Paul N. Johnson

Bet on people who think the future will be better and that they might make it so. Don’t expect much from people who think the future will be worse unless others do something about it.

My hunch is that there's something about human taste/craft/agency that is entirely orthogonal to LLMs, such that no scaling or tuning can even make a dent in it. (This is one reason I don't buy "AGI/ASI" hype. Give me a single good sentence first...)

At best, it feels 1) "barely okay" — but never — 2) "just right." The "okay" results often give me a good idea, but... ^ Whatever that mysterious gap between 1—2 is, my intuition is that it has not closed *at all* despite the massive advancements in models over 3 years.

In daily LLM use since 2023, there's a gap I've never managed to bridge: I've never successfully prompted a model to generate a single sentence for a piece of writing I care about that I'm happy with...

I have a very particular set of /skills

@karpathy my take: the gap is "mental maps" — we need to evolve beyond mono-dimensional mental models/and uis to 2d maps that visualize a "field of play" where any human/agent effort can be placed in meaningful, coherent space. like a telestrator, but for anything.

Human orgs are not legible, the CEO can’t see/feel/zoom in on any activity in their company, with real time stats etc. I have no doubt that it will be possible to control orgs on mobile, with voice etc., but with this level of legibility will that be optimal? Not in principle and asymptotically but in practice and for at least the next round of play.

Expectation: the age of the IDE is over Reality: we’re going to need a bigger IDE (imo). It just looks very different because humans now move upwards and program at a higher level - the basic unit of interest is not one file but one agent. It’s still programming.

[Narrator: little did he know, the shift was back to Command Line Interfaces…]

Had a similar experience: moved my 40+ page/cms personal website off webflow to a vibe coded free service a few weekends ago + added some nice features in about 5 hrs over a weekend and mostly finished it while on the treadmill from my phone. A wild new world we’ve entered no doubt!

I vibe coded an entire headless Wordpress theme from scratch (Created a mockup new TGC website). I know almost zero code and it works beautifully. Instant search, responsive, beautiful typography. Have spent about 4-5 hours total on it. All using Claude Code. Stunning.

@paulg A Timeless Way of Building by Christopher Alexander is a great example of this, and also inadvertently contributed to the development of object-oriented programming

It may be that the best philosophy books are not books explicitly about philosophy, but books on other topics written by such consummate experts that they can take their field's ideas all the way to the highest levels of abstraction.

Good writing is usually about more than it seems to be. Robert Morris introduced me to this idea when he pointed out that K&R was not just about C but about how to program. Since then I've noticed example after example of this phenomenon.

feels like @cursor_ai is beginning to replace @RoamResearch & chatgpt for me, I fear.

Here's the long-read: pathwright.com/mapping-the-ne… (Thanks to @TheJustinHall for the images and interactions!)

I think the next era of the Internet is headed in a different direction than many people think. Made a map of it and would be curious to hear your thoughts.

The breadcrumbs under the table are still more life-giving than analyzing the table.

The popularity of AI slop results from novelty overwhelming our internal quality standards. It's as true for a tweet as it is for the flashiest AI startups. Novelty dominates bubble eras, but patient craftsmanship applied in novel ways succeeds in the long run.

@wmdmark (You can join for free here: paths.pathwright.com/library/pathca…)

My twin, @wmdmark, and I think the Internet is entering a new — yet strangely  familiar — 3rd era. If you'd like a clearer big-picture map of how we see design, learning, & teaching evolving, we'd love for you to join the conversation this Thursday at 12 pm EDT.

Hilma af Klint > Picasso

@MatjazLeonardis "Having a purpose" is more like "having an appetite" or "moving in a direction." I may crave various foods, but that indicates my appetite is flexible, not that it doesn't exist. I can't move without a direction (even if I feel directionless) or go this way then that.

“Having a purpose” is just a useful fiction as is revealed by the fact people have inconsistent purposes. You have no purpose.

It seems that Apple or a carrier could use existing voice-AI to "pick up" spam calls and intentionally waste their time with convincing responses (but of course route to the human if it's not spam). A feature like this would kill telemarketing for good.

The phone part of my phone has become almost entirely useless over the past year or so. Anyone else get spam called all day like this?

Meditation for today: Marcus Aurelius’s Meditations is the only *actual* self-help book ever written.