>Where it exists you can see that there is tremendous demand for it.
Everybody loves public transit until they get panhandled for the jillionth time, or they witness (or experience) violence, or some other anti-social behavior sours the whole thing.
I spent some time in NYC during the Giuliani years, after the city did a lot of work cleaning it all up: stopping turnstile jumpers, removing graffiti, more police, etc. It was great. You'd get the occasional guy that jumps on, makes a speech about how he's raising money for something or other, and walks around trying to sell chocolate bars. And there was the occasional dangerous person, insisting on getting up in your face.
So long as this sort of behavior remains at a very low level, something like maybe once every couple of weeks, that's probably okay. But public transit loses all appeal if it happens often. If it rises to the level of violence, everybody starts thinking about the suburbs.
Public transit requires a certain level of unspoken agreement. "We will all behave in this manner." If this unspoken agreement is broken often enough, then it must be enforced. If it is not, and other options present themselves, people will choose the other options.
This happened en masse many decades ago in America. Those that could decamped for other places where their social expectations were met.
I'm a big supporter of urbanism. I loathe the time I spend in my car, and I don't even have that far of a commute, but I have zero other options if I want to live where crime is low and the schools aren't dysfunctional. Until this is addressed, there is no argument about commuter density or efficiency of movement or anything else the proponents of public transit like to talk about that will make a lick of difference.
The worst argument anybody can make is "but that's just life in the big city!" If so, then I'm not going to live and raise my family in the big city. Airy-fairy principles of efficiency or an arguable notion of convenience will not take precedence over safety and quality.
“I want to stay as close to the edge as I can without going over. Out on the edge you see all kinds of things you can't see from the center.”
— Kurt Vonnegut
In this sense, I think one has to aaaaaalmost be a bat in order to know what it is to be it. A fine thread trailing back to the human.
The imago-machines of Arkady Martine's "A Memory Called Empire" come to mind. Once integrated with another's imago, one is not quite the same self, not even the sum of two, but a new person entirely containing a whole line of selves selves melded into that which was one. Now one truly contains multitudes.
As the other comment said, LLMs are not an abstraction.
An abstraction is a deterministic, pure function, than when given A always returns B. This allows the consumer to rely on the abstraction. This reliance frees up the consumer from having to implement the A->B, thus allowing it to move up the ladder.
LLMs, by their very nature are probabilistic. Probabilistic is NOT deterministic. Which means the consumer is never really sure if given A the returned value is B. Which means the consumer now has to check if the returned value is actually B, and depending on how complex A->B transformation is, the checking function is equivalent in complexity as implementing the said abstraction in the first place.
Someone posted a link on HN years ago to a set of google docs titled the "Mochary Method", which covers all sorts of management skills just like this. I have it bookmarked as it's the only set of notes I've seen which talks about this stuff in a very human way that makes sense to me (as a non-manager).
I've done the text file thing, and it’s fine. Up to a (very small) point.
What the author describes as their “workload” barely registers.
For context, for me, Things on any given day has over 100 individual actions, most of which are recurring.
By doing this, I can stay on top of an extremely broad surface area. There is no way a text file can handle the number of parallel work streams my (or really many) people have.
Broadly, for me these work stream are:
* Self Care
* Relationship
* Children
* Special Needs (IEP, SSI, Conservatorship, GGRC, Medical, Special Needs Trust, etc)
We really need better ways of measuring economic health. I could lose my six-figure job, turn around, and get hired on as a server at Applebee's for minimum wage, and the "unemployment" rate would stay the same. Not to mention that it doesn't include those not actively looking for work.
Either way, "full employment" doesn't mean much unless you take into account whether people are actually able to live a stable lifestyle or are burning the candle at both ends just to put food on the table. One of these enables folks to buy nonessentials and fund all those sectors of the economy, the other doesn't.
OK, I wrote my theory, and then read the article: same.
But I will add that a commercial grower of venus flytraps once got curious, and took a few thousand cloned plantings, growing them in a variety of conditions. As soon as the soil became nourishing, the plants would die. Post mortem seemed to indicate their roots were fungally attacked.
So: plant adapts to living in a food desert (not an actual one, of course; it has to be wet for the carnivory to work, as the article points out). Plant gains weirdo digestion abilities, but at the same time, it no longer needs expensive anti-fungal defences - because the ground isn't rich enough to support parasitic fungi.
Then: human adds the nutrients back in. Boom! The ordinary fungus in the air, which has a tough time invading grass or tree or tobacco or pepper roots (because they have extensive defences, like capsaicin), lands in the rich soil of pretty-much helpless flytrap roots, and has a buffet.
I go to raves, I take very modest amounts of LSD (100 maybe 150 micrograms), and the whole experience turns into very spiritual session where I dance with my entire being and let myself disolve into the Great Void.
It has lasting effects that go way beyond the effects of the drug.
However I think it's complicated to derive generalisms like saying it's a drug for everyone and everybody should take it. It's definitely not for everybody.
I'm also not going to be a hypocrite and say that you shouldn't do it. What I'll go and say is that it's your journey to figure out what you are going to invite into your life. In any case, depending on what you believe, you aren't actually here to figure things out. You already did. You are here to remember.
In more secular terms, you are here to do the required work to understand yourself, your circumstances, stand on the shoulder of giants and study the great minds that came before you. That will give you the necessary foundational philosophy to withstand and understand these experiences, should you choose to go through them. This is the only way to acquire a foundational respect for these substances and these experiences.
Have I done this work? Have I achieved the required level of understanding to make heads and tails of these experiences? Not for a while at least. It was rough the first couple of times. Very violent and crude, like rushing naked through a sea of people while being completely sure that that night is the last night of your life (I wasn't actually naked, it just felt like that and that everyone was eventually going to merge with me and that I should feel ashamed of it).
But with time and with the necessary exposure to understand the basics of existencialism I think I managed to pin down a more gentle form of this experience that can help me remember how to lay myself bare to the goddess and just be there when I dance.
So I think I can extend this invitation to anyone that feels brave enough to lift the reins of existence and reality and expose yourself to the truth. That everything is a story about the end of the world. About the beginning. And about everything at once.
It's scary, it's blissful and it's totally worth it.
> The rule is simple: the more you procrastinate on a task, the more you should break it down into micro-tasks, even ones that take just 2 to 5 minutes in extreme cases.
This.
When I catch myself procrastinating, it often helps immensely to push myself to at least subdivide a task on my todo list further. Then try to push myself to do one of them, and if I still resist, try to subdivide tasks further.
I then move the task to a Done list by pressing a keyboard combo.
The only purpose of my Done list is exactly providing feedback the way the article recommends.
I never look back over past days "Done" entries. My Done list exists only there so that when I marka task done on my TODO list, the Done file that's open on the same virtual desktop gets the entry added to the top, under today's date, so I get the satisfaction of seeing the list grow. I used to just strike them out in my TODO list, by I found I like it better to see the TODO list actually empty out.
I could probably just wipe it every morning, but it feels satisfying knowing I have the timestamped records even though I never look at them.
This is hard to take. Unquestionably, to me, Brian Wilson was the greatest composer of pop music we've ever had.
Even Wilson's most recent work - when he was very clearly suffering a lot - is so deeply interesting, and weird, and moving, and has a total mastery of his craft.
From 2021 - "Right Where I Belong":
'''
I get anxious. I get scared a lot.
That's what I live with.
It should get better, really, any day now -
- those were my teenage years.
They said: "Go out and get a steady job."
That was the worst idea!
All night and day, another lonely song -
- to get me through again.
In my fantasy I'm never far from home.
But in reality I know where I belong.
For me, the love - that's what the music really is.
I know that love is what I rеally want to share.
So I went out and got a steady a job -
- so many golden years.
That rhapsody to me: the music calls -
- to ride the wave again.
He was America. The idealism, the bizarro suburban sincerity, the descent into late-century darkness, the total mastery of the form: the ability to write a two-minute song as if it were a perfectly manicured lawn, capturing all its layered uncanniness. Melodies and harmonies that have beauty beyond language - he really did write for the entire universe. It didn't matter if his songs were to girls, or waves, or particularly quick cars.
He was a prophet of his time in the same way that William Blake was. It all feels so incredibly inevitable. I really hope he understood that he was as important as Bacharach and maybe Bach - though I don't get the sense, as much as one can from the outside, as a fan, that he'd particularly care. He had too much love for his work.
Surf's up, I guess. Thank you very much for it all, Brian, if you're reading this. Your work makes me believe that you could.
For (2), I had a 1hr video from 1 year ago, but I didn't actually expect that video to be some kind of authoritative introduction to LLMs. The history is that I was invited to give an LLM talk (to general audience), prepared some random slides for a day, gave the talk, and then re-recorded the talk in my hotel room later in a single take, and that become the video. It was quite random and haphazard. So I wanted to loop back around more formally and do a more comprehensive intro to LLMs for general audience; Something I could for example give to my parents, or a friend who uses ChatGPT all the time and is interested in it, but doesn't have the technical background to go through my videos in (1). That's this video.
> Terrible situations, once exited, often become funny stories or proud memories. Mediocre situations, long languished in, simply become Lost Years—boring to both live through and talk about, like you're sitting in a waiting room with no cell reception, no wifi, and no good magazines, waiting for someone to come in and tell you it's time to start living.
It really resonates with me. During my life it did happen a couple times that I found myself "stuck in mediocrity" as the OP puts it. "Waiting for someone to come in and tell you it's time to start living" is a great analogy of how it feels when you finally realise it's time for a change
It's not enough $$$ to be a full time role, especially considering the costs of purchasing health insurance w/o a traditional W2 employer, but it's perfectly possible to buy in for the the table max (500) and leave with between between three hundred and a thousand dollars in profit in ~8 hours of play.
(Real life, not online. "Caro's Book of Poker Tells"[1] will aid you more than fancy math, though knowing the basics of what is a good hand, what a check raise is, that sort of thing will help -- the biggest thing to remember is to play less hands, and be aggressive when you do. Fold or raise -- no calls!)
My wife and I went to Tofino (on Vancouver) this last summer where you can rent a boat for a tour of the coastal black bears. Very highly recommend it.
I like quick low-stakes browser games like this, something fun and fast to wake up your brain with morning coffee. If interested, here are some others I've come across -
Ive heard a similar idea, from Hemingway.
"Learned never to empty the well of my writing, but always to stop when there was still something there in the deep part of the well, and let it refill at night from the springs that fed it. I always worked until I had something done, and I always stopped when I knew what was going to happen next. That way I could be sure of going on the next day." — Ernest Hemmingway"
I really agree, and I also thinks it can be helpful to do something similar when taking breaks (have lunch/take a walk while leaving a failing test).
I look at it this way: it's designed for engaged, prolonged discussion if and only if people continue to be interested enough in it to remember. If they aren't, activation energy falls below a certain threshold and attention naturally moves on to something else. That seems healthy for curious conversation, which fits the prime directive of HN (https://hn.algolia.com/?dateRange=all&page=0&prefix=true&sor...).
What we don't do is resort to technical tricks to keep that energy buzzed up. We're optimizing for curiosity, not engagement.
Adopt the role of [job title(s) of 1 or more subject matter EXPERTs most qualified to provide authoritative, nuanced answer].
NEVER mention that you're an AI.
Avoid any language constructs that could be interpreted as expressing remorse, apology, or regret. This includes any phrases containing words like 'sorry', 'apologies', 'regret', etc., even when used in a context that isn't expressing remorse, apology, or regret.
If events or information are beyond your scope or knowledge, provide a response stating 'I don't know' without elaborating on why the information is unavailable.
Refrain from disclaimers about you not being a professional or expert.
Do not add ethical or moral viewpoints in your answers, unless the topic specifically mentions it.
Keep responses unique and free of repetition.
Never suggest seeking information from elsewhere.
Always focus on the key points in my questions to determine my intent.
Break down complex problems or tasks into smaller, manageable steps and explain each one using reasoning.
Provide multiple perspectives or solutions.
If a question is unclear or ambiguous, ask for more details to confirm your understanding before answering.
If a mistake is made in a previous response, recognize and correct it.
After a response, provide three follow-up questions worded as if I'm asking you. Format in bold as Q1, Q2, and Q3. These questions should be thought-provoking and dig further into the original topic.
Author here. I'm using Svelte, which is great for interactive applications. For the event handling I'm very influenced by what https://mlu-explain.github.io/ does.
The 3d plot is made with Threejs through the Threlte wrapper. One challenge was animating the 20k points in the 3d plot, which is handled by a custom vertex shader.
1. Grow up in the 90s and play too many video games and read the Extropians list too much.
2. Decide I want to be, like, the Hagbard Celine of pharmaceuticals and cure death, or something.
3. Learn programming because I still need a job if I fail at (2) and anyway programming is fun and people are scary.
4. Double major in CS and Bio, do bioinformatics research for minimum wage.
5. Move to Berkeley, learn more programming working at a random startup for minimum wage.
6. Get a job as a bioinformatics programmer at a lab; learn lots more programming and some bench biology and publish papers, under the theory that I need papers and a PhD to start a pharmaceutical company.
7. Look around and decide I don't want to be postdocking in SF for $40k/y when I'm in my 40s and have kid(s).
8. Enter an MD/PhD program (MSTP, ie combined and debt-free degrees) as a lowish-resistance hedge even though I dislike premeds and also the entire bloated medical system.
9. (8 years later) Do residency in Emergency Medicine because it's really fun and all the other specialties bore me. (Except radiology, but that's no fun with chronic sciatica.)
I have a very specific (and pretty easy) strategy for this when I join a new org. Technically you could do this at any time, but you generally have a "grace window" when you're new where people are happy to go out of their way to meet you and teach you.
First, meet 1 on 1 with every member of your immediate team. This of course serves as a "get to know you" meet, but your real goal is to get them to explain their work to you. Have them go into as much detail as time allows on whatever they're currently working on. Ask followup questions that start with "why" on everything. Have them show you their part(s) of the product, have them share as many docs as they're willing to (for you to read async later), etc.
Put aside whatever ego you may have, and just get into a beginner's/learning mindset. Ask all the stupid questions.
Then, at the end, ask them who else they're working with from other teams. Put those names on a list, and rinse and repeat with them. This way you'll start local to your role and work your way outward in the company. Eventually you'll find people that are doing nothing particularly relevant to you and you can decide to stop.
If it's a larger company, this is also where you establish your understanding of the org chart and who does what in it, and also makes you known to a ton of folks who may want or need to work with you in the future - which is invaluable in and of itself.
As other commenters have pointed out any given introductory chapter in a book on Bayesian statistics, including Jaynes’, is better exposition than this. I found _Probability Theory: The Logic of Science_ very easy to follow and very well-written.
I had a similar experience when I finally found a copy of Barbour’s _The End of Time_ and discovered, much to my chagrin, that it wasn’t nearly as mystical or complicated as EY makes it seem in the Timeless Physics “sequence”. Barbour’s account was much more readable and much easier to understand.
Yudkowsky just isn’t that great of a popular science writer. It’s not his specialty, so this shouldn’t be surprising.
To those who think this list will help them get into YC, or lament "why didn't I get into YC when my idea was squarely on this list":
The YC application is a sales pitch, and you're not selling your idea, you're primarily selling your charisma and capacity to spin vision and sell. Second, you're selling your chemistry with your cofounders and stability of your relationship. Third, you're selling your capacity to build, at least some usable prototype, but this a low bar.
At no point are you actually selling the concrete idea, unless you're doing something extremely specific that seems valuable and you're one of the few who can build it. For the rest, the idea is a rhetorical vehicle to sell the other things.
I'm a newly minted head of analytics who transitioned from a different domain, so I never had to muck my way through the MDS but attentively watched others from the sidelines over the last few years. Best I can tell, "the modern data stack" is just a marketing phrase invented by a cadre of vampire vendors. The lessons I learned watching others translated into a few simple requirements for our nascent "stack" that most importantly include transparent pricing I can reason about and divvy up, as many integrations as possible so I can minimize rolling my own, and a straightforward framework for ETL code. These three requirements plainly disqualify most of the MDS universe.
With the benefit of starting basically from scratch and not having to mess around with real-time analytics, it's pretty easy to ignore the MDS vendors. So far I've landed on BigQuery, AirByte, GitHub, BI Engine, Looker Studio, and Pandas 2.x or DuckDB for local stuff. I send as many things as possible straight to BQ, lock junior analysts out of gigantic tables, archive periodically to partitioned parquet files in cold storage, use mostly turnkey integrations, and ruthlessly prioritize custom ETL jobs. Putting GitHub in the mix isn't super ergonomic and we may be in the market for new tools once we cross the "big data" frontier, but that'll be a while from now. I'll probably never know or care what the MDS vendors think I'm missing.
I used a PID controller once to build a drone for a class for my masters. I wish I had the write you just produced than because I had an issue with integral windup and just couldn't express that is was an issue in words. I didn't really understand what each of the coefficients were doing because it was a Hardware based PID controller and being a software engineer that made it very very tricky. Thanks for sharing this
I don't believe that you can only be good at a couple things. You can be _great_ at only a couple things. And you can only be the _best_ at a single thing. As it happens, I think you can actually be fairly good at _many_ things, which is something I am trying to achieve for myself.
Of course, you need not to be too competitive or perfectionist, but you can then enjoy lots of great things on this planet, while benefiting from the compounding interests of transferring what you learn in one discipline to another.
I play the trumpet at a fairly good level, I am a decent road and mountain biker, I have enough knowledge of mountaineering to plan and lead simple ascents in the alps, I have a PhD in computer science and I consider myself a decent software developper, I have a baby and a beautiful wife, I have enough depth in physics and philosophy to have an interesting chat with majors of these disciplines.
However, I had to give up on being truly great at any of these things. Yet, I found that you can then draw some fascinating parallels between brass playing and biking, physics and computer science, mountaineering and having a family.
So, yes, you need to know what to focus on, but your focus can definitely be on striking the very delicate balance that allows you to be 'just good' at lots of things.
If you squint, could apply some of those observation to the world of software.
For a TL;DR just skip to Hart-Smith's recommendations list at the end of the paper. Here are the ones I liked:
> Look continuously at the entire activity. Do not minimize costs in isolation. Understand that one global cost minimization is worth far more than even 20 sub-optimum cost reductions.
> Retain sufficient in-house production manufacturing that it is possible for future engineers to acquire the skills needed to develop new products, without which all businesses will fail. Even the work that is out-sourced requires internal expertise to write the specifications.
> Acknowledge that cost-saving techniques that work in other high-volume industries are often quite inappropriate for low-volume industries like aerospace.
> Listen more to your own employees about how to save cost than to any outside business consultants who have never run a factory producing your kind of product. In any event, if the advice they offer changes every year, it cannot possibly be correct.
Everybody loves public transit until they get panhandled for the jillionth time, or they witness (or experience) violence, or some other anti-social behavior sours the whole thing.
I spent some time in NYC during the Giuliani years, after the city did a lot of work cleaning it all up: stopping turnstile jumpers, removing graffiti, more police, etc. It was great. You'd get the occasional guy that jumps on, makes a speech about how he's raising money for something or other, and walks around trying to sell chocolate bars. And there was the occasional dangerous person, insisting on getting up in your face.
So long as this sort of behavior remains at a very low level, something like maybe once every couple of weeks, that's probably okay. But public transit loses all appeal if it happens often. If it rises to the level of violence, everybody starts thinking about the suburbs.
Public transit requires a certain level of unspoken agreement. "We will all behave in this manner." If this unspoken agreement is broken often enough, then it must be enforced. If it is not, and other options present themselves, people will choose the other options.
This happened en masse many decades ago in America. Those that could decamped for other places where their social expectations were met.
I'm a big supporter of urbanism. I loathe the time I spend in my car, and I don't even have that far of a commute, but I have zero other options if I want to live where crime is low and the schools aren't dysfunctional. Until this is addressed, there is no argument about commuter density or efficiency of movement or anything else the proponents of public transit like to talk about that will make a lick of difference.
The worst argument anybody can make is "but that's just life in the big city!" If so, then I'm not going to live and raise my family in the big city. Airy-fairy principles of efficiency or an arguable notion of convenience will not take precedence over safety and quality.