I’m decent at excel, but not amazing. I’ve tried again and again to use LLMs including Claude to solve specific, small, well defined problems in excel with a 0% success rate. My experience so far has been if I can’t do it LLMs can’t either.
If LLMs are a 6/10 right now at basic coding then they’re a 3/10 at excel from my experience.
Filament is really cheap. Basic pla is roughly 2-3c a gram. I couldn't see this being more than 500 grams, so maybe $10-20 for filament, maybe $30 if you use abs or something.
I'm guessing the rest of the price is paying to have it printed by a third party.
My view is the opposite, basically the adage “measure twice, cut once.”
Work without research often is actively harmful in addition to failing and wasting time & resources. Research at least improves knowledge while not wasting other resources besides time.
However, I felt the context of this conversation was self-improvement though. In this particular context, it's easier to get things done without reading any motivational books/articles (in fact, most people get things done without reading about how to self motivate), and the contrary -- reading self-improvement articles -- doesn't mean anything if you don't do the actual work.
Let me quote the initial post of this subthread, which is the sentiment I agree with:
> "The thing about working on yourself is that it’s actually work. Reading an article, or a book on behaviour, self-improvement and what else doesn’t actually change you any more than reading Harry Potter does."
The easiest way to estimate/visualize it is that the normal dump trucks you see on the road are 20 ton dump trucks. So at a maximum it's two dump truck loads, which doesn't go very far. In reality sand is pretty dense/heavy so it's even less than that.
As far as cost goes, I paid $250 for my pixel 3xl in near mint condition. New phones lose so much value in their first year that you can buy them for a fraction of their retail value. I can keep it till it's no longer supported in oct 2021 and sell it and buy another great phone for $2-300 and repeat every 2ish years.
The premium on buying brand new phones is outrageous.
People have learned from what.cd. Lots of people have the same sense of inevitable doom you do so there's more people who have put thought and effort into being capable of rebuilding quickly if that happens again. Just varroa musica alone has created a giant decentralized metadata archive(and that's not even its main purpose just a neat side effect) that would make the restarting process much easier.
The downside is that the post what community is so much smaller. Spotify et al really put a huge damper on the music tracker community, and that makes it less resilient.
> The downside is that the post what community is so much smaller.
People were saying the same thing in 2007 when OiNK shut down and What was starting up. The community is not smaller; it's more fragmented, and that's a good thing. Centralization is no better in our societal constructs than it is in our technological ones. I'll trade some immediacy and convenience for longevity any day.
I disagree, the community is both more fragmented and smaller, what had 144k users on its last day vs the 35k of its spiritual successor today. The few music trackers that have sprung up post what.cd after said spiritual successor are all pretty small. The niche genre trackers all have about the same user numbers since then so they cancel out. Then factor in that waffles has been down for some time now and the difference is pretty big.
I can only estimate because exact numbers are hard to find, but I'd say peak what era music tracker community total users was in the 350k region and we're probably somewhere around 100k now.
Does Spotify really cut that deep into the demographic of hardcore music tracker users? I'd presume those people are chasing rare and lost releases in high quality and Spotify and other streaming services aren't exactly touting rarities. Preserving underground music will always be a job for the fans, not streaming services.
Yes. Music trackers have always been a ton of effort, but in their prime trackers had the advantage of having basically a monopoly on music discovery. If you wanted to find out about citypop in 2011 what.cd was the place, now youtube will recommended it to you on a Joe Rogan video.
People didn't go to what.cd and jump through all those hoops to find rare music, they went to find good music, and now Spotify and YouTube do a good enough job of that with a lot less effort.
As a funny example of that I discovered Billie Eilish when Ocean Eyes first came out from a private tracker and now she's about as big as anyone. So it isn't just about discovering rare czech folk singers, but any music that you might like.
I think the hardcore demographic you mentioned is spot on and does exist, but they were always a minority and the evaporation of the less hardcore users explains why the scene is so much smaller now.
> Does Spotify really cut that deep into the demographic of hardcore music tracker users?
It does because private trackers use buffer as currency. The music trackers have had to adapt to use points systems now to encourage downloading activity. But even still, activity is way down from the what.cd days. It's just easier to stream the easily accessible stuff.
Also if you're not an uploader, building ratio to use the site is hard. Going from ten years of what.cd freeleech of buffer to nothing and facing the prospect of building it up again can be daunting.
I have experience in an unusual domain that's shaped the way I think about truth and misinformation. I play super smash bros melee for the Nintendo gamecube competitively, and I view it as an odd sort of sanity test for the limits of truth.
The deck is impossibly stacked in favor of truth in melee. To start off with we actually have Truth with a capital T. Press Y and dpad down and you bring up white text telling you exactly what state your character is in. Imagine being a psychiatrist and instead of the messy uncertain process of diagnosing patients you can press a button that freezes time and white text from a debug menu god forgot to remove from the public release of life appears. Time switches to frame advance mode and "DepressionModerate 27" floats over your patients head.
Not only do we have the truth, but we have a community that values performance and has a vested interest in the truth.
And not only that, but there isn't even political or partisan resistance to the truth.
Ok, so this is a world where pure objective truth exists, science tools are free easy and available to all, the populace cares about truth, politics doesn't exist, there's no real incentive to spreading misinformation, and there are no large scale or individual actors purposely spreading misinformation. And yet even in this best of all possible worlds microcosm truth always seems to be barely hanging on by the skin of its teeth.
There's always someone around the corner who says that powershielding an attack incurs no shield stun.
I don't think people appreciate just how fragile the truth is even before adding in bad guys. I think we fall too quickly into viewing truth and misinformation through the most exciting narratives. Like getting in a frenzy over shark attacks and neglecting the thousands of people killed by the boring old flu.
I've always been surprised some inspired performance artist hasn't done this to the live sound feed at a big music festival like Bonnaroo or Coachella yet. Audio runs from the stage/band to a sound guy in the middle of the field who controls the mix then from there it goes back across the field to the PA system mounted on stage. The audio cables are literally just big snakes that run right through the crowd where anyone could access them. Highjack a post mixer cable and bob's your uncle.
The only hitch would be that afaik at festivals the pa arrays are passive so you'd need to steal power too so you can power an amplifier and send a powered signal, but if you managed that it'd be even worse than the tv broadcast intrusion because there's no easy way to shut it off, the sound guy has no control because you're after him in the signal chain and nobody onstage is set up to handle something like that. It's not like beyonce is gonna climb the scaffolding and start unplugging speakers when the crab rave starts playing inexplicably.
Two problems with that approach at large scale shows:
1) These days the cabling is run inside a run of barricade bisecting the crowd, so it's in a secure area the entire time
2) Modern PA systems at shows of that size have almost exclusively moved to digital snakes / audio networks, not analog
To realistically pull that off you'd need production access to the event to tap into the audio network and re-route things. That said, given the number of people with appropriate access and the fact that InfoSec isn't a high priority, that actually seems pretty doable.
Once they realized what was going on though, a few breakers flipped would drop power to the amps or speakers (depending on whether using powered or passive arrays) and it would be over (which would happen pretty quickly since the power distro is very well organized and labeled since quick troubleshooting is often necessary).
I'm actually surprised this doesn't happen with wireless microphones more often. While the industry is slowly moving towards digital transmitters, many broadway shows still use old body pack analog transmitters on their actors. Since these shows are stationary they are likely using the same frequencies for each transmitter every night.
I can't imagine it would be too hard to figure out some of these frequencies and transmit over them into the PA. There is a pilot tone but I don't think it'd be difficult to spoof.
I used to work in a shop that rented out audio equipment to broadway shows.
I think applications of AI to creating 3d modeling assets for videogames. Gaming is a >100 billion dollar industry and art can be as much as half of costs for AAA games. It seems to me like it's the lowest hanging fruit of AI in terms of difficulty vs potential revenue.
MS Flight Simulator 2020 seems to be doing this - creating 3D maps of all structures/trees/etc out of normal Bing world flat maps. Not sure how much human touch is required afterwards, but if those early videos are representative then damn that's amazing and brings realism never seen before
That's one example, another one is for smaller assets - trees and vegetation, for example (SpeedTree). There's definitely a market for creating 3rd party tooling to help with developing a game. There's actually quite a few companies and technologies that specialize in one aspect of a game or game's world.
There's a few other interesting technologies that add a lot of content to a game for relatively small investments; think physics and destruction engines, animation technology (iirc there is something developed by EA's sports games branch that can generate intermediate animations for smooth animation transitions), world generation (height maps), etc.
People think touch typing takes a long time to learn, and getting good does take a while, but it only takes about an hour to memorize the alphanumerics and which finger types what well enough to break the hunt and peck looking at the keyboard cycle forever. Once the cycle is broken just typing casually is enough to eventually achieve mastery. It's only that one or two hours that really suck, and then that week or so of being kind of mediocre that keeps people stuck in the suboptimal local maxima of not touch typing.
If you use a computer in any way in the course of your job or daily life, touch typing is the single biggest improvement in productivity and quality of life that you can make. It is SO much faster and easier than hunt and peck.
I had a conversation with a friend who is an attorney and they blew me away with their off the cuff typing speed. It ended up inspiring me to get better at something I did all day....for someone who’s not sure where to start I’d start with gtypist typespeed or speedpad. Don’t forget about mavis beacon!
Also more specific but if you are french the keyboard layout is not great for programming (because []{}';. are not straightforward). I’m more happy since I switched to QWERTY layout for programming. Since I touch type I can easily switch depending on the task.
Just to add that this is not about QWERTY. The portuguese and spanish layouts are QWERTY, but still a pain [1]. Typically the best is to look for the US QWERTY layout or the international layout [2], the ones that have easily accessible brackets.
This is also the case for the german(qwertz) layout because whoever designed it gave zero fucks about people that use it to program. I have been much faster since switching to qwerty. The only issue is umlaute(ä,ü,ö,ß) but I just use compose key(caps lock in my case) + a,u,o,ss to type those.
I learned to touch type via https://www.keybr.com/ - Careful, you can get addicted to this site!
All other software felt like a much bigger grind, with fixed lessons that didn't fit my shortcomings. keybr adapts to your own typing, I made tremendous progress in just a few days.
I still use it a lot but also like to mix it in with https://zty.pe/ - fun little game.
I would add an odd perspective on this. I learned to touch type in grade school because the programming teacher covered all the keys with white stickers. The home keys had different colored stickers. And she setup a sheet of the keys on stands at each machine.
Adding stickers to your keyboard takes very little time, and will force you to learn to touch type. Everyone in our class was a typist within a couple weeks, and doing great by the end of the class.
I combined learning touch typing with switching to Dvorak[1] about 15 years ago.
Doing both together increased the time being annoyingly slow at typing, but it's so rewarding for the gain in comfort. (I don't touch type on Qwerty, but of course with some translating to Dvorak I can pretend.)
https://www.typingclub.com/dvorak is out a good place to start. Don't rearrange the keys on the keyboard! You're not going to look at them anyway.
[1] Nowadays there is also Coleman and variants. I still recommend Dvorak, but at the point someone is deciding between the two it's not worth an online debate.
Most people really could learn touch typing in less than an hour, I experienced this myself. If you type at a decent speed you already know the symbol position already either way, so all that's left is going for more consistent and controlled finger movements.
I easily increased my typing speed by 10WM in a single day just by doing that switch, and I think this is one of the most significant improvements regular computer users can get basically for free.
I really need to do this. At one point I got pretty far on https://www.typingclub.com/, but slid back to hunt-and-peck as soon as I stopped it. I'm learning Vim now (which has a similar effect where you get immediately much slower before you get faster), and one of the guides I was reading said not to waste your time with Vim before you can touch type.
That said, Vim is easier because as soon as you understand how to switch in and out of insert mode it works as a regular text editor. Touch typing really brings me to a standstill.
After a few years of also failing like this I combined learning touch typing with switching to Colemak, and that finally made it stick (though ruining me for normal qwerty, but it was worth it). It takes a while though and there will be a really frustrating period when you can’t type either qwerty or Colemak.
Our school made a concerted effort to have us play touch typing games during our 'computer hour' or whatever it was. It was a great help.
Though by then myself and all my friends were pretty great from playing those old Sierra adventure games. The on-screen action often didn't wait for you to finish typing. So I guess my point is, you can make it fun too if you want.
It definitely took me longer than an hour. I had been using the keyboard wrong for over a decade, so it was hard to undo. I did an hour every day for 2 weeks, I think, after which it stuck. I then wrote at 30-60WPM for a while, which was a bit of a pain.
Now I can get up to 100WPM, definitely appreciate that I took on the challenge.
If LLMs are a 6/10 right now at basic coding then they’re a 3/10 at excel from my experience.