we should all be thinking less

draft from 2023

(i wrote this 2 yrs ago & forgot about it lol)

I’ve been a bimbo in training for the past couple years. I can’t say I’ve nailed all of it but I’ve definitely learned some esoteric bimbo knowledge, principally: how to think less.

1. at the individual scale: Always ask yourself if you’re thinking about the right things, you wouldn’t want to waste your precious compute on something irrelevant/useless/harmful.

2. at the communal scale: Bounce ideas. Offload your compute whenever you can.

3. at the broader scale: consume all the best precomputed answers. aka read and listen.

I would rather not need to know everything and trust the systems around me. Sorry not sorry, we don’t all need to care about climate change, wars, etc etc. Before you cancel me—of course I care about climate change, I care about war and death, but I also care about solving the problems I am most immediate to and after a certain point we’re just overstimulating ourselves and the returns from thinking about “how afraid I am about climate/war/etc” vs developing niche mental models to understand the world around me are… well, one will give me unactionable-anxiety and one will make me Cool.

One thing I love doing at parties in san francisco is offloading my most pressing questions onto other people, brilliant people, (it’s free compute, what developer would ever turn that down??) this has included questions like:

• “how could we make the economy maximally efficient?”

• “what are the most important memes to teach to kids?”

• “how does my 20 step plan for being benevolent dictator of the world sound? anything im missing?”

• “should I get a face tattoo?”

Since this is a public forum and I don’t see a reason to gatekeep my thoughts: if you would like to do some open source computing for me here’s what I’m thinking about:

TechnoDarwin

Can we create evolutionary principals for technology? I’ve been imagining a web/tree tracking all of it, and a set of many principles that guide its evolution, just like Darwin’s. (ex: new chip fab technology creates a new eco[no]logical niche for chip design, new battery technology creates new demand for raw materials, etc). Can we build a conceptual taxonomy for companies? ie: Database company, Discord company, ChatGPT wrapper, etc. Can we map their relationships? Can we predict humanity’s technological evolution? Identify where things are bottlenecked? Can we explore alternate histories that would have led to different technological pasts and futures? Could we build a dynamic “Periodic Table of Innovation” and map the relationships between effort, capital, companies, and human wants and needs. The periodic table analogy is interesting because we were able to make a map with spots for elements we hadn’t discovered yet, but knew existed because we knew the underlying rules. I think this would bring light to how much human capital truly goes to waste. I think only ~1% of human effort matters to humanity in 100 years. Imagine how much cooler 100 years from now would be if that were 2%. The future will be full of industry births and deaths which could be accelerated through calculated invention, investment, and legislation, which is only possible if you have a good enough map of the territory.

Meme Based Language

I’ve been using a meme language in my mind. It’s fun. Thinking in memes makes my subconscious speak to me in memes.

Memes are an underrated form of language. They’re used on the internet but not in any serious way because it would take too much time to speak. You’d have to make the meme and then assume the other person already has familiarity with it if it isn’t self explanatory.

A meme feels more like a piece of software than a sentence to me. for example: a 2x2 grid is a program. you have axes and values and use it to convey a relationship between things. You could try to explain the same thing with words but it wouldn’t be the same. Images are more sticky than words and can communicate more more succinctly. Instead of letters they use vibes and only through the interplay of each element do they form incredibly specific meaning. The scope of what is communicable through images is greater than what’s possible through words (words are a subset of images, you can put words into images but not the reverse). I’m imagining a hilarious new Hieroglyphics language. It would be cool if we could write with memes+ (and if it was a new language with a vocabulary and grammar). I think it would unlock more depth and absurdity in our communication.

We have memes because we have computers and the internet. It will be cool to see how language evolves once we get BCI and more.

Can we solve Internet Fights All Debates? How?

I love a good internet fight, but once we have AGI won’t all sides of an argument be computable and a true answer be reached & explainable to all sides? Can we distill information well enough to convey the nuance of the answer to a point all parties would be satisfied? Or say with certainty that something is unknown?

Will people still engage in pointless debates for fun? I think so. Do you disagree? Wanna fight about it?

The part I am most curious about is how to get everyone to agree, or in nerd terms “Interpretability” aka the “we traced the edges, there are no holes, we promise. are you sure? let me see. ok yeah, ship it.” My suggestion: Find all the patterns in explanations, make short codes for them, shortcut the loops. Basically hold court using a new Code of Laws for Code and compress the results so they’re easy to verify.

What I think this will look like: a finite list of questions to ask to sus out if something is a good answer and like 200 possible responses to them. Some answers necessitate follow up questions. Imagine this done by a computer for every debated subject.

a note from 2025: I was really obsessed with TRiZ back then & liked the idea that flow charts and checklists could be built out of principles identified from looking over a lot of data related to whatever subject matter was at hand for use in “filling in the blanks.” and that it could be used for settling debates, but I think this approach would solve debates more by giving both parties a common enemy than by providing data driven truth.

Which will be solved first? AGI or Search?

I have a feeling it will be AGI. Can we build better search/lookup into AI? “Compute once, reference later” doesn’t make sense for most things because compute is cheap, but it makes tons of sense for uniquely human or difficult work (like photography, math overflow answers). I wish we could better index human labor so repeated work can be shortcut. We capture so little of it currently. I’m really worried that the quest for Good Search will lose steam after AGI. Perfect search is almost impossible and good search is complicated by economic incentives. What we have right now is pretty-terrible search. I’m saddened that we might not know what we’re missing because we would’ve never learned the habit of searching for it.

2025 update: I am more hopeful about search—internet search because of exa and personal search because of ememe. I’m also aware of a lot more nuance to this and think it’s a bad question because of the lack of defined goalposts for search and agi. they’ll assist each other a ton as they develop.

I wrote this post while experimenting with thinking less and silencing my mind. I did this through the lens of being a bimbo instead of being a zen monk because it was more fun that way. Sadly I’m wired to overthink, so I thought a lot about not thinking. About how the future is better where we don’t have to think about as much as we have to now, where past and other humans have done the thinking and we get to think about new things. About how to accelerate the pace of this, how to invent new internal mediums of thought, how to write/compute proofs for debate and decision conclusions.

It’s funny to reread because these ideas still influence my work today & back then they were just lil seedlings. I could leave this in my drafts, but there’s no reason not to post. not going to overthink it. send tweet.