This is really slick. I remember when I used to have the F1TV version pulled up on my laptop during sessions - eventually I decided it wasn't that helpful. The radio and race control panels on this are awesome.
Definitely giving it a test run during the race tomorrow!
Really, really love the interface and the way you have the calendar, FIA docs, etc available too. Shout out for the delay feature -- I was able to sync it perfectly with my TV stream!
The big problem I had was desync. Every 5-10 minutes, random parts of the interface would stop updating. Sometimes the whole page would blank out. I was able to fix it once by loading one of the other panels (like the FIA docs) and then going back to the live data. Eventually it broke and I wasn't able to fix it (around lap 18 or so). Thinking maybe sockets aren't reconnecting when their connection drops from the client side? And it seemed like there was some issue that prevented it from fixing with a simple refresh.
That's right. I'm truly sorry. This race had three times the traffic of a normal race and twice the traffic of the previous day's qualifying, so i had to urgently upscale our server instances, and i confirmed that several instances experienced issues..
I plan to implement stabilization measures throughout the live timing server code during this summer break.
I sincerely apologize for any inconvenience caused during this important race.
The best way to make a happy, healthy person into an unhappy, unhealthy person is to keep them lonely and keep them still. It should come as no surprise that the inverse also tends to hold true.
On a tangent, I think that's part of why volunteering can be so rewarding.
There were no restrictions on running or biking in the pandemic. Quite the opposite. Because indoor entertainment venues like bars weren't available, I saw way more people embracing the outdoors.
I am Canadian. Running, biking etc outdoor activities is impossible in our winters for 5-6 months of the year. Since gyms were closed, we also didn't have access to weight training which is also very essential.
Canadian here. This may be true for some (many) parts of Canada but certainly not all of Canada. For example, with proper gear you can run outdoors in Vancouver year-round. When I lived in the GTA, I routinely ran outdoors through late December and almost always resumed by mid-March.
Point being: dressing appropriately can extend your running season dramatically.
Even in your example, there's a 4 month period where it's impossible. Snow + ice makes it dangerous to be running outside during winters. I have friends who run marathon and one has had knee surgeries and therefore afraid to run in even a little bit of ice.
This is an interesting comment because I've lived a lot of sides of this.
At my first job (where I kinda 'weaseled' my way into doing software vs my job title) it was an incredibly collaborative experience. It started with finding ways to make tools that helped my colleagues do monotonous tasks faster. Which then evolved into fun dialogue. "Hey can you make a button to do X" and we'd get to talk about it, I'd hack the feature together, hit publish and wait for the team to give feedback. "Oh I got this error" I'd get up and walk over. It wasn't perfect but I was never lonely and only as still as I wanted to be.
At my second gig, It started a little lonely but thankfully the culture was just laid back enough I got to socialize (thankfully it was a shop full of fun and interesting people!).
Third gig, Uggh it was very 'heads down' for most of my time there, nobody liked small talk except the conspiracy theory guy. I learned a lot about what I did and didn't like in company culture there. It did get a little better before I left...
Fourth gig was a dream. It was the second place where I didn't just get to collaborate with my team, but the first place where it was a lot of software engineers. We even had a teams room for nothing but sharing music and it was always heartwarming to see a reply to some obscure tune and someone would reply with something that yes you would absolutely appreciate given what you originally posted. And it was hectic enough that I did get a reputation for being a 'floor runner'.
Fifth Gig... well it was 100% remote. And in fact one day I was so focused on a problem I sat in the wrong position too long and permanently fucked up my left ulnar nerve... But that was such a good group, and Ironically I was able to -take- the lessons from #4 and #2 and turned them into traditions that stuck around even after I left (hell even after they fired everyone, we kept doing the 'game night' for a while...)
Won't say anything about my current place, that's all still a work in progress <_<
The discussion of how the medium affects what you build reminds me a bit of The Beginner's Guide (https://en.wikipedia.org/wiki/The_Beginner's_Guide). If I recall, it had a bit of dialogue about how the game-creation tool made it easy to build straight, square corridors... And so the map had a lot of straight, squared corridors.
I don't really remember much of that game, but for some reason that part stuck with me. It's a good bite-sized insight into creativity (to be aware that constraints, while limiting, will also guide you).
Describes the growing use of surveillance and AI for things like predicting crime, detecting a migrant's region of origin using their dialect, etc. The video in the article has a lot of detail and discusses the potential benefits and drawbacks.
Seems somewhat relevant to the ChatControl discussion on here the other day.
I really miss some of the companies from that era... Red Storm Entertainment, Tom Clancy's vision, comes to mind. The early Ghost Recon and Rainbow Six games had a dedication to immersion that their modern counterparts totally lack.
Ghost Recon (2001) runs perfectly through proton on my linux desktop. I still fire it up from time to time.
At our local lans in the early 2000s, Rainbow Six Rogue Spear was probably the most popular game we played, and still there doesn't seem to be any tactical shooters like that available today, that focuses on realism over cosmetics and arcade-mechanics.
Also makes me think of the countless hours of fun Westwood Studios provided me with in my youth, real shame they didn't get to survive longer :/
I was worried about that a couple of years ago, when there was a lot of hope that deeper reasoning skills and hallucination avoidance would simply arrive as emergent properties of a large enough model.
More recently, it seems like that's not the case. Larger models sometimes even hallucinate more [0]. I think the entire sector is suffering from a Dunning Kruger effect -- making an LLM is difficult, and they managed to get something incredible working in a much shorter timeframe than anyone really expected back in the early 2010s. But that led to overconfidence and hype, and I think there will be a much longer tail in terms of future improvements than the industry would like to admit.
Even the more advanced reasoning models will struggle to play a valid game of chess, much less win one, despite having plenty of chess games in their training data [1]. I think that, combined with the trouble of hallucinations, hints at where the limitations of the technology really are.
Hopefully LLMs will scare society into planning how to handle mass automation of thinking and logic, before a more powerful technology that can really do it arrives.
The RAG technique uses a smaller model and an external knowledge base that's queried based on the prompt. The technique allows small models to outperform far larger ones in terms of hallucinations, at the cost of performance. That is, to eliminate hallucinations, we should alter how the model works, not increase its scale: https://highlearningrate.substack.com/p/solving-hallucinatio....
Pruned models, with fewer parameters, generally have a lower hallucination risk: https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00695.... "Our analysis suggests that pruned models tend to generate summaries that have a greater lexical overlap with the source document, offering a possible explanation for the lower hallucination risk."
At the same time, all of this should be contrasted with the "Bitter Lesson" (https://www.cs.utexas.edu/~eunsol/courses/data/bitter_lesson...). IMO, making a larger LLMs does indeed produce a generally superior LLM. It produces more trained responses to a wider set of inputs. However, it does not change that it's an LLM, so fundamental traits of LLMs - like hallucinations - remain.
There are some things that you still can't do with LLMs. For example, if you tried to learn chess by having the LLM play against you, you'd quickly find that it isn't able to track a series of moves for very long (usually 5-10 turns; the longest I've seen it last was 18) before it starts making illegal choices. It also generally accepts invalid moves from your side, so you'll never be corrected if you're wrong about how to use a certain piece.
Because it can't actually model these complex problems, it really requires awareness from the user regarding what questions should and shouldn't be asked. An LLM can probably tell you how a knight moves, or how to respond to the London System. It probably can't play a full game of chess with you, and will virtually never be able to advise you on the best move given the state of the board. It probably can give you information about big companies that are well-covered in its training data. It probably can't give you good information about most sub-$1b public companies. But, if you ask, it will give a confident answer.
They're a minefield for most people and use cases, because people aren't aware of how wrong they can be, and the errors take effort and knowledge to notice. It's like walking on a glacier and hoping your next step doesn't plunge through the snow and into a deep, hidden crevasse.
LLMs playing chess isn't a big deal. You can train a model on chess games and it will play at a decent ELO and very rarely make illegal moves(i.e 99.8% legal move rate). There are a few such models around. I think post training messes with chess ability and Open ai et al just don't really care about that. But LLMs can play chess just fine.
Jeez, that arxiv paper invalidates my assumption that it can't model the game. Great read. Thank you for sharing.
Insane that the model actually does seem to internalize a representation of the state of the board -- rather than just hitting training data with similar move sequences.
...Makes me wish I could get back into a research lab. Been a while since I've stuck to reading a whole paper out of legitimate interest.
(Edit) At the same time, it's still worth noting the accuracy errors and the potential for illegal moves. That's still enough to prevent LLMs from being applied to problem domains with severe consequences, like banking, security, medicine, law, etc.
> people aren't aware of how wrong they can be, and the errors take effort and knowledge to notice.
I have friends who are highly educated professionals (PhDs, MDs) who just assume that AI\LLMs make no mistakes.
They were shocked that it's possible for hallucinations to occur. I wonder if there's a halo effect where the perfect grammar, structure, and confidence of LLM output causes some users to assume expertise?
Computers are always touted as deterministic machines. You can't argue with a compiler, or Excel's formula editor.
AI, in all its glory, is seen as an extension of that. A deterministic thing which is meticulously crafted to provide an undisputed truth, and it can't make mistakes because computers are deterministic machines.
The idea of LLMs being networks with weights plus some randomness is both a vague and too complicated abstraction for most people. Also, companies tend to say this part very quietly, so when people read the fine print, they get shocked.
> I wonder if there's a halo effect where the perfect grammar, structure, and confidence of LLM output causes some users to assume expertise?
I think it's just that LLMs are modeling generative probability distributions of sequences of tokens so well that what they actually are nearly infallible at is producing convincing results. Often times the correct result is the most convincing, but other times what seems most convincing to an LLM just happens to also be most convincing to a human regardless of correctness.
> In computer science, the ELIZA effect is a tendency to project human traits — such as experience, semantic comprehension or empathy — onto rudimentary computer programs having a textual interface. ELIZA was a symbolic AI chatbot developed in 1966 by Joseph Weizenbaum and imitating a psychotherapist. Many early users were convinced of ELIZA's intelligence and understanding, despite its basic text-processing approach and the explanations of its limitations.
Its complete bullshit. There is no way anyone ever thought anything was going on in ELIZA. There were people amazed that "someone could program that" but they had no illusions about what it was, it was obvious after 3 responses.
Don't be so sure. It was 1966, and even at a university, few people had any idea what a computer was capable of. Fast forward to 2025...and actually, few people have any idea what a computer is capable of.
My experience, speaking over a scale of decades, is that most people, even very smart and well-educated ones, don't know a damn thing about how computers work and aren't interested in learning. What we're seeing now is just one unfortunate consequence of that.
(To be fair, in many cases, I'm not terribly interested in learning the details of their field.)
If I wasn't familiar with the latest in computer tech, I would also assume LLMs never make mistakes, after hearing such excited praise for them over the last 3 years.
It is only in the last century or so, that statistical methods were invented and applied. It is possible for many people to be very competent at what they are doing and at the same time be totally ignorant of statistics.
There are lies, statistics and goddamn hallucinations.
Have they never used it? Majority of the responses that I can verify are wrong. Sometimes outright nonse, sometimes believable. Be it general knowledge or something where deeper expertise is required.
I worry that the way the models "Speak" to users, will cause users to drop their 'filters' about what to trust and not trust.
We are barely talking modern media literacy, and now we have machines that talk like 'trusted' face to face humans, and can be "tuned" to suggest specific products or use any specific tone the owner/operator of the system wants.
It's super obvious even if you try and use something like agent mode for coding, it starts off well but drifts off more and more. I've even had it try and do totally irrelevant things like indent some code using various Claude models.
My favourite example is something that happens quite often even with Opus, where I ask it to change a piece of code, and it does. Then I ask it to write a test for that code, it dutifully writes one. Next, I tell it to run the test, and of course, the test fails. I ask it to fix the test, it tries, but the test fails again. We repeat this dance a couple of times, and then it seemingly forgets the original request entirely. It decides, "Oh, this test is failing because of that new code you added earlier. Let me fix that by removing the new code." Naturally, now the functionality is gone, so it confidently concludes, "Hey, since that feature isn't there anymore, let me remove the test too!"
Yeah, the chess example is interesting. The best specialised AIs for chess are all clearly better than humans, but our best general AIs are barely able to play legal moves. The ceiling for AI is clearly much higher than current LLMs.
> you'd quickly find that it isn't able to track a series of moves for very long (usually 5-10 turns; the longest I've seen it last was 18)
In chess, previous moves are irrelevant, and LLM aren't good with filtering out irrelevant data [1]. For better performance, you should include only the relevant data in the context window: the current state of then board.
Interesting. I don’t use GPT for code but I have been using it to grade answers to behavioral and system design interview questions, lately. Sometimes it hallucinates, but the gists are usually correct.
I would not use it if it was for something with a strictly correct answer.
I picked up Player Piano, which had been collecting dust on my shelf, yesterday, and quickly found myself having blitzed through it in its entirety by the evening.
The book has many elements and predictions that feel dated, and a lot more that feel like they might've been made yesterday. Highly relevant to the cultural effects of automation. It puts its finger on the nose of the obvious questions that have somehow gone un-asked in current dialogues -- importantly, as the linked article quotes, "What are people for?"
I wonder if the sunk cost fallacy - that usually refers to an abstract cost, like time or money - would truly be the same effect as an aversion to retracing a path in 3D space.
Possible, or even likely, but interesting nonetheless. Towards the end of the article, they describe an interesting other direction of their research that's not so directly correlated with sunk cost:
> More recently, we’ve been examining a related form of hesitation. This time, it’s not in switching paths, but in committing to one at all.
> “While it might seem that having enticing options (e.g., a great apartment one could rent, a fun event one could sign up for) would make commitment easier, we’ve found that it’s often the loss of a great option that finally pushes people to choose. People often hold out for something even better, but the disappearance of a pretty good option inspires some pessimism that encourages people to grab onto what is as good as they can get for now.”
Definitely giving it a test run during the race tomorrow!