Legibility of Effort

So:

• Lots of work is easier because of LLMs, and
• Lots of heuristics we used to gauge effort are broken because of LLMs, so
• Conveying effort to others is harder because of LLMs

And:

• “Someone spent time on this” is a proxy for “this is worth engaging with” and
• I appreciate projects that take a lot of effort and
• I like it when people appreciate that I spent effort on something

But also:

• Making a good thing still requires making lots of decisions, and
• Making lots of decisions is still very hard, despite LLMs, so
• Making good things is still very hard

As a creative technologist I’ve found this paralyzing. And as someone who loves novel internet projects I’ve found it frustrating. I’ve heard similar things from many of my friends.

What software (and writing, to an extent) is missing now is legibility of effort - the ability to tell at a glance whether something took a human meaningful work.

I typed these words with my fingers

I wrote this blog post in my text editor. It took me a while.

Until recently, “someone cared enough to write this” was an ok heuristic. Plenty of writing on the internet was bad, but you could convince me that you cared about something just by writing it down.

Of course, generating plausible-looking text - or a plausible-looking website - is trivial now.

For me, this means it’s harder to find good stuff worth engaging with online. And on a larger scale, systems built on easily gauging effort at a glance are falling apart.

Open Source

In January, the (incredible) whiteboarding tool tldraw started automatically closing pull requests from external contributors because they’d “recently seen a significant increase in contributions generated entirely by AI tools” and were struggling to manage their contributions.

It’s disappointing - if not surprising - that people are spamming open source projects with slop.

But the problem isn’t just the spam! It’s that identifying low-effort work is difficult.

(Internet bloggers who aggregate cool stuff have the same problem; here’s Robin Sloan on the scourge of AI self-promotion. I’ve heard similar complaints from the amazing Andy Baio)

Some maintainers are inventing new ways to identify low-effort work like “poisoning” agent context so that poorly-managed agents add files that say “I am a sad, dumb little AI driver with no real skills.”

This is very funny! Unfortunately I can’t filter all of my browsing this way.

But I can learn to notice the signs.

Vibecoding

Some friends of mine have a discord thread where we log signs of vibecoded websites. And our examples are startlingly similar to what Wes mentions here:

Wes’s tweet was a hit - lots of people have noticed the same things. I think we’re paying attention to these details because they make effort - or the lack of effort - more legible.

There’s nothing objectively wrong with making a website with a warm-cream background and hero text in a sans-serif font with a single accent word that uses an eye-catching color and a different font.

But when I see a website that has the default Claude style I assume that the author put ~no thought into how the site should look. And I often assume that the author didn’t put too much thought into the site at all.

That’s not fair of me! But “someone made this website” is no longer enough to tell me that the website was important to them. So “default Claude style” is one of my new heuristics.

Hard, useless things

A few years ago I made a version of Flappy Bird that ran inside MacOS Finder.

It took a bunch of work!

This was the first project I made that got attention on the internet. My writeup was on top of Hacker News and got mentions from blogs like John Gruber’s.

John called Flappy Dird “useless but completely joyful,” and many comments elsewhere said the same thing.

But the joy of Flappy Dird wasn’t just that it was useless. It was that it was useless and that I had spent meaningful time and effort building it.

I love expending effort to make an excellent but useless thing, making my effort legible by writing it up, and then seeing strangers appreciate that effort.

And now I feel less confident that I can make my effort legible to others. For some projects I’m not sure I’ll have to expend effort at all! So I spend less time blogging about making computers do funny things.

However, I haven’t seen a flood of other people running games inside Finder.

Taste, etc

The New York Knicks recently won the NBA finals. This was a big deal in NYC; I spent my evenings running around watching the revelers in the streets.

One fun way I kept up with the celebration was GardenCam - a website that let you watch the celebration outside Madison Square Garden using traffic cameras.

GardenCam is a great, simple idea. I suspect a whole bunch of people could have vibecoded a version of it.

But they didn’t! The creator of GardenCam is my friend Morry Kolman, who had previously created a wildly popular way to take selfies using NYC’s traffic cameras.

Morry was really good at repurposing online data for creative projects before vibecoding came around, and he’s still really good at it, and that’s still important.

The most boring people in the tech world now talk a lot about the importance of “taste” - I truly hate to join them. But I can say that people I know who were good at an odd, specific thing a few years ago are still good at that odd, specific thing now.

At the same time, I’ve noticed my own taste and preferences changing.

Changing preferences

It’s still challenging to make great software. In some ways it’s probably more challenging than it’s ever been:

But some things are easier. And making my effort legible when I write software is getting harder.

Recently I’ve found myself drawn to hardware. It’s new to me (challenging!), making a physical object still takes real effort (legibility!), and LLMs make it more approachable (convenient!).

I’ve heard similar things from several others. My friend Brian Moore put it this way:

And I’ve noticed my preferences changing outside of the tech world as well. I’ve been drawn to weightlifting and to playing my piano, both of which are just as hard now as they were 5 years ago. And like my friend Kelin, I’ve found sports refreshing:

Societal preferences changing as technology changes is nothing new.

Re-adding friction

I attended a liberal arts college in 2010 and enjoyed reading Pitchfork. So I have a record collection.

The reappearance of record collecting in the 2010s is a funny phenomenon. But viewed through the lens of legibility of effort it makes a lot of sense.

The proliferation of digital music and streaming made having a music collection easy and frictionless. And so a subculture evolved to re-add that friction.

And in small ways I think you see the same things happening now.

I’ve seen people joke about adding typos to emails to prove that they wrote them. MS Paint-style image macros read as more human than detailed, funny images (the image could be AI slop). Websites that look intentionally bad are more interesting than websites that look beautifully bland.

Conveying our humanity in the face of LLMs is a hard, new, interesting problem. I’m interested to see what we produce as a result.

Wrapping Up

I don’t have a solution here. I’ve been thinking a lot about legibility of effort and I wanted to write those thoughts down in case they resonated with other people.

So I hope you enjoyed this post. I worked hard on it. I hope you can tell.