This Has Happened Before

im a software engineer with about 20 years of experience. im tired of spending my best thinking hours in a day in service of a small part of someone significantly wealthier than me that doesn’t even know i exist.

i would love to turn my own ideas into reality. these ideas do not intrinsicly make money. i dont want that to be my goal. that said, there are the realities of daily life like bills or caring for my family.

what are my options to quit my day job and work on these ideas full time? be as creative as you like - ive wracked my brain for years and come up with a viable plan that doesn’t seriously change my lifestyle.

i’m already doing a few things like trying to diversify income, but for some reason, i “can’t” actually get myself to quit my job. a big portion of it is uncertainty about how to replace the loss in income.

I had this prompt in an unsaved buffer that I contemplated sending to an LLM for weeks, maybe months. If you’ve known me at all over the last 10+ years, you’ll know that these thoughts are not new. After pouring myself into this craft, projects that I am truly proud of that I can call my own are few and far between. The ones I do complete can only be described as miracles. For a time, I started to feel like the living embodiment of this poster.

Over the last couple years, technical advancements have changed the whole calculus on side projects. For some time there, this was awesome. I finally got the Go diagramming idea out of my head and used it to clean up Way to Go in a way I had been meaning to do for years.

Naturally, I started plotting out larger projects on one side and vibing out all sorts of silly things on the other (little games, ideas, whatever I could effectively one-shot). I felt like I was finally starting to get some traction on doing the things I wanted to do in the way I wanted to do it. The obvious thing to do as a followup was to start plotting out all the ways I could make this my full time gig - building out cool stuff on my terms. I planned out dozens of projects that seemed to have a market need that I could do ranging from developer tools to Etsy dashboards, all using AI to do research.

Still, there was a nagging feeling - if this was so damned easy for me, what’s stopping literally anyone else from doing this? The rationalization that I made was that somehow my tenure in the industry afforded me a small moat of sorts. After all, I can write code and design the systems that need to run the code. I could even do some data analysis, product planning, and design work. Maybe AI would get to the point where that moat disappeared, but I had a window to work with so I plodded forward.

Something big is happening.

This post by some AI startup founder made the rounds. I’ve seen a lot of the advancements that Shumer mentions first hand. While some of what he’s saying seems like speculative hype (i.e. bigger than Covid, mass white-collar job displacement), the main catalyst is hard to deny. AI will only improve.

In late 2004, my college roommate introduced the game of Go to me via an anime called Hikaru no Go. This inspired me to actually learn to play, but there weren’t any local clubs. As well, I didn’t like the idea of mingling online as a weak beginner. A common recommendation at the time was to download Gnugo - one of the strongest computer programs at the time.

    White (O) has captured 0 pieces
    Black (X) has captured 0 pieces

    A B C D E F G H J K L M N O P Q R S T
 19 . . . . . . . . . . . . . . . . . . . 19
 18 . . . . . . . . . . . . . . . . . . . 18
 17 . . . . . . . . . . . . . . . . . . . 17
 16 . . . + . . . . . + . . . . . + . . . 16
 15 . . . . . . . . . . . . . . . . . . . 15
 14 . . . . . . . . . . . . . . . . . . . 14
 13 . . . . . . . . . . . . . . . . . . . 13
 12 . . . . . . . . . . . . . . . . . . . 12
 11 . . . . . . . . . . . . . . . . . . . 11
 10 . . . + . . . . . + . . . . . + . . . 10
  9 . . . . . . . . . . . . . . . . . . .  9
  8 . . . . . . . . . . . . . . . . . . .  8
  7 . . . . . . . . . . . . . . . . . . .  7
  6 . . . . . . . . . . . . . . . . . . .  6
  5 . . . . . . . . . . . . . . . . . . .  5
  4 . . . + . . . . . + . . . . . + . . .  4
  3 . . . . . . . . . . . . . . . . . . .  3
  2 . . . . . . . . . . . . . . . . . . .  2
  1 . . . . . . . . . . . . . . . . . . .  1
    A B C D E F G H J K L M N O P Q R S T

black(1):

Gnugo’s starting interface

I downloaded it and I was off to the races. Over the next several weeks, I played and lost 50 straight games. On my 51st game, I took my first win, a bit of a fluke. As it turned out, Gnugo was sometimes bad at playing out precise sequences. I accidentally killed a large group that Gnugo could have saved.

White (Gnugo) controls much more territory and is clearly ahead and can stay ahead if it stays connected here.

Now these stones are isolated and dead. Black (me) takes the lead.

So that was the state of the art in 2005. A beginner whose training consisted of just playing a bunch of games could score a win against the strongest computer of the day. It was unthinkable that this version of computer Go could ever beat a competent human.

Over the next several years, I immersed myself in the game, got decent at it, and eventually plateaued. I joined the community, dabbled in coding projects around the game, and spent time teaching and volunteering.

Meanwhile, the application of Monte Carlo tree search to the game produced a huge breakthrough. The computer could now play at a strong amateur level. It got so strong that in 2014, Crazy Stone managed to defeat Ishida Yoshio with a 4 stone handicap. Ishida is a Go master who was one of the top players in the world in the 70s and 80s, coincidentally nicknamed the “Human Computer” due to his endgame counting precision.

Still, the Go community was split on whether computers could ever beat humans in an even game. Even Remi Coulom, the author of the Crazy Stone, thought it might be 10 years before computers could challenge humans at Go.

Fast forward a couple years to 2016, AlphaGo was ready to take on Lee Sedol, one of the top players in the world at the time. In the lead up to the match, many thought there was no way AlphaGo would stand a chance. After all, just a few months prior, AlphaGo managed to beat a European champion in Fan Hui, but had noticeable flaws in its game.

What happened next is history. AlphaGo dominated the 5-game series 4-1. Instead of despair, onlookers took that one win and spawned LinkedIn articles and inspirational branding. Lee Sedol had somehow become a study in human resilience.

Move 78 by Lee Sedol which turned the game dramatically in his favor.

The fallout for those actually affected was both practical and profound. Go masters whose livelihood depended on teaching found their market had dried up as AI teachers proved to be more practical and available. As a whole, they mourned the loss of something more fundamental.

With AI, however, we all realized that the best way to reach the highest possible level of Go is not through thinking about it for a lifetime. It’s actually to buy more powerful GPUs and a well-trained deep neural network and have it play Go.

Lee Sedol would retire in 2019, citing undefeatable AI.

I wish I could say that everything will be OK. That our livelihood as engineers will stay intact. That white-collar work as we know it will not face the disruption that Shumer predicts.

At the very least, I wish that the story arc of Go and the disruption caused by AI could somehow help us figure out how to proceed from this point.

Honestly, though, I’m not sure what happens next. I don’t think anyone does. The Go masters didn’t see it coming, but we get front row seats and we’re encouraged by everyone around us to throw fuel onto this fire.

What will it look like when our AlphaGo moment comes? Will it have helped to see it coming?