alexpotato

I was reading about Tao's efforts to get more people to use Lean and apparently a big roadblock for people is that Lean uses very specific static typing.

e.g. to use a very simple example on a white board "3" is "overloaded" as:

- the integer 3

- the rational number 3

- the whole number 3

- etc

When you write a proof in Lean, you have to specify the the type of "3" you mean.

Having using Python/Perl and Java over the years, I get that some math folks found handling this daunting or at a minimum friction to getting into using Lean.

LLMs seem to have been a big help here just for the "translate my math notation into a proof" feature.

I work with several researchers (primarily working on consensus protocols etc).

We were discussing research in general and I asked them: "Do you prefer the writing of the papers or the research?"

They, almost unanimously, agreed that they preferred the research. This makes sense as if they preferred writing they probably would have chosen another profession.

I say this b/c having LLMs available to turn research diagrams, code etc into a paper (or at least the starting point of a paper) will probably lead to MORE quality research papers. This is b/c I'm sure there was some friction in a researcher's mind of "I would love to do the research on this but don't want the trouble of writing the paper".

Put another way: on a 2D plot with one axis being the skills as a researcher and the other being hatred of writing, LLMs may "unlock" the people high on both axes to get more papers out.

Post Script: I agree that this could also lead to more BAD papers but the net may turn out to be positive in the long run.

Basically, what happened is that they rebuilt it from proof of ownership and testimonies of the people

In a similar vein, I was once curious how you would prove your identity if ALL of your relevant documents (passport, driver's license, birth certificate etc) were lost in some kind of cataclysm e.g. a house fire pre-digital etc

Turns out there is actually a mechanism for this:

- get multiple people to sign sworn affidavits that you are who you say you are

- that begins the "paper trail" of evidence that allows you to start getting the rest of the document chain

- you rebuild from there.

If you're married, there is already a similar process for when a spouse takes the last name of the other spouse. The marriage certificate is the first step and then it goes from there for driver's license, passport, credit cards and so on.

Romanian friends have told me that this is really due to corruption.

Specifically:

- government gives IT/data contracts to cronies

- cronies don't actually do any real security work to protect the data

- things like this happen

This reminds me of the stone carvings found in Scotland showing grape vines growing there.

In other words, it used to be MUCH warmer in the northern latitudes.

Best quote over on distributed systems as applied to humans (from Dan Luu):

"Everything we've looked at so far is a technical problem. Compared to organizational problems, technical problems are straightforward. Distributed systems are considered hard because real systems might drop something like 0.1% of messages, corrupt an even smaller percentage of messages, and see latencies in the microsecond to millisecond range. When I talk to higher-ups and compare what they think they're saying to what my coworkers think they're saying, I find that the rate of lost messages is well over 50%, every message gets corrupted, and latency can be months or years"

- from https://danluu.com/sounds-easy/

This is an excellent used of student activity funds.

In fact, a lot of colleges have a pool of money for just this kind of activity.

Rutgers used some of this money to organize a paintball trip. On the way back from that trip is how the club got started which then turned into a team etc. I was one of the first officers and that's what led to me working in pro paintball several years later.

Celebrities and hedge funds will often do things like "rent out Disneyland for the day" as well.

One of their front-page selling points is that they use "ancient techniques" from books written prior to the year 2000, because presumably everything newer than that is bad?

I get what you are saying here.

At the same time, was reading Deming's last book [0] and there is a great line that I have never seen in any recent business books:

"The organizational chart is usually a pyramid. The main benefit of this style of chart is to clarify who reports to whom. Notably absent from this chart: the customer."

I've also worked at firms filled with very smart people who had no concept (or were not incentivized) to think about the Theory of Constraints, Total Quality Management or that the firm is supposed to work as a team and not a collection of fiefdoms.

What I'm saying is: just because a book was written before 2000 doesn't make it irrelevant.

0 - https://amzn.to/4waEjA2

Paintball had a peak in the mid-2000s (I would say 2006) due to a combination of:

- Millenials who were kids of the baby boomers being in their late teens early 20s

- Disposable income due to the real estate bubble / positive consumer sentiment

It dropped off a lot after the 2008 GFC though.

BUT

A lot of those kids playing in the mid 2000s are now parents of ~10 year olds so apparently there is a bit of a resurgence going on.

So I wrote a whole bunch about college paintball statistics here: https://www.pbnation.com/showthread.php?t=3949120

Then I wrote some more about pro paintball stats in the below three Reddit posts:

1. https://www.reddit.com/r/paintball/comments/1h17f2m/intro_to...

2. https://www.reddit.com/r/paintball/comments/1jy5xqp/paintbal...

3. https://www.reddit.com/r/paintball/comments/1k6bzi7/paintbal...

Some highlights:

- I started with just pen, paper and a stopwatch (as a college coach)

- I assumed paintball would be more like football where it's hard to track individual effects

- Turns out it's a surprisingly simple and stable "state machine". e.g. the odds of winning with +1 body (e.g. 5v4, 4v3 etc) is, in college, about ~75%

- Paintball is one of those sports where "the weakest player determines the outcome". Why? b/c if 1 player gets out early, you are fighting out of a hole.

It also made me appreciate that as good a book as Moneyball is, reading it after you try to create analytics for your own sport makes it 3x as enjoyable/insightful.

One downside though:

I would watch games and I got so good at internalizing the stats per state of the game that it was like watching the world series of poker where I could see both player odds of getting eliminated and probability of winning over time charts as I watched the games. Made it harder to be the "come on guys! we can win this" coach when we were down on points + bodies.

The new model is notable for the pricing: $3/million input tokens and $15/million output tokens, putting it at the same level as Anthropic’s Claude Sonnet series and making it the most expensive model released by a Chinese AI lab to date

Interesting to see that prices are converging to an "equilibrium price" regardless of being US or Chinese.

But, for many tasks, ultra mode is possibly worse and certainly more expensive.

Went through something similar. Fable would just spends minutes thinking, processing, confabulating etc.

I dropped down to Haiku and got an answer in >30 seconds.

so advanced that it's just as readable to me as Greek

I used to feel this way about statistics.

The language and terms are hard to understand and many of the formulas are taught as "just memorize this" instead of building up from first principles.

But then I started using statistics to analyze something I cared a lot about (paintball) and I quickly realized it's like learning anything new:

- there is jargon

- and core concepts

- when you learn the above, it suddenly makes a lot more sense.

One use of these old laptop/netbooks is to have an always on server that can even survive a power outage of a couple hours (due to the onboard battery).

As others have mentioned, the RAM is really low for a desktop but perfectly fine for running a FTP/File/PiHole etc and are usually more powerful than a RaspberryPi.

They also have multiple USB A ports if you want to add storage, ethernet etc.

I think there is a bit of a spectrum e.g.

- I don't want EVERY SINGLE notification from my phone to go to my watch

- I also don't want zero notifications b/c it's very useful to get notifications from my kids, important calendar reminders, my boss sends me a slack etc

Also, the watch allows me to still be "connected" for important events but not distracted by my phone.

(I should add that Apple watches give you a LOT of control over the notifications from your phone to your watch. Not sure if that's true with other watches)

There is an interesting point about this from animation.

Imagine you are seeing a pendulum clock and it makes a "tick" on one extreme and "tock" on the other.

When they first started doing animation + sound they noticed that if you play the "tock" sound at the exact same time the pendulum hit the extreme, people would think it was delayed.

Research showed that humans required a small amount of time to "context switch" from one stimulus to another. I think it's about 1/16th of a second.

Some interesting other observations about time perception here: https://en.wikipedia.org/wiki/Time_perception

There is a story from the early days of Napster where music industry executives discovered the service.

One of the executives describes it as:

"We had a laptop open and we tried to play 'stump the Napster' but not matter which obscure song someone threw out, it was there!"

To put it in modern terms Napster indexed:

- every song

- on every user's computer

- and then indexed all of those songs on a central server

I suspect people would pay for a service that offered this (same for movies) but a combination of licensing, IP protection etc don't allow that to happen.

I will occasionally mentor folks of varying ages.

They usually end up talking to me because some project or career path they were on didn't work out how they were expecting.

I mention this b/c at some point I ask them: "Hey, do you have a blog post or website about this failed project?"

They usually say "yes, but who would want to hear about this??"

Me: "You would be ASTOUNDED at what people find interesting. Success or failure, there is always something to be learned from someone else trying to solve a problem."

Hopefully, we see more of these writeups as they are VERY inspirational (successful stories like this one or not)

You could even have a decision tree in code where the leaf nodes are primarily "run this script with these params" but some of the leaves are "ask the LLM"

100% agree that the more deterministic code the better up to the limit where you need the LLM's ability to be non-deterministic to kick in.

There is this ACM blog post called "Manual Work is a Bug" [0] that was originally written to help humans automate processes using code. I find it just as applicable today as when it was written. You and the LLM look at what has to be done and then figure out the scripts/tools to make it happen. You then tie those tools into a system.

The more I use the above the more it makes sense and the worse the whole "just commit the prompt" seems like nonsense.

0 - https://queue.acm.org/detail.cfm?id=3197520

PSA:

If you ever call 911 and order a pizza, many of the dispatchers are trained to recognize that as "I am in danger and I need an officer to come to the house immediately. The person threatening me can hear this conversation."

It works b/c it gives you a plausible reason to give your address to someone over the phone and they can give you an ETA which is also plausible of "the pizza guy will be here in 20 minutes" etc.

If people like this post, they will probably like the below post about the typography used in the movie Alien.

The site has typography analysis from several other scifi films too.

https://typesetinthefuture.com/2014/12/01/alien/

I have a whole thread about how Western Union would select new telegraph operators: https://x.com/alexpotato/status/1806069602063159676?s=20

And for those that don't have Twitter: https://threadreaderapp.com/thread/1806069602063159676.html

One of my favorite stories about Morse Code:

Alexander Graham Bell (inventor of the telephone) [0] went deaf later in life. He would go to meetings with his assistant and when some asked Bell a question, he would pause, say let me think about that, pause again, and then give a response

What was actually happening was that his assistant was using Morse Code + Phillips Code [1] to tap out the question on Bell's leg under the table. Apparently, no one ever figured out they were doing this.

Also, I HIGHLY recommend the Victorian Internet [2] which is about the telegraph and how it was discovered, adopted and then became ubiquitous. I first read it in the late 2000s and assumed it had been written a year or two earlier but was surprised to learn it had been written in the 1990s given how prescient it was. e.g. it mentioned how local newspapers shut down because who cares about local news when you have global news?

There is also a mention of how being a telegraph operator was VERY similar to be a SWE in the late 2010s in that it paid well, job mobility was very high and it was a hard skill to learn. Edison actually got his seed capital by being a telegraph operator who pivoted into repairing telegraph equipment. There are also many comparisons that can be made to AI and how it's being adopted in the economy and affecting SWEs.

0 - https://en.wikipedia.org/wiki/Alexander_Graham_Bell

1 - https://en.wikipedia.org/wiki/Phillips_Code

2 - https://amzn.to/4wx75KY (Victorian Internet)

For some background on me to set the context for the following opinion:

- been a SRE/DevOps at banks/hedge funds for almost 20 years

- now work in L1 crypto

- have been coding since I was 12 and have also been using frontier models for the past year (including running multiple agents at the same time etc).

My thoughts:

The models are indeed amazing. They can read large codebases, find bugs, infer the root cause of an issue from partial logs etc etc.

They do still hallucinate. WAY less than they used to but it's still non-zero. In a way that's worse b/c the model will spit out a complex piece of software and say "Yep, no mistakes. I even wrote tests and they all pass!" You might think "Phew, that's great!" but in the same way we've all found bugs in production code written by smart people, there will be bugs here too.

I say this not to imply that you have to read all of the code. I say if only to underline that for big complex systems, the "let's write unit tests for the parts that ABSOLUTELY HAVE TO BE CORRECT" is still just as important as it ever was. I'm thinking of examples like:

- the order and execution handler of a trading system

- avionics flight controls

- healthcare related medical devices

- etc

As an example: I was working on a complex system. I wasn't sure if the LLM code was actually correct so I wrote up a quick script that I checked, line by line, to be 100% sure it was working as I expected. I then used that script to double check the LLM. I didn't read all of the code the LLM created. The sense of "ok, now this works" was astounding.

I'll add, a lot of the developers I work with are going this "hybrid" route too where they will have the LLM write code and tests but then go back in and double check.

In closing, a lot of these big rewrites with LLMs are possible only b/c the devs KNOW, FOR A FACT, that the unit/integration tests are correct. I'm still not convinced that you can have LLMs write all of the code and all of the unit tests and be 100% sure that it's all correct. (I will admit that this has always been difficult and even the pre-LLM days were not a guarantee that all of the code wa s correct)

I am a big Twitter user and recently gave it up for lent as an exercise in "resisting temptation" (for lack of a better term).

I found myself reading a LOT more (mostly via the Kindle app on my phone).

It actually inspired me to get an actual Kindle which does help cut down on the "let me flip to social media" when reading.

Exactly this.

Plus the notifications for chat groups are basically:

- show me everything

- don't notify me at all

Back in 2005-ish era, I helped reboot a college club (I was the coach/advisor).

We started out using forum software to co-ordinate what we were doing but eventually (2008-ish) switched to Facebook as the president of the club pointed out "Alex, everyone is already on Facebook and the notifications from us are in the middle of the notifications for when the next party is" etc.

Fast forward to today and the club is rebooting again. I asked the current club president "What social network is everyone on these days?" His response: "Really there is no one place where everyone goes anymore." I then asked him how clubs share their info etc and he says "The bulletin board at the student center?"

While social media definitely has its downsides (echo chambers, extremism etc) I do feel like it's a bit of a net loss to not have a "commons". That model makes it super easy to start up new organizations, get the word out etc.

Part of me hopes that we got back to the late 1990s dedicated websites/forums. That seems to be the Discord model but let's see.

There is an old quote:

"Add comments to your code under the assumption that the next person to maintain it is a homicidal maniac who knows where you live"