It can be highly variable. For example in the morning or right after a nap I can visualize in extreme detail, but when I'm awake and at my most alert it will become a lot more basic.
1,024 karma · joined July 27, 2015
Apparently South Africa was the world's second-largest market-economy producer of asbestos from the 1950s to the 1980s and the largest producer of amosite and crocidolite (I had to google those and it appears among asbestos there are a few different types/lengths of fibers).
The big mines all closed around 1992/98, with a complete cessation of all asbestos mining in 2002/3. I found this scanned document from the University of Cape Town that goes into detail on the discovery, mining and a few epidemics as they called them: https://vula.uct.ac.za/access/content/group/9c29ba04-b1ee-49...
The numbers are crazy, for example in 1931 mill workers (in a sample of 100) only 14% had more than 5 years of service (the single longest having been there for 9 years). "Yet every worker experienced a cough productive of mucoid sputum, 72% suffered dyspnea on exertion and 47% reported loss of weight".
In less than 1 year 36% showed early signs, 9% advanced signs, 45% any abnormal sign. In 3 years that went up to 55%, 24% and 79% respectively; and by 5-9 years 36%, 64% and 100%.
I spent a lot of time collecting and breaking down game elements from all board games I could find, but as things go with ADD I then ran out of steam before I had any kind of functional prototype.
As always I highly recommend The Player of Games by Iain M. Banks, and although its not his best book I really enjoyed the game aspects in it.
- Googling for water outages gives a lot of results in just the last few days. In the NorthWest for example there are a lot of failing municipalities which are relying on government assistance to just make it month to month. Water trucks are a common occurrence all over. The official numbers on connection to water, electricity etc. are pretty much a joke.
- Loadshedding is indeed no more: Up to about 10-15% of households are now living off-grid, while in the industrial sector I can link you any number of metal processing plants that have closed down, the same for mines, car manufacturing etc. In the last few years our electricity bills have about doubled, rates and taxes aren't far behind either. That's not a win in the least.
- Healthcare: A few of the more well funded public hospitals are ok, but just from Tembisa approximately 2 billion Rands have been siphoned as of recently. Impressive isn't the word to use. Google for images to see the conditions of the hospitals and what the people who go there are experiencing, while on the other hand you can see videos of tenderpreneurs riding their Lamborghini's with police escorts via dirt roads in the outlying areas.
- Violent crime has nothing to do with apartheid (apart from the occasional incitement by political parties etc). We have crime because somewhere between 33-43% of the population is now unemployed, along with having only a barely functional police force. The people stuck on the bottom have no hope of changing their circumstances, which in turn is fueling crime (and violence).
- What makes you think there's less corruption now? The fact that more and more of it is coming to light? As long as the governing party allows it to happen its going to cascade down into all facets of life/business etc. They've begun to realize that they are losing the vote (and with it the power), but we're still a long way off from having any change on the horizon.
- Single anecdotes are pointless, some businesses will naturally grow while other decline, a lot of it is just random luck based on the type, area, time etc. Foreign investment is down something like 29% in just the last two years while we've taken on more than R25 billion in loans just recently.
But having said that, there's probably a few openings for research in there where you can experiment with different field/magnet shapes, sizes and counts.
There are cases where even though the sort executes more instructions that the size of the elements/code still fits into some Ln cache level and makes it faster, but in general the prefix approach comes out ahead.
They really upped the level and detailing on the art and the audio/music is quite good as well.
Design-wise they added a button to perform a forward jump which is really welcome.
For only $162 you get: "The Single Use Finger Toy consists of a singular rubber appendage attached to a retractable wire harness designed specifically for individual fingers. This ergonomic accessory provides hours of solitary stimulation through intricate finger movements, gradually detaching itself after extended use due to over-tensioned joints."
- Why long pants instead of shorts? It does make them look more dignified, I suppose.
- Why white, you know that's just going to stain the feet something terrible.
Firstly, if we don't remove the Google AI summary then as you rightly say, it makes the experience 10x worse. They try to still give an answer quickly, but the AI takes up a ton of space and is mostly terrible.
Googling for a Github repository just now, Google linked me to 3 resources except the actual page. One clone that was named the same, another garbage link but luckily the 3rd was a reddit post by the same person which linked to the correct page.
GPT does take a lot longer, but the main advantage for me comes in depending on the scope of what you're looking for. In the above example I didn't mind Google, because the 3 links opened fast and I could scan and click through to find what I was looking for, ie. I wanted the information right now.
But then let's say I'm interested in something a bit deeper, for example how did they do the unit movement in StarCraft 2? This is a well known question, so the links/info you get from either Google or GPT are all great. If I was searching this topic via Google I'd then have to copy or bookmark the main topics to continue my research on them. Doing it via GPT it returns the same main items, but I can very easily tell it to explain all those topics in turn, have it take the notes, find source code, etc.
Of course as in your example, if you're a Doctor and you're googling symptoms or perhaps real world location of ABC then the hallucination specter is a dangerous thing which you want to avoid at all costs. But for myself I find that I can as easily filter LLM mistakes as I can noise/errors from manual searches.
My future Internet guess is going to be that in N years there will be no such thing as manually searching for anything, everything will be assistant driven via LLM.
But for every positive case there are just as many negative ones. I have an intrinsic need to answer questions and will often spend an inordinate amount of time doing so, when in fact I could have used that time far more productively.
Which if you think about it is a real issue. Imagine applying at a courier company for a developer role and they keep asking you about the tracking software you've built, parcel measurement integration you've done etc., instead of asking you about your development skills. Having done those things is of course a huge bonus, but excluding 100% of people that don't have that experience excludes a great majority of candidates that could have been a great fit.
The problem is even bigger than that if I think about it. In this example they don't want to know about individual pieces of industry relevant software that you've built, they are expecting you to have shipped enterprise wide solutions that fit the criteria and that doesn't match your skillset. The role they're advertising might be a senior tech lead/developer, but you're not being hired as a programmer, you're being hired as a game maker. They want you for the games you've shipped, not for the code you've written.
Does your little games have "juice"? That's going to get you hired 100%, but mainly because of your skills as a designer, artist, tester, audio engineer etc., coding only made up 20% of that package.
But then during a break the other day I read up on Radix sort and then right thereafter implemented a prefix sum for spatial partitioning that also incorporates a bit table, CAS operations for doing multithreaded modifications etc. After learning the core Radix concept I sort of came up with the idea of using it that way myself which was quite pleasing.
Props to the author, I'll definitely be spending some time scanning the collection to find some alternate options.
I recently implemented both, first I started with OKLab which turned out really well, the gradients you get from it are amazing and the usual color sets (analogous etc.) produce really pleasing sets.
However I quickly ran into the main problem with it, which is that fiddling with its Lightness, Chroma and Hue dials doesn't produce human understandable results. For example sometimes changing L or C induces a color shift, or for some given values changing L only gives midrange values that doesn't go up or down all the way.
I then implemented OKLCH on top of that, which was the way I assumed everyone was doing it. Just have it act as the controller for the human layer, then convert to OKLab for creating gradients etc. The article doesn't really go into it, but having OKLCH as the frontend controller fixes the LCH sliders such that they produce values that make sense to us humans, while still having the superior OKLab setup in the back.
You can essentially rush to the ultimate Reaper character by just having everyone pray, which is both a game ender and a spoiler, because it really makes the game boring.
Having said that, this was for the original version when it came out, I don't know if they ever did anything about it and/or for the later GoG etc. versions.
A great example is that little children have no protocol, they won't walk up to someone and say things like hi, how are you, how's the weather, etc.
Instead they'll just walk up and either stay mute, or just talk directly about what's currently happening, without any introduction or pre-amble.
I can ask it to help me code for example a physics engine, so we're talking really hard and intricate code and it'll come up with some amazing optimizations, we're talking (recent) research paper level implementations.
Then I ask it to work on something that's relatively trivial, let's say we need a flowfield. It'll think and reason about it just as well as in the first example, but then it'll start spitting out a lot of subpar code. Its error rate will increase 10x while the global cohesiveness of the produced code will be substantially worse.
As to why that's happening, maybe its being trained on a lot more as well as worse examples of the second, whereas the first is relatively "pure".
These programming competitions are pretty much the same thing in my opinion. For us humans its a hard challenge, but in general they're asking the same-ish questions, just in different formats. They should add some questions where the participant has to invent something new, or alternatively use two or more existing concepts in a totally novel fashion.
- Light: If they make a clanking noise every 5 seconds when you pick them up and put them down its just distracting.
- Plastic: There are other materials that work, but in general metals and fabrics react too strongly with your fingers over time, whereas plastic can be easily cleaned.
Having a little water window isn't a problem, but the water is also not as directly cool looking. Usually in the '90s you'd get those mixed color oil/water toys where the colors would make them stand out more, but shaking them would mix the substances causing them to lose function.
As such the electronic version is quite durable in comparison.
There were a few small text mistakes and the image isn't quite as good as I've seen before, but overall it delivers on its promise.
When I was a kid and just started working I would still often code with whatever I came up with initially, but then you go back to that 3 months later and you have to throw the whole thing out because it's impossible to maintain or add to.
On the other hand there are sometimes additions to a language that's just so useful that you have to expand your vocabulary, for example in C# that's happened a few times. One of the notable additions there was LINQ that made manipulating data so much easier. It can become dangerous though, the same as for example with a complicated DB stored procedure.
Tcl was by far the easiest to use, while Perl sat on the other end of the spectrum for both skill needed to write it as well as to just be able to read it.
As the mantra goes, "Power, but at what cost?!?"
As for the last part, I've recently been getting close to 50 and my eyes aren't what they used to be. In order to fight off eye-strain I now have to tightly ration whatever I do into 20 minute blocks, before having to take appropriate breaks etc.
As a result of that time has become one of the biggest factors for me. An LLM can output code 1000x faster than a human, so if I can wrangle it somehow to do whatever basics for me then its a huge bonus. At the moment I'm busy generating appropriate struct of arrays for SIMD from input AoS structs, and I'm using Unity C# with LINQ to output the text (I need it to be editable by anyone, so I didn't want to go down the Roslyn or T4 route).
The queries are relatively simple, take the list of data elements and select the correct entries, then take whatever fields and construct strings with them. Even so, copying/editing them takes a lot longer than me telling GPT to select this, exclude that and make the string look like ABC.
I think there was a post yesterday about AI's as HUDs, and that makes a lot of sense to me. We don't need an all-powerful model that can write the whole program, what we need is a super-powered assistant that can write and refactor on a very small and local scale.