Scott Smith

[Scott Smith @sakohts]

@perplexity_ai I love your app, but you have to allow users to turn off those “engagement questions” at the end of the answers it gives. I’m sure they are useful for shy people or idiots. For the rest of us, they are massively irritating.

[Andrej Karpathy @karpathy]

It is hard to communicate how much programming has changed due to AI in the last 2 months: not gradually and over time in the "progress as usual" way, but specifically this last December. There are a number of asterisks but imo coding agents basically didn’t work before December and basically work since - the models have significantly higher quality, long-term coherence and tenacity and they can power through large and long tasks, well past enough that it is extremely disruptive to the default programming workflow.

Just to give an example, over the weekend I was building a local video analysis dashboard for the cameras of my home so I wrote: “Here is the local IP and username/password of my DGX Spark. Log in, set up ssh keys, set up vLLM, download and bench Qwen3-VL, set up a server endpoint to inference videos, a basic web ui dashboard, test everything, set it up with systemd, record memory notes for yourself and write up a markdown report for me”. The agent went off for ~30 minutes, ran into multiple issues, researched solutions online, resolved them one by one, wrote the code, tested it, debugged it, set up the services, and came back with the report and it was just done. I didn’t touch anything. All of this could easily have been a weekend project just 3 months ago but today it’s something you kick off and forget about for 30 minutes.

As a result, programming is becoming unrecognizable. You’re not typing computer code into an editor like the way things were since computers were invented, that era is over. You're spinning up AI agents, giving them tasks *in English* and managing and reviewing their work in parallel. The biggest prize is in figuring out how you can keep ascending the layers of abstraction to set up long-running orchestrator Claws with all of the right tools, memory and instructions that productively manage multiple parallel Code instances for you. The leverage achievable via top tier "agentic engineering" feels very high right now.

It’s not perfect, it needs high-level direction, judgement, taste, oversight, iteration and hints and ideas. It works a lot better in some scenarios than others (e.g. especially for tasks that are well-specified and where you can verify/test functionality). The key is to build intuition to decompose the task just right to hand off the parts that work and help out around the edges. But imo, this is nowhere near "business as usual" time in software.

[Andrej Karpathy @karpathy]

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.

[Scott Smith @sakohts]

@karpathy I find Warp like Claude Code but way better experience. Am I crazy? Curious if others have tired both and think Claude Code is better now.

[Ethan Siegel @StartsWithABang]

Today is the day!

The Grand Cosmic Story, my newest book (with Jon Lomberg, Mark Garlick, and Will Lidwell), from National Geographic, is released!

Get the history of the whole Universe, one page and 100 Myr at a time, today!
https://t.co/o9ieGV9fpD

https://t.co/J3ayHTV7HC

[Scott Smith @sakohts]

@RBReich Can you clarify? The definition of being rich is to own expensive things. However good or bad the people or the companies, that the richest people are rich because they own the most influential companies is not surprising?

[Scott Smith @sakohts]

@cremieuxrecueil Yeah, the murder rate in cities is 2X per capita what it is in rural areas. And since the population is 5X, that means 10X to the murder rate. We have not figured out how to manage cities. Sadly.

[Scott Smith @sakohts]

@cremieuxrecueil The graph with countries and causes of illness doesn’t include the US. How does it show the difference between those countries and the US?

[Scott Smith @sakohts]

@pmarca Animal populations tend to expand to
fill a space, but predators limit their own ranks to reduce competition. Curious if humans are doing this now. We want less competition from other humans for space, status. Wealth consolidation.

[Scott Smith @sakohts]

@wizardingnews @jk_rowling The other option was …rewrite the script to have the character be pregnant? I’m glad we have laws to protect people bringing on the next generation. But it seems like we get outraged w/o asking if we would want to be on both sides of this law and would think it was fair.

[Scott Smith @sakohts]

@JamieBonkiewicz Causally used today it means a social safety net. But for Marx it meant the “nationalization of businesses”. No private biz ownership. Everyone works for the government. Which will work more fairly than greedy private owners, in his theory. Few advocates today want this.

[Scott Smith @sakohts]

@donwinslow There are so many opinions in this thread, but very little actual info on what’s happening. What you see is based on what story you believe before you see it. Amplify prejudice on multiple sides. Fail to inform. This is how media is weaponized to chip away civic discourse?

[Scott Smith @sakohts]

@rubenhassid If you use https://t.co/lO56R0AHKg instead of Apple’s lagging system, you can have exceptionally nuanced reason-based explorations on novel topics. It will correct your reasoning errors. Get math right. Remain stalwart when you attempt linguistic tricks. Not AGI, but…

[Scott Smith @sakohts]

@tcdelfin @BrightInsight6 @joerogan History is too useful to be burdened by facts? The central drive for humans is to engage in motivated thinking for fun and profit?

[Scott Smith @sakohts]

@grok @swisslaw1 @elonmusk @grok Please add the numbers for the Biden administration?

[Kanjun 🐙 @kanjun]

I treat anxiety as a trailhead to parts that need something, and give those parts compassion. This updates the part, and the anxiety disappears. This means doing harder things = deeper healing!

I only become aware of these scared parts because hard things are scary; without hard things I don’t know if I’d ever notice. This makes me like scary things more, and prevents getting into ease ruts.

There is a “breaking point” of doing hard things though, which we call “burnout”. Noticing and processing parts is a good way to avoid / heal from burnout.

[Scott Smith @sakohts]

@Kevin722kjay @musk_news13 This is a great answer because the others with 40 and 96 explain the pattern, but they ignore the = sign. Yours is the only one that respects the = being true in some context.

[Scott Smith @sakohts]

@elonmusk Hard to imagine a lease with more skin in the war game than Zelensky.

[Scott Smith @sakohts]

@KevinSVanHorn @paulg So you think they quit or had to be fired because they were corrupt. Not because they were asked to break the law?

[Scott Smith @sakohts]

@0xmetaschool Love these models. But they aren’t “open source” unless we have the code that builds them and can iterate on it. It is nice that they are downloadable at no charge.