It's so sad that we're the ones who have to tell the agent how to improve by extending agent.md or whatever. I constantly have to tell it what I don't like or what can be improved or need to request clarifications or alternative solutions.
This is what's so annoying about it. It's like a child that does the same errors again and again.
But couldn't it adjust itself with the goal of reducing the error bit by bit? Wouldn't this lead to the ultimate agent who can read your mind? That would be awesome.
> It's so sad that we're the ones who have to tell the agent how to improve by extending agent.md or whatever.
Your improvement is someone else's code smell. There's no absolute right or wrong way to write code, and that's coming from someone who definitely thinks there's a right way. But it's my right way.
Anyway, I don't know why you'd expect it to write code the way you like after it's been trained on the whole of the Internet & the the RLHF labelers' preferences and the reward model.
Putting some words in AGENTS.md hardly seems like the most annoying thing.
tip: Add a /fix command that tells it to fix $1 and then update AGENTS.md with the text that'd stop it from making that mistake in the future. Use your nearest LLM to tweak that prompt. It's a good timesaver.
While this may be the end goal, I do think humanity needs to take the trip along with AI to this point.
A mind reading ultimate agent sounds more like a deity, and there are more than enough fables warning one not to create gods because things tend to go bad. Pumping out ASI too quickly will cause massive destabilization and horrific war. Not sure who against really either. Could be us humans against the ASI, could be the rich humans with ASI against us. Anyway about it, it would represent a massive change in the world order.
The problem is the demand for dynamic content in AAA games. Large exterior and interior worlds with dynamic lights, day and night cycle, glass and translucent objects, mirrors, water, fog and smoke. Everything should be interactable and destructable. And everything should be easy to setup by artists.
I would say, the closest we can get are workarounds like radiance cascades. But everything else than raytracing is just an ugly workaround which falls apart in dynamic scenarios. And don't forget that baking times and storing those results, leading to massive game sizes, are a huge negative.
Funnily enough raytracing is also just an approximation to the real world, but at least artists and devs can expect it to work everywhere without hacks (in theory).
Maybe I'm missinformed, but doesn't starlink have to comply with rules made by local authorities? Afaik, when the internet goes down, like it happened in Thailand, Starlink can't be used too, because it always roots traffic to a ground station near the source.
If this would not happen, I would agree that Starlink is the future. But as it is right now, I don't see the point, unless you are living in or travelling to remote places.
I also thought about this a lot. Some things about slow thinking are great. I truly believe that it helped me thrive as a software developer.
But social interactions are awkward. I can't really come up with things to say easily and lots of times I can't respond in ways to keep the conversation going. Only after the fact I get lots of ideas of what I could have said. I'm truly impressed about others who can just come up with interesting or funny things to say on the spot.
I'm a tad older, so I stopped caring about it and just accepted my slow thinking. But I'm sure that I also missed out on a lot of opportunities regarding friendships or work. I still think, that others perceive me as awkward or just not fun and it's hard to just ignore that.
Funnily my wife is completely opposite to me and we have the greatest time.
> I'm truly impressed about others who can just come up with interesting or funny things to say on the spot.
As Winston Churchill once said when asked “what are you doing” –> “Oh just preparing my off-the-cuff remarks for tomorrow”
I’m one of those weirdos who does public speaking sometimes. Even 8 hour workshops. You cannot prepare for an 8 hour speaking engagement. Not really. But you can accumulate a plethora of anecdotes, metaphors, and remarks that you weave into the narrative or in response to questions.
You can build frameworks that are similar to code. Prepared functions/coroutines/objects that you run in appropriate situations. Works pretty well especially in mentoring/teaching/consulting situations. This is also how comedians prep their sets.
The key is that things you say are new to the audience, but not to you. It can be the same metaphor you’ve fine-tuned over dozens of interactions. And the person you’re talking to thinks “Wow that guy is so quick on his feet, how did he come up with that so fast!?”
You can also spot this if you watch talks by popular presenters (Simon Sinek is a good example). You’ll notice the same 2 or 3 core stories getting polished and fine-tuned over years of talks and interviews.
When I was super young, I used to think my dad (who everyone I met seemed to think was extremely funny) had a huge repitoire of hilarious stories, but after a few years I noticed him repeating them and realized he just had a few specific ones that he would re-use with new people, like you mentioned. As someone who tends to be pretty slow to learn how to navigate new social situations, it was eye-opening when I recognized this was something I could do. What's amusing to me at this point is that I'm still not sure he fully realizes that this is something he does sometimes, because he'll still sometimes try to whip out one of the stories when talking to me and then genuinely be surprised when I remind him of some very minor detail about it that he forgot to mention this time he told it.
My dad recently told me a funny story about something that happened to him. Except it actually had happened to me and I had told him about it years before.
I recognize this in my self, so I usually preface the story saying something like "this may or may not have been me, but I'm telling it from the first person". First person stories are funnier. This is part of the "telling a yarn" or "tall tale" tradition I grew up with (in Texas).
Heh, my wife was recently telling a story about how she taught our son how to run into a hug. She did not, I actually did, but I let her have it since she was so excited about it.
Haha, this is my dad too. He doesn’t seem to realize it. It’s me as well. These are stories that tend to strongly relate to our identity. Like, if you get to know me I will talk about meditation or another story like me training with Wim Hof for an experiment at the Radboud Universiteit.
Jeremy Vine once wrote a story about Boris Johnson which I thought was the pinnacle of this. It was published on his Facebook page and I've since lost the link, so you're going to have to read it on Reddit where someone has posted the whole thing again.
It tells the tale of how the man who was going to become Prime Minister of the United Kingdom used to play the improviser and ex tempore comedian, in a practiced and automatic way.
> “Oh just preparing my off-the-cuff remarks for tomorrow”
That actually reminds me something I head from Corridor Crew. Wren was referencing Brandon Laatsch (timestamped link to clip [0])
| "Shower thoughts only work, when you put in the work." You have to actually spend time trying to brainstorm something, trying to think of something, and at the end of it you'll've made no progress and be discouraged. But then randomly, later, in the shower, doing something completely unrelated, the idea hits.
It's weird, but I think it works. My best guess is that (part of this can be explained that) by spending time thinking about the solution space, your brain will then subconsciously start to generalize. Which would mean this is, to some extent, a trainable skill. I'm sure Churchill was also a master of moving conversations to topics or subtopics where he could more quickly make off-the-cuff remarks. But I think even the "slowest" person will recognize that they are much quicker in certain categories. Slowness might not mean they lack having thought about those topics, but might just mean they've thought less about quip remarks in that domain (or even quip remarks in general).
I think we would be naive to assume quick responses are a good measure of one's intelligence[1]. I know this is common, but I think it is missing the same thing that quick responses also tend to miss: depth. You can be fast and deep, but more often people are fast and wrong[2]. More complex the topic the easier it is to be unaware of how wrong.
[Edit] I wanted to add that I found this method highly effective during my PhD. It requires a balance of churning the wheels and walking away. Progress is invisible until the finish line is in sight, so you need to spend time pushing even if it looks like you are getting nowhere. But at the same time, you need to walk away. If you keep pushing you'll never have that time for those random thoughts. There's a laundry list of famous physicist[3] who used to "only work" for a few hours a day and then do things like go on long walks or play tennis. I think that fits into this model. It seems to be a critical aspect for any creative work. Honestly, I would find that the most common mistake I would make is sitting at my desk for too long. It results in a narrowing of focus. There's a lot of times we want that narrowing, but there's also plenty of times we want to think more broadly. I think this is very true for programming in general. I can sympathize with managers who look at people doing these things and interpret them as being unproductive. But I think the reality is that productivity is just a really hard thing to measure when you're not a machine stamping out well defined widgets. I think this ends up with us just making fewer "widgets" and of lower quality. I mean it isn't like you can measure quality by anything as simple as the number of lines of code or number of Jira tickets knocked off. Hell, if you are too narrow your solutions are probably creating more tickets than you're knocking off! But that's completely invisible, only measurable post hoc, and even then quite difficult to measure (if not impossible).
We often talk about current "titans" and all of them boast their long hours and "dedication." People like Elon suggesting 120hrs or the growing 996 paradigm. But I'm unconvinced this really checks out. If anything, it appears much more common that Nobel scientists worked fewer hours, not more. We're all not working on Nobel level work, but it does beg the question of what the most effective strategy actually is. Certainly we can't conclude longer hours at the desk yields better output. We can't counterfactually conclude that Dirac would have been even greater had he spent 16 hrs a day working rather than a handful. "More hours" just seems to be a naive oversimplification, highly related to these "shower thoughts"
[3] Dirac is a famous example, who colleagues would also jokingly use the unit "Dirac" in reference to "one word per hour". Notoriously "slow" thinker, but a surefire candidate for one of the smartest humans to ever exist. Poincare famously worked 10am till noon then 5pm till 7pm. Darwin followed a similar model.
People like Elon suggest 120 hrs or 996 for the employees that work under him implementing his ideas-- the people rolling up their sleeves and putting hammers to nails. Most of the people in an org do not need to be involved in deep level thinking.
> nobody ever changed the world on 40 hours a week
- Elon[0]
Yet... I listed a few... He goes on to suggest 80 is good with spikes into 100. I mean Elon is notorious for putting in those long hours himself[1], but he's definitely wrong in the quote.
So... who do you think those demands are for? He seems pretty clearly to be demanding it from engineers to execs. That also matches the experience of everyone I've known to have worked at SpaceX, including both programmers and aerospace engineers. Same with Tesla.
Also, thought I'd drop a link to this 996 HN post from the other week[2].
Honestly, I'm not sure who you're referring to, because when not taken literally that would seem to cover literally every employee.
Elon does this because being able to order all those people around makes him feel special and important, not because it actually works.
Study after study proves that productivity drops off rapidly when people are tired and stressed - which should be common sense, but apparently it's too common for some of our notables to understand.
So instead of actual work output you get productivity theatre. Everything is dramatic and shouting happens, but - for example - Tesla still doesn't have anything resembling FSD while more modest companies are much further along.
Musk's use of social media and his involvement in recreational activities is way too much for 80h/week. He's clearly more of a 4 day workweek type tho shy about it.
In a way it shows what a craft it is to not only have jokes but also be able to present them again and again in a way that appears fresh.
Being a comedian is hard not because writing jokes are hard but because it's actually an intersection of a number of hard skills that require a LOT of practice to hone.
Do you have any advice for accumulating and sharing relevant anecdotes? I struggle with sharing anecdotes that have an "Aesop", or directly relatable point, even if I've lived such experiences.
> Do you have any advice for accumulating and sharing relevant anecdotes?
For me blogging for the past 15 or so years has been the secret ingredient. I regularly sit down and distill things into an approachable form then send off to my audience to see if it lands. If yes, I mentally add to my reference list. If not, I engage in some clarifying back-n-forth and try again next time.
These days in a more leadership position I get a lot of reps of this synchronously as I work with younger or less experienced folk.
There are sometimes long pauses before my response or even mid-speech, during which I’m thinking about what’s said. But the delay is often interpreted as a cue for someone else to respond or change the subject, which often leads to not being able to say anything that i’ve spent so much glutamate to process.
I used to say “one moment” every 5 seconds while I think, but that was distracting.
Sometimes, I do this thing with my eyes jumping them around as if I’m reading a book; that gives people something to look at while they wait, like a spinner indicator.
As as over-thinker myself something I didn't appreciate until too late in life is the necessity of practice.
If you want to be able to hit a ball it doesn't matter how much thought you put into it - the learning is all about programming your lower instinctive brain and it only has the input device of repetition. This brain level has the ability to work at much lower latency - which is critical for reactive physical tasks.
I suspect it is the same here. You can certainly learn to speak using different levels of your brain as well. Case in point public speaking - the reason this is hard is generally you have to trust your mouth on automatic mode to follow behind and using the thinking part of your brain to better plan (or remember) ahead to build a narrative path.
Yeah. That does work. I do that sometimes, though my eyes start to hurt when I roll them up for that long.
I also find it easier to do something with my eyes than to do nothing while thinking. That’s probably just me.
In the future, I might want an led embedded in one of my temples, that will blink like a network switch port or hdd light that indicates brain activity.
Also filler text while you think is good to practice. “That’s actually a well informed question, what I’ve seen is….” Buys you 5 seconds if you can say it on autopilot and think while your mouth is moving
lol, yeah the trick is to use them as sparingly as possible and have a dozen or so to rotate. I'm not sure if I qualify as a slow thinker the way OP and some others here are discussing it. But, I do frequently need more time to answer complex problems. I've also started just feeling OK telling people I can follow up with them on certain questions as it's too complex to answer off top of my head. As with anything, it's a balance, I find if you can't answer simple questions people will lose confidence in your answers.
That is a hilarious bodymod. And considering that is is possible to get rough brain activity indicators in a non-invasive manner, something like it could actually be made.
To an extent it’s a skill you can practice if not learn.
By nature I’m a slow thinker but I can mode switch if I need to but it’s exhausting after a while in a weird way I put it down to working in the trades before switching to programming full time, some of the fastest funniest people you’ll ever meet are tradesmen on job sites (introversion doesn’t mean poor social skills after all though they get conflated).
If you are generally happy as you are don’t sweat it, be a boring world if we where all the same.
I am also the slow-thinking dev married to a quick thinker, and it's a good pairing. I know couples where they're both quick thinkers and things are so mercurial it's hard to believe they're still together, but maybe the excitement keeps it going.
I enjoy watching Harry Mack videos on YouTube where he freestyles and can work in something that happens like someone walking into the frame into literally the next line of his raps. This capability is so absolutely outside of the realm of possibility for my brain I almost feel like he's a different species.
I‘m very similar. I noticed that people who are very easy to speak to share one trait: they have no shame to tell you the same story multiple times. It bores the hell out of me every times. If i try to do it, i get bored as well.
My boxing coach once described fighting as a conversation. I am inclined to agree.
In boxing you don't have the luxury of taking your time to think otherwise you get punched in the face.
Improving at conversation is like boxing - it can be reduced into structures and scenarios. Combinations and responses can be drilled in. Ultimately once the foundations are bedded in there is plenty of room for self expression and creativity.
The funny thing about social interaction is that we all talk to each other but there are people who live breathe and hone the art whether formally or informally while plenty of us just stumble along doing just good enough...
Due to this, career success is going to be highly dependent on chosen profession. Some professions are inherently more social and require heavy networking. Software and engineering as a whole, is a good profession for slow thinkers to excel. It can also be meritocratic so any personality quirks or pedigrees get normalized when ranking by performance.
I'm self taught programmer, I don't code professionally but feel confident that I can build any software related widget I ever dreamed up. That said, coding interview stuff I have seen is like gibberish and I don't see how it correlates to the day to day job of building software. Maybe some types of software (lower level systems stuff and such come to mind). It seems like everyone online discussing it feels about the same too. So honestly not sure if that's signal or noise for you, lol.
I think this is not really a bad trait. If you think about it from the other person’s perspective, they really don’t expect you to make jokes or entertain them
> they really don’t expect you to make jokes or entertain them
Oh, but they do, if you want to have future conversations with them. As a slow thinker with the same social issues as OP, trust me, they do. Nobody wants to keep talking to someone they consider boring, and first impressions are still the most important impressions.
you can try getting into improv comedy to develop this sort of skill. I'm also generally a slow thinker, but i dont actually think we think slower, I think we have too high a barrier for what we allow ourselves to say. We're afraid of making a mistake or saying something stupid, but most people just blurt out the first thought that percolates from their subconscious.
It's a shitty system, if one side just needs to succeed one time while the other side needs to succeed over and over again.
What really should be done is to disallow proposals, which are kinda the same. Once a mass surveillance proposal like this is defeated, it shouldn't be allowed to be constantly rebranded and reintroduced. We need a firewall in our legislative process that automatically rejects any future attempts at scanning private communications.
> What really should be done is to disallow proposals, which are kinda the same.
This very much exists in a lot of parliamentary rules authorities, but it's usually limited to once per "session." They just need to make rules that span sessions that raise the bar for introducing substantially similar legislation.
It can easily be argued that passing something that failed to pass before, multiple times, should require supermajorities. Or at least to create a type of vote where you can move that something "should not" be passed without a supermajority in the future.
It is difficult in most systems to make negative motions. At the least it would have to be tailored as an explicit prohibition on passing anything substantially similar to the motion in future sessions (without suspending the rules with a supermajority.)
I don't know as much about the French Parlement's procedure as I would like to, though.
Is there no way to codify a negative right, like “The right of the European people to privacy in their communications and security in their records through encryption shall not be infringed?” Negative rights reserved to the people should be more important than positive laws granting power to the government.
Yes; they could amend the definitely-not-a-constitution (for branding/eurosceptic-appeasement reasons, the EU constitution was rebranded as the Lisbon treaty before adoption). Arguably such a right may exist already and this legislation might find itself on a collision course with the ECJ if it passed (notably the ECJ nuked _another_ intrusive law, back in the day: https://en.wikipedia.org/wiki/Data_Retention_Directive).
In some ways yes but we've already seen with covid that governments are happy to behave unconstitutionally even when it's clear they will eventually lose in court - by then their targets have already been dragged through the mud.
This rule can really hurt. e.g. Theresa May tried passing a deal to keep the UK in the Customs Union. The speaker wouldn't allow it because the same deal had previously been rejected, even though she now had the support for it in the house.
I wonder if it'd be possible to fix a lot of these issues by having a constitution with damn near impossibly strict standards for changing it that rely on the entire population agreeing (or close to it)?
So there might be a right to privacy or freedom of speech enshrined in law, and the only way to change it would be for 90+% of the population to agree to change it. That way, it'd only take a minority disagreeing with a bad law to make it impossible to pass said law. Reactionaries and extremists would basically be defanged entirely, since they'd have to get most of their opponents to agree with any changes they propose, not just their own followers.
I know this behavior all to well. I definitely think that this can be used to your advantage: You learn alot in the struggle and get a deep understanding of the problem.
But you really need to step back once in a while and contemplate, if the thing you're doing is really worthwhile.
One of the most precious resources is time. I didn't appreciate this insight a lot while I was young. But as I grew older I needed to be more careful how you I was spending it. In this regard like I like the saying "youth is wasted on the young". But this also enables you to be more focused in your approach. Fail fast is a lot better than spending years on a problem with bo end in sight
Wouldn't it just solve a whole lot of problems if we could just add optional type declarations to json? It seems so simple and obvious that I'm kinda dumbfounded that this is not a thing yet. Most of the time you would not need it, but it would prevent the parser from making a wrong guess in all those edge cases.
Probably there are types not every parser/language can accept, but at least it could throw a meaningful error instead of guessing or even truncating the value.
I doubt that would fix the issue. The real cause is that programmers mostly deal in fixed-size integers, and that’s how they think of integer values, since those are the concepts their languages provide. If you’re going to write a JSON library for your favourite programming language, you’re going to reach for whatever ints are the default, regardless of what the specs or type hints suggest.
Haskell’s Aeson library is one of the few exceptions I’ve seen, since it only parses numbers to ‘Scientific’s (essentially a kind of bigint for rationals.) This makes the API very safe, but also incredibly annoying to use if you want to just munge some integers, since you’re forced to handle the error case of the unbounded values not fitting in your fixed-size integer values.
Most programmers likely simply either don’t consider that case, or don’t want to have to deal with it, so bad JSON libraries are the default.
This is actually a deliberate design choice, which the breathtakingly short JSON standard explains quite well [0]. The designers deliberately didn't introduce any semantics and pushes all that to the implementors. I think this is a defensible design goal. If you introduce semantics, you're sure to annoy someone.
There's an element of "worse is better" here [1]. JSON overtook XML exactly because it's so simple and solves for the social element of communication between disparate projects with wildly different philosophies, like UNIX byte-oriented I/O streams, or like the C calling conventions.
reply