martythemaniak

I'm dealing with this now because I'm working on something that is very far outside my familiar domains so even if I had the time to hand-code it, I wouldn't be able to do it.

My answer so far is to dedicate a substantial token budget to continual polishing, refactoring, debugging, cleanup and explanation. Some practical steps:

1. output has to be perfect. Anything that looks even slightly off get's logged and a session is started to investigate, this usually leads to a bunch of other issues.

2. Ask the AI if the code works the way you think it does -- you should still have some model of how the software works in your head

3. Ask for documentation and explanation around interfaces and make sure you understand the interfaces. It's ok if something's a black box, as long as you have a good understanding of what goes in and what goes out.

4. continually ask for performance improvements and assessments.

5. In addition to agents and readmes, maintain an ADR so all decisions and past constraints are readily available to be enforced on new code.

6. Put in as many hard rules as you can about things that need your explicit approval.

Fundamentally, my current job as a "AI-native builder" (cringe) is way more similar to my previous position as EM than anything resembling an IC.

Is there anything noteworthy about Minister Michel?

Not crazy at all. Tesla gave up being a car company a few years ago, but there's lot of inertia and the 3 and Y are still competitive products in the marketplace.

They're not, it's an illusion.

My older sibling went to HS in Eastern Europe and kept the textbooks, because my family had this same reverence towards math over there. I went to HS in Canada and in grade 12 when I was doing Calculus I had a look at the old textbooks and they were the same thing. Different styles certainly, but the problem sets were basically interchange. In Algebra and Geometry, the Canadian books seemed to have much more emphasis on algebra, whereas the eastern european ones had much more emphasis on geometry.

I think the illusion comes from the different rates at which you're expected to pick things up. Grade 8 Canadian math was much easier, so I think the Canadian curriculum was backloaded - easier in the early grades and much harder in the later ones, whereas the eastern european one seemed much more even.

I'll repost a comment I made a while back describing the process I use, which I think balances AI in keeping your voice.

If you care about your voice, don't let LLMs write your words. But that doesn't mean you can't use AI to think, critique and draft lots of words for you. It depends on what purpose you're writing it for. If you're writing an impersonal document, like a design document, briefing, etc then who cares. In many cases (scientific papers, legal documents) you already have to write them in a voice that is not your own. Go ahead and write these in AI. But if you're trying to say something more personal then the words should be your own, AI will always try to 'smooth' out your voice, and if you care about it, you gotta write it yourself.

Now, how do you use AI effectively and still retain your voice? Here's one technique that works well: start with a voice memo, just record yourself maybe during a walk, and talk about a subject you want, free form, skip around jump sentences, just get it all out of your brain. Then open up a chat, add the recording or transcript, clearly state your intent in one sentence and ask the AI to consider your thoughts, your intent and ask clarifying questions. Like, what does the AI not understand about how your thoughts support the clearly stated intent of what you want to say. That'll produce a first draft, which will be bad. Then tell the AI all the things that don't make sense to you, that you don't like, just comment on the whole doc, get a second draft. Ask the AI if it has more questions for you, you can use live chat to make this conversation go smoother as well, when the AI is asking you questions, you can talk freely by voice. Repeat this one or two more times, and a much finer draft will take shape that is closer to what you want to say. During this drafting state, the AI will always try to smooth or average out your ideas, so it is important to keep pointing out all the ways in which it is wrong.

This process will help you with all the thinking involved being more up-front. Once you're read and critiqued several drafts, all your ideas will be much more clear. Then, sit down and write your own words from scratch, they will come much easier after all your thoughts have been exercised during the drafting process.

So why did this get [flagged], is Gemini moderating HN now?

There's no reason to think it didn't.

The media is mostly MAGA owned. Here's a short list:

- TikTok is controlled by Larry Ellison

- CBS news is controlled by Ellison's son

- The Ellisons will soon own CNN as well

- Twitter is controlled by Musk

- Facebook, Instagram and Threads are under Zuck

- WaPo and Amazon Prime by Bezos

- Old time favourites like Fox and WSJ owned by Murdoch

The owners of this media are also very explicitly involved in pro-Trump editorial content. In short, the playing field is wildly lopsided and unreliable and no tech (decentralized social media, lol) solution will work. Ultimately people will have to elect politicians to explicitly break up and clamp down on this hive of scum and villainy, but people aren't there yet, mentally. Such a solution seems wildly inappropriate today, but my guess is that another 10 years of the last two years will change a lot of minds.

A fun idea I've been playing around with and testing on folks: We need a neo-temperance movement.

The first reaction is a scoff and a laugh of course - wasn't the original temperance movement an utter failure, prohibition didn't work etc. But no, the movement was far wider and successful in a multitude of ways. And a new temperance movement isn't about blindly copying something old, but situating yourself around the turn of the 20th century, understanding the multitude of interconnected problems people faced, the solutions they came up with and which ones worked and didn't.

America at the turn of the 20th century was a nasty place - there was wild corruption, politics was run by mafia-like patronage networks, wealth was extremely concentrated, rich people were unaccountable, working people suffered under low-paying, dangerous jobs, women were excluded from public life, and alcohol was merely the symptom of a very sick society, and yes indeed it was a massive problem, the way alcoholism is a massive problem in Russia today.

The original temperance movement was bound up tightly with the Suffarage movement (giving women a voice), with Taylorism (running organizations rationally), with income tax (a fairer less corruptible source of income for the federal government than tariffs and excise taxes).

Today's problems are thematically the same - massive concentration of wealth is making an unaccountable elite wielding power against the wishes of the public, creating dangerously addictive digital products, massive increase in gambling, restricting the rights of poeple to own and repair products, working to disenfranchise people, working to roll back women's participation in public life, etc etc. So sex, alcohol and drugs are not an issue today, but gambling, addiction, disenfranchisement, exploitative business practices, mind-boggling amounts of corruption - they're all back in the same form and for the same reasons, and we'll need the same type of movement to fight against them.

What the progressives of the early 20th century did was not abolish capitalism, but tame it. We don't need to go on a jihad against AI, but we do need to yoke it and disempower the current slate of people who own it and direct it.

Do this exercise: Take your company's market cap and divide it by the total number of lines of code in the company's repo. You'll discover that don't need that much code to create value. For Google (somewhat publicly somewhat known), this is something like $2000 per line of code. Your company is probably not too different, certainly within an order of magnitude.

So let's assume 1 line of correct code is approximately 10 tokens and $1000 of value. If you could theoretically write the correct code, $10 per 1M output tokens should get you $100K worth of value, or 10000x return! Obviously, thought experiment, very theoretical limit no one can achieve, doesn't include thinking tokens, etc, but it's worth thinking about how much your code is actually moving the needle.

Maybe I'm doing things wrong, but I don't need that much code and I don't spend that much time producing code (maybe 1000 lines in a productive day), because the bulk of my time is spent researching, testing, evaluating, thinking, etc to find out how to write only very valuable code. 1000 lines is laughable compared to what tokenmaxxers do, but over the course of a month that's a 200-300K codebase and anyone who has worked on something of that size knows it can support a sizeable business.

Sure will, but there's plenty of private gain for the right people.

No pain, no gain.

Pretty insane that this isn't available for Pro subscribers on Antigravity, like not even a taste. Antigravity comes with a sample of Opus 5.5 too!

Of course (the watermark is in the lower right), it's just a visualization. I'm spending my money buying parts for the prototype instead of professional renders.

I'm a robotics noob currently building a prototype of a snow sweeping robot (www.frost-e.com) using my own design and I'm encouraged to see similar work coming from real places. Namely:

- Zero shot control using a visual harness and action primitives. I've seen this from a few other places as well (https://tml.stanford.edu/homebody/) and it's essentially having a generalist LLM command a robot via a set of predefined actions. Many robots and many tasks can probably be done with this approach, since not everything needs super intricate 6DoF arms or dexterity

- simulation bring-ups: I have a pipeline that takes in satellite imagery and reconstructs the lots and houses in a 3D environment and snows on it so that my virtual robot can learn to clean. With these latest models, you can probably create a higher fidelity digital twin more directly.

- self-improvement: These models (Argon, Astra, etc) can't operate the robot itself, but would be able to create data to distill into smaller models you can actually utilize.

Assuming the expected thing happens and Trump gets shellacked in the midterms, do Americans think this will have some impact (ie reverse or accelerate) the current trend of cuts to science, cracking down on foreign students, cuts to NIH, HHS etc? Those being big recent change in this area.

As I said in my original comment:

> The problem is that Americans are very far from realizing that they have a problem

So I understand and I certainly agree that you're with the majority of Americans thinking everything's basically fine.

Totes for reals, yes, because I understand the difference between de jure and de facto.

The failure mode you're observing in the US is that one partisan hack can write a memo in the correct linguistic style, submit it to a council with a majority of other partisan hacks, and like magic anything can become legal or illegal, because everything was done in the correct linguistic style. Actual practice does not matter. There are lots of examples of this over the last two years, perhaps the case against the 14th amendment is a good starting point to see how this works.

Sure there is, many countries have successfully thrown off personalist authoritarian regimes and transitioned to being inclusive democracies with a strong rule of law. The problem is that Americans are very far from realizing that they have a problem, the magnitude of their problem, and the time and effort it'll take to fix this problem. It's a generational project that starts with throwing most of the current cabinet in jail, however this view is very far from the mainstream and things will keep getting worse before more people come to the same conclusion.

Nice, it's got real Interdimensional Cable vibes.

They got nothing on American Elites - the current regime is packed to the gills with TV personalities who are incapable of actually doing their nominal jobs, yet they've all done extremely well for themselves.

This article really should not be surprising - the elite go to school to learn to rule - to socialize and develop networks of trust. Doing stuff is for the class below (PMC in today's terms).

I hosted a party for some friends and their families yesterday and this also came up, particularly in regards to Roblox. It is possible to tune Roblox and make it safe for your kid, but is is a lot of work.

Fundamentally, tech companies are user-hostile because their growth, revenue, stock price, (and for some of them now, political power) depend on a shitty experience. Every tech company wants to automate other companies's products, but will use every tool available (including legal ones) to stop you, the user, from automating and controlling their product.

The examples are endless: Roblox creates a kids casino, but if you as the parent want a safer, saner experience for your kid you have to navigate endless settings and there's still tons of shit that falls through the cracks, so you have to manually oversee what they do and manually ban bad games. However, if you try to use automate this process for yourself, there's no API, or you'll get banned for screen scraping or atypical behaviour that triggers some automated system. Facebook aggregates other websites, becomes a centre of social life, then they'll shove endless slop at you, but if you try to use your own app or agent to create your own experience so you're not exposed to that, you get banned.

A while back I vibe-coded a little parental control for my kid's YouTube because I wanted to give him autonomy to choose his own stuff, but also make sure he's not watching crap. I took his YouTube history then had Gemini automatically watch, rank, categorize and flag things for me (https://github.com/drashkov/slopstopper), and the analysis was pretty great. But taking actual action (banning channels, automated runs) is a major pain, because YouTube fundamentally does not want me to be in control of my kids's experience, they guard their turf jealously.

IMHO, the tech arms race is doomed, things will keep getting worse until enough people are pissed off to demand political changes.

The underpants gnomes was supposed to be a joke.

Adding my anecdote, because it amused me: I finished wiring up the compute/sensor box for my robot, ssh'd in and told agy "I have a Livox Mid 360 Lidar connected to this Jetson orin nano, setup a full environment with docker, cuda, ros2, foxglove and get it all working so I can see the lidar output". It did all the local config for the lidar, setup docker and the ROS2 environment, then told me "open up this url in foxglove" and sure enough everything worked. Whole thing used up 6% of my weekly limit.

Well, if I share the exact same expectation/prediction as Musk, but we appear on opposite sides of the chart, is the chart really useful? What's it really trying to say?

Doom or Bloom for Whom?

Say that I believe that AI is an ordinary technology (in the same way as the internet or transistors) that will be used to concentrate massive amounts of wealth and power into a small set of unaccountable oligarchs and their handmaidens while everyone else gets endless enshittification.

Pretty great outcome if you're one of the lucky few. Why wouldn't Musk or Andreesen think that's "bloom"? The exact same outcome OTOH seems pretty doom to me (a random dude)

This is incredibly obvious and believable. The reason why American elites, particularly the SV executive class, went so hard for Trump is because he gives them complete lack of accountability. Whether it is sexual predation, financial regulation, crypto regulation, anti-gambling, it is all the same. Anybody putting any kind of rules in place for them is an affront to their 'freedom'.

This is a much more in-depth article that grew out of that bluesky thread.

Sounds like malicious compliance, they should get a fine for that as well.

Why would you not take him seriously?