It is absolutely astounding how much of them run on code that is:

- very reliable aka it almost never breaks/fails

- written in ways that makes you wonder what series of events led to such awful code

For example:

- A deployment system that used python to read and respond to raw HTTP requests. If you triggered a deployment, you had to leave the webpage open as the deployment code was in the HTTP serving code

- A workflow manager that had <1000 lines of code but commits from 38 different people as the ownership always got passed to whoever the newest, most junior person on the team was

- Python code written in Java OOP style where every function call had to be traced up and down through four levels of abstraction

I mention this only b/c the "LLMs write shitty code" isn't quite the insult/blocker that people think it is. Humans write TONS of awful but working code too.

Someone asked Yuval Noah Harari, author of Sapiens, his thoughts on LLMs and how easy it was to create fake news, ai slop etc.

His response:

"People creating fake stories is nothing new. It's been going on for centuries. Humans have always dealt with it the same way: by creating institutions that they trust to only deliver factual information"

This could be government departments, newspapers, non-profits etc.

A personal note on this:

There is a Christmas card my grandfather made in the 1950s by "photoshopping" (by hand, not the software) images of each member of the family so it looked like they were all miniature versions of themselves standing on various parts of the fireplace. The world didn't collapse due to fake media between the 1950s and today due to people having that ability.

They're essentially a combination of a plane, spoke-shave, draw-knife and gouge but all in a one handed tool. They were primarily used by Native Americans to build things like canoes, snowshoes, baskets etc. I first found about them from reading John McPhee's Survival of the Bark Canoe [1] but there are lots of uses of them on video on the website below (which I created).

If you want to get into woodworking but want only a few tools and/or a very portable tool, highly recommend.

e.g. in theory you could build an entire canoe with an axe, crooked knife and 3 or 4 sided awl (and a lot of time, patience and materials)

0 - https://crookedknives.com/

1 - https://amzn.to/3NSj4T3

There is some nuance here.

Manufacturers didn't know if people preferred shorter, higher quality (Beta) or longer, less quality (VHS). That's partly why there were two formats.

Most people like to say VHS "won" but what it really won was the consumer market. Beta won the professional TV news market because it turns out news stations had a high demand for short, high quality video storage.

I point this out only to say that winning isn't a one dimensional/binary outcome. You can "lose" in one market but still be very profitable in another market.

There is a line in the book Accelerando about how evolution did this with biological vision.

It's basically the highest bandwidth sense we have and evolved AFTER smell (chemical based) and auditory (gas pressure based) senses.

e.g. let's say you sue and then win: that's now in the public record (which any new hiring company can see).

Yeah, "all bets are off" once you are a FTE.

I wasn't sure how they could figure this out at the time until someone later pointed out that many corporations do a credit history check on you as part of the background check. This gives them access to past compensation.

The information asymmetry here is, as with much of hiring, pretty bonkers when they had both the current and past comp history during negotiations when you have just yours. You might also have the comp history of your friends too (if you share) but that's still tiny compared to the corporations.

0 - this was in NYC where it's now no longer allowed.

- first used copy and paste in and out of Grok

- started using CLI tools e.g. Claude and OpenCode

- move up to using 3 and sometimes 4 agents at the same time

- considered going to the agents managing agents

- have settled on having LLMs build tools that are both deterministic, usable by humans and the agent, and also faster (b/c there is less "back and forth")

Honestly, it feels a LOT like when Kubernetes came out. e.g. you stopped running containers on a box using Docker Compose plus scripts/configs etc. Instead gave a large part of the operation to an "agent" (in this case k8s) that managed all of the details you didn't need to care about anymore.

I've also realized that while the LLMs can crank out code at a very high rate, someone still needs to make sure everything is running, debug issues etc. You could set up agents to monitor what the agents do but then you still end up with someone needing to keep an eye on everything. If anything, you need MORE people b/c now you can just keep spinning up new components etc.

Also, was in a discussion with one of the best developers I've ever worked with. It came down to the following point:

"Programming is rapidly becoming a hobby. Software engineering is becoming more important than ever."

https://superuser.com/questions/545329/how-do-i-make-a-machi...

This is the exact point I make whenever people say LLMs aren't deterministic and therefore not useful.

Yes, they are "stochastic". But you can use them to write deterministic tools that create machine readable output that the LLM can use. As you mention, you keep building more of these tools and tying them together and then you have a deterministic "network" of "lego blocks" that you can run repeatably.

Was recently watching a video on the RFID tags that Bambu Labs use on their spools and not only is the tag data encrypted, it's signed so even if you bypass the encryption, you still don't have a way to spoof the signature.

I mentioned in a sibling thread that I do this on Twitter (and it's a lot of fun to get to re-use old threads for new audiences).

> Using LLMs would in theory save some effort by rephrasing to it doesn't look copy pasted but I am strongly opposed to the mild reality distortion they are prone to doing

Same thoughts from me. There is also a bit of "John Henry" [0] in that I want to keep my brain strong in this skill versus letting the machines take it away.

0 - https://en.wikipedia.org/wiki/John_Henry_(folklore)

I do this on Twitter.

Specifically, I'll retweet the thread or tweet I already have written on the topic.

Retweeting is, IMHO, the best part of twitter as it lets you "weave" a narrative of old tweets and threads. It's also why I think articles are dumb b/c you can't like to the specific part of an article like you can with a tweet thread.

It's funny you mention this b/c "yes". Both of my parents are big storytellers.

Never realized this so thanks for pointing it out!

It’s all manual and I guess just how my brain works. My wife actually calls it “the database” because I can quickly access stories and I apparently tell them in a very similar way.

I’m just as impressed that you noticed and had the Déjà vu.

My first full time job (early 2000s) was working for a firm that did online cybersecurity related investigations for Fortune 500 companies (generally via a 3rd party law firm they had retained).

A big part of this was running investigations into people running "pump and dump" stock schemes on Yahoo message boards. We would generally start by scraping all of the posts for a user who had instigated one of these and then handing off the posts to an analyst.

It's amazing:

a. how much info people give out even when they think they are being careful

b. related to a, how even small tidbits combined over time can build a pretty accurate picture of who someone is.

e.g. they post "oh man, the Cubs lost", then a year later "went for a walk on Lakeshore drive", another year later, there was a fire at my local subway stop etc etc and you pretty quickly narrow down the rough neighborhood where they live in Chicago.

Combined with tools like Lexis Nexus and you get a list of people that you can narrow down by age, sex, occupation etc and we could narrow it down to <20 people based on other info they had shared.

Then you fold in their posting patterns and it's pretty obvious who is at work (posting 9 to 5pm) vs home (posting 7pm to 1am).

Again, you keep adding constraints and the intersection of the Venn diagrams gets smaller and smaller.

This was all in the early 2000s before we had cellphones that tracked your location and ad infrastructure that followed you around the internet.

I've been using AI to help me write it and I've come to a couple conclusions:

- AI can make working PoCs incredibly quickly

- It can even help me think of story lines, decision paths etc

- Given that, there is still a TON of decisions to be made e.g. what artwork to use, what makes sense from a story perspective

- Playtesting alone + iterating still occurs at human speed b/c if humans are the intended audience, getting their opinions takes human time, not computer time

I've started using this example more and more as it highlights that, yes, AI can save huge amounts of time. However, as we learned from the Theory of Constraints, there is always another bottleneck somewhere that will slow things down.