elliotto

Bike nerd + AI nerd here. The author's observation that all bicycle images face right almost certainly has to do with the convention to photograph a bicycle from the right. From the right, you see the drivetrain - this is good for aesthetics, but also for marketing - the drivetrain is branded and labelled and a buyer will want to know what model it is.

There is a bunch of guidance online on how to photograph bikes, and every sales image of a bike will be from the right. You can anecdotally observe this by google imaging 'bicycle for sale'.

Hi, Co2 levels like that are severely high and will cause you to have a lot of issues. I had some issues with poor ventilation and headaches in my apartment, and my solution was to run the bathroom fan all the time - this gave me enough ventilation to feel much better. You're basically suffocating at 3000 ppm.

It's important to measure this somehow - I do this with a $100 Co2 sensor and display I got off amazon, but you seem to already have these sensors available.

Bravo Garry, net worth $x00m, having the integrity to go after public school teachers.

This rule does not apply in the main trekking areas that everyone goes to.

In the areas that few people go to, it isn't actually enforced.

The Nepalese government just introduces this rule every few years in response to a missing trekker in the guise of safety.

I've been to Nepal a bunch of times and I usually recommend just passing quickly through KTM to get to where you are going. The dust can be terrible and it is loud and polluted - the opposite reason to why most people generally want to go to Nepal. Better to spend more time in the mountains or Pokhara

I worked on a startup experimenting with using gemini-2.0-flash (the year old model) using its full 1m context window to query technical documents. We found it to be extremely successful at needle-in-a-haystack type problems.

As we migrated to newer models (gemini-3.0 and the o4-mini models) we again found it performed even better with x00k tokens. Our system prompt grew to about 20k tokens and the bots were able to handle it perfectly. Our issue became time to first token with large context, rather than the bot quality.

The ultra large 1m+ llama models were reported to be ineffective at >1m context. But at this point, it becomes so cost prohibitive to use anyway.

I am continuing to have success using Cursor's Auto model, and GPT-5.1 with extremely long conversations. I use different chats for different problems moreso for my own compartmentalisation of thoughts, rather than as a necessity for the bot.

It used to be that the bots had a short context window, and they struggled with getting confused by past context, so it was much better to make a new chat every now and then to keep the thread on track.

The opposite is true now. The context windows are enormous, and the bots are able to stay on task extremely well. They're able to utilize any previous context you've provided as part of the conversation for the new task, which improves their performance.

The new pattern I am using is a master chat that I only ever change if I am doing something entirely different

Yes, this is AIMD and it's well formalised and understood.

This is the TCP backoff algorithm, specifically the slow start to find the optimal bandwidth. In your analogy, it would find the optimal amount that a person is willing to reciprocate.

Not only does this algorithm exist, but we're using it to communicate right now!

https://en.wikipedia.org/wiki/TCP_congestion_control

I wrote this comment in a response to his second chapter, where he presents criticism of the political role of the company as cynical, and then later where he presents a perspective on some tech company anti-union behaviour being conspiratorial.

I definitely took an uncharitable reading, but man am I tired of being told big tech is neutral. I will continue to be cynical and I will continue to gnash my teeth at anyone who tells me otherwise.

I think there are reasonable things to expect from someone's morality calculus. Leaving the country you were born in for moral reasons is a complex and life changing undertaking and beyond reasonable expectation for anyone not extremely politically motivated, let alone resourced enough to do so. Not working for a company whose moral values you disagree with (when you have an extremely lucrative skillset) is a smaller and more reasonable ask.

I'm also not really asking that people leave these roles - everyone has their own path to take. Just that they don't make posts dismissing criticism of these structures as silly cynicism. Or else they will have to contend with me writing a comment disgreeing with them.

I don't think he's morally bankrupt. I am disagreeing with his attempt to handwave away a moral analysis of these organizations as 'cynicism'. I think these analyses are really important.

I don't live in the US. But if I did, and I was capable enough to be a successful software engineer, I would try to work for an organisation that was not implicated in abhorrent behaviour. If I was to work for one, I would not attempt to dismiss criticisms of it as cynicism.

The author seems like a nice guy, but perhaps a bit naive regarding the efforts big tech companies go to to crush employees (https://en.wikipedia.org/wiki/High-Tech_Employee_Antitrust_L...). They appear to be a staff level engineer at a big tech company - I don't know how much money they make, but I suspect it's an ungodly amount.

The organisation he works for is implicated in surveillance, monopoly exploitation, and current military action involving particularly unpopular wars. No one forced him into this role - he could have made less money elsewhere but decided not to. He has decided to be a cog in a larger, poorly functioning machine, and is handsomely rewarded for it. This sacrifice is, for many, a worthwhile trade.

If you don't want to engage with the moral ramifications of your profession, you are generally socially allowed to do so, provided the profession is above board. Unfortunately, you cannot then write a post trying to defend your position, saying that what I do is good, actually, meanwhile cashing your high 6-7 figure check. This is incoherent.

It is financially profitable to be a political actor within a decaying monopolist apparatus, but I don't need to accept that it's also a pathway to a well-lived life.

The author postulates a few ideas about manners and courtesy, and starts to recognise that business transactions (employment relationships) don't actually care about these things, even though the human beings who populate these systems hold these values.

The nash equilibrium in a buyer-seller market like the employer-employee relationship is for both sides to defect. Humans don't behave optimally, because they aren't pure rational creatures, they are imbued with some socialisation and cultural memory. So humans try to treat with these organisations as though they are other humans, and will respond to good-will with good-will, but this is not rewarded, and ultimately they change their behaviour in response to a poor environment.

Capital does behave short term optimally. Optimal economic behaviour is to betray the person opposite you, and violate and exploit the commons until the commons collapses entirely, like what we see today. At some points in the past, capital has been subdued by a human operator who will apply courtesy and social norms to prevent these ugly actions, but capital has now become too intelligent to bother with this, and the result is a sequence of increasingly insane and inhuman processes, such as what we see here with the job market.

Many countries have financial incentives provided to its citizens to have children. Requiring half of a citizens median salary to be given to a faceless middleman to provide this service seems untenable. I cannot imagine a society that does this would be able to survive.

This is a good article that is critical of narratives around behaviour within organisations. I particularly enjoyed his criticism of the 'morality tale'.

The author then postulates some guidance for how to survive in organisations more generally, working above these strange social structures largely unique to silicon valley. It wasn't the purpose of the article, but I wish he was a bit more critical of these structures in general.

Under this definition, could any tool at all be considered to produce more value?

I completely agree with this.

I think the writing style the LLM produces is an artistic decision made by committee prioritizing for inoffensiveness - what a coincidence that it comes out sounding like LinkedIn slop.

I don't really see any innate reason an LLM couldn't write well - it's an active decision by its creators to tell it not to.

I don't quite understand this post. Wouldn't rolling out fiber infrastructure early have been proved to be visionary and made the UK a serious technical force?

In Australia, we went through a similar journey where fiber to everyone's home was planned and then politically destroyed. Except this happened in 2010 and has been a significant factor in our inability to retain a technical edge.

Beats giving all the money to the person who says the word 'blockchain' the most.

The title of the post is 'the kind of company I want to be a part of'. This presents a more abstract philosophical question of what one should do and how one should be. I clicked the article expecting a piece about social utility, intellectual stimulation, or the role of firms in an increasingly complex moral environment.

Instead the author posited a point about pluralizing nouns.

This is the Californian ideology - do not engage with fucking anything at all, because we're all getting rich off pluralizing nouns.

We are working on a similar agent for general AI analytics at https://www.truestate.io/

We have a similar experience where it's shocking how much users prefer the chat interface.

Hi HN, after reading through the comments sections on blog posts about RAG, I was finding a lot of differing views. Many people have had an entirely unique experience trying to implement a RAG system, due to the nature of their data, the scale and their goals. The underlying tech also changes so quickly that new methods show up every day, and old methods become outdated quickly.

I put together this blog post trying to capture some of the experiences we had building a RAG system at TrueState, and trying to put together a centralised, up-to-date review of the concepts.

If anyone has any feedback, suggestions, or wants to flame me for something I've gotten wrong, please let me know!

Technofeudalism by Varoufakis is about this N% cooperation. Growing wealth concentration means this collusion becomes possible with smaller and smaller N% cooperating. If it's game theory optimal to cooperate I have no doubt Thiel will be releasing the robo hounds the minute he can.

The Grapes of Wrath by Steinbeck is largely about this concept. If these social transitions were as rosy as you paint them I think the book would be called Grapes of Pleasant Improvements in Living Standards.

GPT-5 can't handle 2 things: an esoteric quine or an aquatic equine

Very neat! A lot of small LLM's have a similar failure mode where they get stuck and repeat a token / get stuck in a 2-3 token loop until they hit the max message size cutoff. Very ironic that it's about a quine.

I tried it with thinking mode and it seems like it spiraled wildly internally, then did a web search and worked it out.

https://chatgpt.com/share/68e3674f-c220-800f-888c-81760e161d...

https://chatgpt.com/share/68e3674f-c220-800f-888c-81760e161d...

With thinking it spirals internally, runs a google search and then works it out.

https://chatgpt.com/share/68e366b2-0fdc-800f-9bf3-86974703b6...

GPT-5 Instant (no thinking) spirals wildly. Poor bot