@meduka's banner p

meduka


				

				

				
0 followers   follows 1 user  
joined 2022 September 05 09:44:41 UTC

				

User ID: 520

meduka


				
				
				

				
0 followers   follows 1 user   joined 2022 September 05 09:44:41 UTC

					

No bio...


					

User ID: 520

The fundamental issue is, assuming HBD is correct, anywhere that adopts "colorblind meritocracy" will see massive disparities across racial lines. How do you expect blank slatists to reconcile this outcome with their beliefs? They will believe the system is flawed, racist, nepotistic, etc. Meritocracy + blank slatism is not a stable equilibrium: you need some kind of explanation as to why there are more losers among some groups than others, or else the legitimacy of the system will be called into question.

"Cultural differences" is the major non-HBD "defense", and personally I don't think it's robust enough on its own: e.g., even rich blacks score worse than poor whites on the SAT.

Eco

That brings me back. I no-lifed Eco for a couple of weeks on a public server with a few friends close to the Steam release and it was a surprisingly fun game to optimize (until our group got called into a "tribunal" for suspicion of exploiting, because our 4 players had progressed much further than the rest of the 200+ player server: supposedly we were abusing a bug that let you respec professions for free, in reality we simply specialized purely into unlocking better tech and completely ignored all of the efficiency bonuses, powering through with raw playtime). The player economy stuff was great: our currency become the de facto standard by cornering the market on higher-tier tech, and then we could discreetly mint more coins as we needed to buy up raw resources for "free". I haven't kept up with the game since, but as I understand it there were quite a few updates to prevent that particular style of gameplay.

Re: Palworld, I think your assessment is largely on the mark, though with a caveat that Palworld, mechanically, does interesting things that other (multiplayer) survival/base builder games don't. Or at least it tries to, anyway -- there are obvious issues that may or may not be addressed. Sure, Conan Exiles had a similar "recruitment" feature where you kidnapped enemies and forced them to work in your base, and you could even bring them along in a party, but it was quite horribly executed (IIRC with default settings you needed to "break them on the wheel" in classic Conan-fashion for literal hours: in Palworld you just toss a sphere and pray to RNGesus). Pals have needs and will path around your base to fulfill them: if you drop resources, they'll actually go pick them up and deliver them. This isn't technically impressive or anything like that, but it brings Palworld a step closer to something like Kenshi vs. a game like Conan where the NPCs are just crafting boosts. I recently played Enshrouded, another new Early Access survival game, and while I love the voxel-based building system (though I wish there were a few Valheim-style restrictions, like with smoke/structural support) the "NPCs" there are literally just crafting stations: you dump them in a spot and then they're completely static. At least in Terraria NPCs wander around a bit! It is shocking that no other game attempts what Palworld does. Hell, even Starfield, which is singleplayer(!!) has an utterly undeveloped base-building system compared to Palworld, and that's a game that had both Fallout 76 and Fallout 4 to build off of, but somehow managed to be a major step backwards.

Victoria 3 is pretty interesting. It's easy to make decisions early on that completely cripple your industrialization, leave you permanently running a deficit, etc. I'd recommend playing as one of the Canadian provinces until you have a handle on the economy: this gives you a benevolent protector (in Great Britain) and access to their market means you can drain their population via immigration once you improve your standard of living and make your "nation" more appealing.

Generally you want to look for goods that are overpriced and then focus hard on developing your production of those goods. Larger factories/extractors = better economy of scale which means you can start to snowball, so a single size 20 steel mill is generally better than various size 3/4 factories. You will want to focus on raw resources initially, and then move up the chain and start producing manufactured goods from those resources (there are efficiency bonuses for using local resources IIRC). Then you plow the resulting revenue into increasing your construction capacity (this is the engine of your economy -- employs lots of workers, but more importantly, consumes raw resources -> driving up their demand -> higher wages, richer capitalists -> more taxes, faster expansion).

Once you have an understanding of how to get your economy off the ground and into a positive feedback loop, everything else kind of clicks into place. It will take some trial and error to figure out what the optimal path is, but the many, many tooltips will provide enough information for you to figure out when something's going wrong. Depending on how well things go, you might be able to form Canada pretty early and opportunistically invade the USA during the civil war and seize some valuable territory. From there the sky's the limit.

Texas is a fun alternative semi-"challenge" start, after you have some experience with the game's systems. Mexico is a prime punching bag with lots of gold mines that, once conquered, can sustain your entire economy for decades with minimal taxation. You won't be strong enough to win a war against them immediately: your first decade or so should be focused on building up the warchest and the technological advantage necessary to steamroll them. If you stay independent (and maybe form an alliance with a European great power to ward off Yankee aggression) you can take California and Arizona (more gold!). Then you can slowly conquer your way down Central America and into South America.

I didn't have much trouble getting a 4090FE at launch simply through conventional means (make sure you have a store account setup, payment attached and working, etc) and being ready with my F5 key. Getting a PS5 ~6 months after launch was considerably more annoying -- I ended up sitting in stock tracking discords waiting for pings (because you'd never know when a new shipment is going to drop, or on what site) and usually those would clear out within 30 seconds or so. Which is totally doable if the ping hits while you're at your computer, less so if you're literally anywhere else.

I switched from using a pretty beefy "custom" Linux laptop from System76 to a shitty $300 Chromebook (old laptop was ~4 years old and I started having some trouble with the charging port and headphone jack) and while my expectations were low, I was pleasantly surprised: the experience is much better than expected.

The key thing is, I think, I mostly use my laptop for reading the internet and watching videos in the living room or in bed, or rarely when I go out. I have a top of the line desktop PC for gaming and coding, and Windows 11 has solid SSH and SMB support built right in so with barely any configuration I can access all my files, run whatever (CLI) programs I want, etc. Battery life is much, much better than my old laptop, something like 16+ hours of active use, and, while I'm sure this goes for most recent laptops, USB-C charging is great, because it simplifies cable management so much -- I can charge my phone, laptop, tablet, ereader, all with a single cable (and charge them off of each other if the power's out or whatever).

Probably my only complaint is the default out of the box ChromeOS Linux distro of choice is Debian (and an ancient release, at that), so I've run into some package issues when I try to do anything too interesting. Coming from Arch, Debian feels positively unusable, so I mostly just SSH in to either my VPS or my Windows PC if I need to do anything "serious".

I wouldn't be able to use a Chromebook as my only PC. But as, basically, a web browser/YouTube machine with a keyboard, it's great. I'm not sure what black magic optimization they've done to get Chrome with a hundred tabs running about as well as it does on my 13900k desktop, on this piddly i3-10110U chip, but it works. Best part is the price: I basically do not give a shit if it breaks or gets stolen or whatever, as I could easily get a replacement for 1/4 the cost of a decent mid-range laptop.

Yep, this is the right call for sure. Mumble is easy to set up and run locally, and there should be basically zero latency on a LAN.

I thought Fallout 4 was pretty decent, mechanically anyway. The gunplay was fine ("best in series" is not saying much, but I'll say it anyway), the settlement building and crafting were shallow but offered a decent respite from the endless "walk from person a to person b" quests or clearing what looks like the same dungeon post apocalyptic factory for the 10th time. I didn't play the story through to completion (and probably the less said of the writing, the better) but it was a reasonably memorable ~30 hours before I got bored. I liked the power armor. Solid 7/10.

Starfield seems considerably streamlined, even compared to Fallout 4. The loss of attributes means the only thing differentiating your character build is your choice of skills now, and unfortunately I have terminal RPG brain and cannot justify taking anything that doesn't grant me more utility (e.g., better lockpicking to open more doors and explore more locations, higher persuasion chance to open up new quest options, etc) and the combat isn't exactly difficult (you can spam medkits to brute force any encounter, even on Very Hard) so I can't see myself dropping a point into any of the weapons skills until, like, level 30+.

Companion AI seems even more braindead than I remember in Fallout 4, with followers regularly getting stuck on geometry, and they don't teleport to you until you change location, so they're useless in most fights. The space combat is basically just a minigame (and a hard DPS check if you're up against >3 enemies, as there aren't any useful mobility options, cover/asteroids are rare and get destroyed almost immediately) but the lego-style ship customization is still pretty fun to toy around with. Jump jets are cool, different planets having different levels of gravity makes combat feel a little different depending on where you are.

It feels extraordinarily casual. This is not necessarily a criticism, it's just a reasonably well executed AAA video game, with all that entails. My biggest complaint is that the design is unambitious: it's Fallout 4 In Space. Any time they had a choice of introducing more systematic complexity, they chose against. With the extended development cycle I was hoping we'd see something genuinely novel, but alas. I think they either experimented a lot (and cut a lot) or spent most of their time on content (and from what I can tell, there is a ton of it). Overall, it seems competent. It's not God's Gift to Gaming or whatever people were hyping themselves up for: it's a mainstream Bethesda game with as many rough edges filed down as possible. I'm still having a good time and would recommend it to anyone with an interest in an open world sci fi light RPG shooter.

When normies hear "ethnic cleansing" they think of ovens and Auschwitz. Hanania's (psuedonym's) actual phrasing there is much less inflammatory ("get them to leave") and while I'm sure you can find any number of progressive sources, ideologically captured historians, etc, who will claim that these things are identical, I don't think most people are going to buy it.

Personally I just use the built-in search plugins in qBittorrent, sort by seeds, set the filesize threshold to a couple of gigabytes so most of the awful encodes get excluded. For some really niche stuff (or if you just want more trackers for marginally faster downloads) you can run Jackett (also supported in qBittorrent), which searches more sites.

Well. I tried Bing Chat just now and got this.

It is worth noting that the settings besides "Creative" tend to have worse performance for these sorts of tasks. You may want to rerun it on that. Personally I don't have any difficulty believing LLMs can perform some semblance of "reasoning" -- even GPT-3 can perform transformations like refactoring a function into multiple smaller functions with descriptive names and explanatory comments (on a codebase it's never seen before, calling an API that didn't exist when its training data was scraped). It is obviously modeling something more general there, whether you want to call it "reasoning" or not.

I'll be concerned if these limits last longer than a couple of days. I doubt the GCP stuff is related (even the article you link connects it to Twitter Trust & Safety while this outage is affecting Twitter's core infra). FWIW Twitter seems fine for me now. It was severely degraded for a few hours on both app and web earlier but looks to have improved. I suspect someone fucked up deploying the new login-wall and they're running damage control, and possibly using the situation to run some experiments (or as leverage for their negotiations with API customers). This is the first major service interruption since Musk's takeover and (unless it persists!) I really think you're catastrophizing too much. I mean, Reddit was completely unusable for several days just a couple of weeks ago (though that was due to managerial incompetence rather than technical); 12 hours of degraded service is a bad look for a major tech company but hardly apocalyptic like you seem to imply.

The egg thing always bothered me, as someone who's had family members that raised chickens. Modern hens produce a shit ton of eggs and will literally go insane ("broody") and starve themselves to death sitting on unfertilized eggs if you don't collect them. What's the vegan rationale for refusing to eat them? I can see being against factory farming or whatever (I don't agree, but I can acknowledge that there exists a consistent ethical system to be against it), but just flat out refusing to consume all animal products regardless of context seems overly dogmatic.

The closest thing to a steelman I can come up with is something like, the chickens didn't consent to be your pet so it's unethical to raise them in the first place. But given that the majority of vegans I've met have pet cats, I don't think that's the logic. If it's ethical to raise and provide for an animal (with conditions superior to what can be found in the wild!) and to do so you need to perform some caretaking task that creates something usable as a byproduct (eggs from chicken, wool from modern sheep) it certainly seems as though you could reconcile eating eggs and wearing wool with being vegan, unless you're willing to bite the bullet and just admit that the modern domesticated breeds of these animals are unfit to survive and should go extinct, which is... a take.

I'm reminded of a recent viral story about OpenAI's use of Kenyan labor to train ChatGPT to avoid generating offensive content: https://time.com/6247678/openai-chatgpt-kenya-workers/

I thought about making a top-level post at the time because the "outrage" was actually infuriating to read. Somehow we live in a society where paying people to read some vulgar words is conflated with literal slavery. I couldn't summon the necessary restraint to write a neutral summary so I refrained from posting at the time, but man, I want off this ride where it is apparently commonly accepted that written words are equivocal (or, you know, in some cases, worse) than actual violence.

I was watching a streamer play the Japanese dub and the voice is normal there, so the only "tell" that the character is intended to be trans is that they're just kind of masculine and ugly.

Trans characters in general seem rather difficult to just toss in for extra inclusivity points. As you mentioned, the aspiration of most trans women is to be indistinguishable from an actual biological woman, not clearly identifiable as some weird third neither-man-nor-woman thing that disgusts normies. But if you include a trans character that obviously passes, then you need to be really hamfisted with the dialogue to make sure whoever's playing understands that yes, this character is trans and it's still likely they'll miss it. So if you want to inject your political views and make sure they actually get across you need to make it obvious through e.g., voice or appearance.

There is a dark humor here, where apparently even in Harry Potter's world of magic and wonders, trans women still look and sound like men in dresses. I don't know that I'd be happy with this depiction if I was a trans activist.

The most impressed I've been with ChatGPT has been when I've pasted in 100-200 lines of my own (uncommented, with not particularly descriptive variable names) code and had it accurately explain precisely what it does, and if prompted, offer reasonable-sounding suggestions for improvement, as well as answer more abstract questions about it. I had a somewhat lengthy Jenkins file, activated by a github webhook, which pulled code, ran a formatter, compiled the Rust code, zipped the resulting executable up, copied the zip file into a directory on a web server, grabbed the last 20 git commits with git log, wrote them into an .html file to provide a quick list of what the newest build contains, and then finally sent a message via webhook to a Discord server to notify users that a new build was available.

ChatGPT had no trouble at all recognizing all of this and even proactively recommended that contributors take care not to leak credentials in git commit messages, and could perform simple transformations on the Jenkins file (like adding a rule to send a different message if the build failed) with no errors. I also had success asking it to rewrite .bat build files to use a Makefile and clang instead of MSVC. I could see ChatGPT in its current state easily shaving 5-10 minutes off simple (but annoying) tasks like this, and also helping with boring API wrangling that is not technically difficult but requires tediously scouring docs to find the appropriate functions. Asking it to write more complex code whole cloth was less impressive and for meaningful contributions to a larger codebase I'd expect you'd quickly run into issues with its limited context window. I'd say it's roughly on the level of a novice (maybe 10th percentile?) programmer with access to Stack Overflow, but it provides solutions instantly. It's certainly more competent than some people I've had the misfortune of working with, though that probably says more about my former coworkers than it does about ChatGPT's capabilities.

I got 14/20. After getting properly calibrated it was not too difficult but the sample size is a little too small. Constantly being thrown new examples that shift the entire distribution around is rough early on. Like, someone who would clearly be left-aligned in the Anglo world is apparently right in Finland, okay, which means this guy who is pretty borderline is actually on the right, etc. I think with 50 or so examples most people would be able to achieve 80-90% accuracy.

I'd say it's pretty common knowledge for, well, anyone who's worked in any large software company. The whole "10x engineer" thing is a bit of a meme (though they certainly exist -- competent people who really are just more efficient, produce better code, solve problems more effectively, and singlehandedly do more than entire teams of other programmers -- and without any of the normie cope about how they can only achieve these results by playing loose with standard practice or eschewing maintainability or extensibility) but you don't even need to believe they're out there when the existence of 0.1x engineers is clearly self-evident. "Rest and vest", whole communities based around juggling multiple WFH jobs simultaneously, etc. Then there are the employees who are literally worse than dead weight, the -x engineers, who somehow seem to stick around due to petty political bullshit, and everyone who actually does work wishes they'd just... stop, because their incompetent meddling just creates more work for people who actually know what they're doing.

The question is whether Elon culled the right people, and it's far too soon to know the answer.

Would you be willing to make a few concrete predictions (e.g., where you think Twitter will be in 6 months, a year, 2 years, in terms of userbase, revenue, whether it is still owned by Musk, moderation policy, etc?)

A quick google suggests you can sideload iOS apps by toggling into developer mode (although this apparently doesn't work if Apple revokes Twitter's dev certificate, as they did with Epic & Fortnite). This is actually better than I expected: I had the impression iOS was totally locked down -- looks like it's only mostly locked down and subject to Apple's whims.

Yes, I'm in complete agreement with you there. The existence of scalpers is a strong signal that the original price was too low and I'd much prefer the counterfactual world where Nvidia raised its own prices 30-50%, generating more profits (thus delivering more value to shareholders). I can't blame scalpers, though -- their actions are not wrong any more than picking up a $20 bill from the sidewalk is wrong -- it's Nvidia's pricing error that they're taking advantage of, so it's Nvidia's fault. Personally I find the people complaining about Nvidia's higher prices to be more infuriating than scalpers. As you say, at least when Nvidia charges more, that value is going to R&D, manufacturing, stakeholders, etc.

This is not really correct. Scalpers, by definition, have no interest in the goods they're scalping -- they don't want them. Their only objective is to arbitrage the price people are willing to pay and MSRP. If there wasn't demand for the products at the price the scalpers were asking then they wouldn't sell and they'd be forced to lower their prices. There is no scenario where scalpers are "distorting the market", they correct pricing errors and make the market more efficient.

Voting is primarily a means of electing people who decide how to allocate public funds. I don't see why individuals who live off the government, at great cost to the taxpayer (something like $100k/yr per inmate in California) should have any input into how other people's money is spent. Determining who has the "right" to vote should probably be rolled under the IRS' jurisdiction anyway, with a voter card sent out with your tax return. If a felon gets out of prison and makes enough money to actually pay taxes, then sure, let them vote, just like any other taxpayer.

There's a separate argument, one that I'm partial to, where franchise should be tied to your net tax contribution (and if you're at a deficit, with more benefits received than paid, then you get no vote until you make up the difference), but that's even further outside the overton window, so.

I wrote maybe half a post about TI last week and never sent it because it seemed kind of unfair to criticize a game I've played and (mostly!) enjoyed for a hundred hours. It's deeply flawed and I can't in good conscience recommend it to anyone unless you're already so deep down the GSG rabbit hole you know what you're getting into. Yet it does so many truly novel, interesting things, in a setting that may be poorly written and poorly explored (although, as far as video games go, I don't think it's that bad, it's just not even close to the greats) but still compelling -- I think it's worth playing.