but I really like how I can zoom infinitely out to keep adding stuff. I prefer this setup over regular whiteboards for solo brainstorming sessions

100% agree on the launch times, it gets nuts even for smaller games - I'm not sure if they're working on this.

Have you tried Unity Collaborate or Git LFS for version control? What other improvements would you like to see?

The Stockholm syndrome comment made me lol, It's just very satisfying to make games with Unity that I'm willing to tolerate/ignore the issues with it.

Most of it is a lot more impressive than projects completed with React.

Unity has a high learning curve once you want to move away from tutorials. I believe React is the same, I kept trying to learn it but would get discouraged with the new tooling and eventually give up.

Any reasonably complex interface is essentially a game whether it's a music composition app, simulator, MMO or even a basic form.

The basic abstractions of entities and components sometimes feel like they could apply to any software project. It's seamless to animate something, add physics, style in ways that would seem mind boggling in other environments I've worked in like Python or Javascript.

You can learn so much going over basic Unity tutorials on Youtube and disassembling Unity games on Steam.

The asset store has many people aggressively experimenting with better user experiences for doing certain tasks whether its character skinning or node based programming, many of the best assets end up added to Unity by default.

Unity is not mostly used for mobile games, some insanely complex games are made in Unity. You can check out this list on Steam https://store.steampowered.com/curator/31285130-Unity-Engine...

Unity ML agents is the single most exciting ML project out today which helps you create the most complex Reinforcement Learning environments you can dream of.

Unity ate the indie game market but I think they will also eat Robotic simulations, Animation, interior design, VR, Web apps and Desktop apps.

Can't wait to buy shares.

Focusing on the direct cause of the blast is a huge distraction from understanding why the 2,750 tons of ammonium nitrate were stored in the middle of Beirut for 7 years.

The president Aoun and his senior leadership were all aware of this problem but said they didn't have the authority to do anything about it. IMO, this is a hilariously bad argument that's deflecting who the most likely owner is. Aoun and his lackeys apparently have the authority to start a state of emergency and shoot protesters but don't have any such authority to prevent half of Beirut from being nuked.

The director of the Beirut port Badri Daher has been running bazaar ever since he's been in that position, regularly stealing supplies from shipments, suing reporters for defamation and beating up investigative journalists. The port director also reports to the Amal party which is closely allied to Aouns.

However, my take is that people that have mastered subjects are the ones that can create stuff with that information so I believe the goal of "education" should be to graduate people to creators as soon as possible. Instead of flashcards why not graduate people to messing around with a quantum simulator. I'm not even sure how important memory is these days.

As far as I'm concerned some of the best educators out today are the Zachtronics and the Factorio devs. I hope more people follow in their footsteps.

While I do see the value in a programmatic diagram tool, I'm not sure I'll use it over something like Inkscape if the visualization is fairly simple.

EDIT: Scroll down a bit, the original poster made their account private a few moments ago

My process so far has been I keep a daily journal where I write down the most interesting things I learnt or did everyday. If my daily notes are getting unwieldy then I use a Markdown file which I sync on Github. And when I really want to make sure that I've understood something I'll either write a blogpost or stream myself going through a library.

I would definitely appreciate the ability to search my old notes better since they got too long to efficiently parse a long time ago but this process lets me structure the really important concepts I learn without adding too much overhead when I'm just learning random stuff. I also expect my notes to outlast many knowledge base projects and part of the appeal of markdown to me is that it's probably gonna be around for a while.

I'm curious if any writers here have found personal knowledge bases to be worth the initial overhead.

This is obviously true for non fiction programming books where you can spend years developing a library off of ideas from a single chapter but I was also surprised to learn that it's also true for fiction books. My girlfriend doesn't mind spoilers in novels because she reasons if a book isn't good enough with spoilers it's probably not good enough without the spoilers too.

The goal should be to enjoy reading not read, it's deceptively unproductive to force yourself to finish a book.

What researchers want is discoverability and credit. Some researchers are already famous and own their own distribution via arxiv or Twitter but for new researchers publishing a paper at a prestigious conference is a way to bootstrap a reputation.

If you really wanna get rid of large publishers what you really need to solve is discoverability and you can solve it with a combination of virtual mini conferences and callouts on Twitter and Hacker News.

I've gotten some of the best constructive feedback on my blog posts from Hacker News directly, I don't see how that's different from peer review. I don't find it very likely that someone would even be willing to give me feedback unless they themselves are already subject matter experts.

But really the impact of a piece of computer science research is directly correlated to how many people directly use the OSS inspired by the research. People in industry know this, what's left is is for tenure committees to also take software impact into account.

Worth every penny.

If I'm rendering a forest of course I don't want to copy the data for trees a 1000 times so of course I need to use the flyweight pattern

The decoupling patterns chapter is particularly good and will help you understand how to turn each entity in a game into a server which sends events to other entities. That way you end up with many small pieces of a code as opposed to a single giant game loop. The entity component system (ECS) is one of the main reasons why programming games in Unity is so pleasant.

If it wasn't obvious by now, this is not a book about game programming patterns it's the best book on software design patterns that I've ever read. Most software design books seems to think I'm only interested in designing accounting or banking software, why not a game?

But honestly the book is very cheap relative to how good it is IMO