Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost

Enderzero posted:

in a demonstration of the dictum above that we choose not to, it don’t cost 1400. we are charged 1400

yeah, small peptide synthesis means prolly cogs is like a dollar

Adbot
ADBOT LOVES YOU

Beeftweeter
Jun 28, 2005

a medium-format picture of beeftweeter staring silently at the camera, a quizzical expression on his face
the system works

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

bob dobbs is dead posted:

yeah, small peptide synthesis means prolly cogs is like a dollar

well thats a lot of extra economic activity and growth compared to if they only charged a few dollars a month

bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost
cogs like a dollar, amortized r&d like two hundred bucks, sg&a prolly like 500 bucks on top of that

Enderzero
Jun 19, 2001

The snowflake button makes it
cold cold cold
Set temperature makes it
hold hold hold

fart simpson posted:

well thats a lot of extra economic activity and growth compared to if they only charged a few dollars a month

number go up?

number go up!!

BMan
Oct 31, 2015

KNIIIIIIFE
EEEEEYYYYE
ATTAAAACK


according to a new yorker article I read, you can get it much cheaper by having a sketchy pharmacy send you a kit with the salt form of the drug

Juul-Whip
Mar 10, 2008

Sapozhnik posted:

the 20th century brought us radio (with its beginnings at the tail end of the 19th century), automobiles (boo hiss), aviation, applied nuclear physics, computing, and transistorized integrated circuits.
you forgot lasers motherfucker. you philistine

bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost
20 amino acids, so each amino acid is 6 bits tops. in practice its less because of nonuniform distribution. semaglutide is 31 amino acids, tirzepatide is comparable but weirder. both are basically human glp1 with some weird poo poo tacked on to make it last a week instead of 10 minutes

synthesis has fewer than 30 steps, if that one paper was right

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

it was right

LanceHunter
Nov 12, 2016

Beautiful People Club


BMan posted:

according to a new yorker article I read, you can get it much cheaper by having a sketchy pharmacy send you a kit with the salt form of the drug

I've got a friend who is a nurse practitioner and has had to deal with insurance companies denying patient after patient that he tried to get on Wegovy. It turns out that even if a patient meets all the clinical requirements, if you aren't an endocrinologist they're likely to reject the prescription. As a result, his clinic is starting to talk to a compounding pharmacy who can basically assemble a version that's essentially identical but will cost about $200 for a four-week supply.

Being on the stuff myself, it's not so much a cure for fat as it is something that makes the cure for fat that we have always had (eating fewer calories than you burn, and then only eating the number of calories you need to maintain the weight you've reached) actually feasible for people whose sense of satiation has been destroyed. I kinda knew how insanely over-blown my sense of hunger was, but it was only after taking the drug that I actually understood how far it was from normal. It was a bit like when I first got glasses and could suddenly see that there was all this detail in everything that had just looked blurry before.

bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost
that just means bein fat is an endocrinological disease not a dietary one

LanceHunter
Nov 12, 2016

Beautiful People Club


bob dobbs is dead posted:

that just means bein fat is an endocrinological disease not a dietary one

I mean, yes and no. If there were some hormonal imbalance that caused people to repeatedly smash their hands with a hammer, and the more they smashed their hands with a hammer the more likely they were to do it in the future; you wouldn't say that nobody needs to worry about the effect of hitting your hand with a hammer because these people's broken fingers are actually the result of a hormonal imbalance. Diet is still the mechanism by which all of this occurs. Also, having a better diet when young means that it is less likely that the endocrinological disease will develop.

the panacea
May 10, 2008

:10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux::10bux:
does it work with weight gain related to anti-depressants etc?

bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost

LanceHunter posted:

I mean, yes and no. If there were some hormonal imbalance that caused people to repeatedly smash their hands with a hammer, and the more they smashed their hands with a hammer the more likely they were to do it in the future; you wouldn't say that nobody needs to worry about the effect of hitting your hand with a hammer because these people's broken fingers are actually the result of a hormonal imbalance. Diet is still the mechanism by which all of this occurs. Also, having a better diet when young means that it is less likely that the endocrinological disease will develop.

but you will agree that recommending hand-armor or softer hammers as treatment is worthless, just like excision for syphilis sores instead of antibiotics is basically worthless

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

rotor posted:

the tech industry should spend more time on weird input devices imo

Sooooo there's like, a huge unsolved problem right now at the heart of gaming, where you essentially run out of axis and its kind of why game design for control has plateau'd in a really specific place.

its the literal conceit at the heart of why "genred" gaming exists and nobody talks about this and it makes me so drat mad. its very clear projects which mix or ignore genre alltogether and just do a "whatever-fest" hold player attention a lot longer and its one of the missing pieces of the puzzle for pretty much any sandbox game other than letting players build their own rule-constructs and win-conditions which is my other pet-peeve but i won't get into that

for context, integrating even something like a roll command or a throttle in a game is really really non-trivial, and each is competing for the available triggers which players have said pretty clearly they want to think of as their left and right arms in 90% of games.

this is before we even begin to joke about moving the neck independent of a lead camera direction, which you actually have to do tank controls with the left stick to even perform.

we're still making games where the face-buttons and d-pad see regular use for basic tasks, so you're switching stick to buttons to stick and that's... Kind of not very good?

I get like, *why* though, because paddles for example conflict with the task of holding up the controller to begin with.

game controllers are in a weird locked in place of "enormous enshittification" where adding meaningfully good features to solve these limitations isn't really incentivised because most of the people working on controllers have no idea how we got here.

for example: The prevalence of shooters is mostly down to the fact right thumbs overall tend to have more skilled movement than left thumbs, and movement is a left thumb task because movement has been a left joystick task.

why?

arcades put it on the left instead of the right to make games harder to control so you'd give them more quarters.

as a result, game design changed within 5 years to be more about making precise inputs with buttons, than precise movements.

a few did super resist this and kept putting their sticks on the right: Those are often bullet hell games.

if we ignore the dominance of twiddle-sticks in games, there are people trying to find weird new ways of doing mouse input on the desktop and I think their attempts are at least significantly more trial/error battle hardened than Steam's work on their controller was. the fact this guy isn't literally having wads of cash thrown at him by xbox's team or something for some sort of "pro fps controller" concept that dorks would go gaga for for this thing baffles me until i remember the massive resentment this would generate in terms of community fairness.

i think "disruption" is a loving stupid term mostly used by idiots but that's sincerely one of the times i think something would be disruptive, and it shows that there's a legitimate incentive *not* to make controllers better: competitive players and their resentment mean any real advantage ever had is inherently going to get bitched and moaned at by piss-babies lambasting their lovely k/d ratios on someone elses controller instead of accepting responsibility that they themselves loving suck at the game

coming back to weird pc input devices, i fukken love my 3dconnexion spacemouse (it sits next to a Belkin N52 Nostromo, and at some point I am going to get off my rear end and 3D print a combined case at some point like so many have done before with keyboard segments) but i am so buttmad more stuff doesn't support it out of the box. the fact there's no codified standard for dealing with inputs from nonstandard devices, parsing and then translating them for software on ANY os that doesn't require scripting or programming is absurd. adding functionality should be as simple as ticking a box in a menu somewhere and then selecting a profile

but still, in 2023, unreal engine needs a plugin and zbrush still does NOT support it because the folks working on zbrush hate their users because they are literal fukken aliens from outer-space

seriously

- saving a document means saving the arrangement objects on screen
- saving a file means saving the actual polygons you are sculpting
- a tool is a layer/object with your polygons on it
- a layer is totally useless and indecipherable if you're doing 3d not 2d work in zbrush
- you have to hold and release the buttons on the keyboard to move the camera (no lmb control for you!)
- you have to drag menus around to do things
- there is no search functionality

aliens made this and its awful

hardcore "not made here" syndrome but for ux

vr also has a similar sin: "let's not have functional parity with regular controllers for no obvious reason whatsoever" -- result is instead of getting a full range of options *and* the ability to use gesture, you lose instantaneous control for lots of input options which limits the range of how you can express those gestural and positional inputs

the biggest complaint i hear when someone is commanding units or flying planes or in a vr game a lot of the time is "this is very cool but its a bit simple, isn't it?"

and i get why this exists too. they didn't want devs going "ok we'll just make a regular game with regular inputs with one small vr gimmick" so they deliberately handicapped the ability to do direct ports. i get it, i really do

but we now like, see what the value add of vr is. can we have "real" controllers now please, which are also good for regular stuff too at the same time? that would really be very cool guys.

rotor posted:

i just cant get over the idea of a coop rhythm mech fighting game, its amazing

so what like, doing an authority score where the similarity of a posture drives the capacity of it to push influence to a body?

like you'd need code which comps out for different body shapes and stuff like that, and some forgiveness for timing but you could also incorporate a gimmick which recognizes who is leading and who is following and provides appropriate encouragement to get the sync closer though then there's issues of like if the players aren't in the room together, and they'd need some visualization of eachother in vr clearly that can show the limb furthest from themselves otherwise *doing* syncronization is going to become really hard

from a purely programming standpoint i wouldn't know how to do posture check matching for differently shaped bodies, but i can very clearly in my head feel how the movement would work.

like the issue which results is you do need to limit some of the inputs so the robot doesn't just like, fall over or w/e

like do you see this being "we're driving a slightly simulated body" or "we're unlocking a goal action" like synchronized body positions become the equivalent of making an input sequence?

like if you're doing purely input sequences (like idk, doing a particular punch together triggers the robot to then do that entire punch animation independently) results in a lot of rapid gratification but it gets really annoying in the long-term because then the players will go "well why am I not doing this with a button"

there's sort of a running joke that this is the "nintendofication" of motion inputs in a lot of stuff (eg, waggle stick to do thing), because you give up any fine control in timing, or spatial/duration control of the input (you don't get say, Mario's 'versatile jump' where how long you hold the button changes its influence and outcome. this "zeroth step of motion control" basically soured motion control for... pretty much forever in the mainstream as a gimmick unsuitable for serious games, if I'm honest I think is a really huge loving shame



the thing is nobody really knows what motion control "is": gyro-aim, infared sensor, acceleration and orientation censors and positional sensors all get thrown into this category, but there's no clear definition. its a bit like the word "woke" in that sense: its "whatever gamers say they don't like that involves any kind of movement", so let's get that yuck out of the way and slightly go into why because whenever someone comes to me with vr game, i find it really helpful to explore this stuff to figure out how to conceptualize their game design in clear terms

even Nintendo's attempts to make up for this with the wii motion-plus were essentially reading an acceleration threshold so you have to meet that minimum rate, and then it turned the direction into playing back an animation along an arc -- and then when not in use, reading the orientation of the controller to change link's posture. it was "better", but it wasn't really motion-matching with positioning info because you only have orientation and acceleration information from the second sensor against the first in the controller. I could call this the "first step" of motion control.

with the "second step", you got motion-matching (which requires precise position info) such as in vr, you're not just dumping an orientation into the input but now you're actually reading finer info and thus you have precise interaction. its still not a trivially solvable thing for obvious reasons of sensor design and how references are detected and stuff, but motion-matching is cumbersome but very cool. even modern vr sadly has huge precision input issues, due to the way certain signals don't get properly filtered on say, the chips valve are using in their controllers resulting in "vr-shakes" which are an unaddressed issue by hardware manufacturers that a lot of gamedevs then have to write their own noise filtering algorithms for.

making a lot of motion inputs "work" is all about "not nintendofying your motion control" with context stuff like reaching for a backpack depends on an overlap check which is spatial, or swiping a sword is motion matched with a velocity and acceleration calculation involved so if the player does some baby fling swing, the sword stops when it hits something, but if the motion is flowing and intentional, then the power in the virtual swing is big and carries into the target and that's how you do good sword-play in vr

ok good, so this is giving us a frame of reference for the kinds of levels of scope we can deal with and we've spotted some pitfalls so let's try this:

in this sense... i guess the way to do this would be to read all the look-at vectors between both players filtered through unit-vectors from each joint root to each joint goal for a similar expressed rig, and then do a dot product between Player 1 and Player 2 ... Then tweak them for bias based on what's fair in different weird edge-cases, and then use that to drive the system?? and then to make it feel a bit more fair, you add a bias curve that's sorta like, preferring the highest score of the dot product for each joint along those unit vectors, and then only letting the other negatively affect the sum of it goes above some really off value?

i think it would require lots of fine tuning but yeah it doesn't sound like, *difficult* to do provided there's a comparison of the histories of those measurements to see if the shape of the graph itself is good, rather than just like, perfectly synced values -- and then when you do get near perfectly synced values you replace slumpy heavy robot movement with something swifter and cleaner and more proficient and more visceral reward like higher damage rates against opponents?

like you could also have a shadow only the player can see of the other player's body and then onion-skin it a little so you can feel the movement better, but that risks becoming then a crutch so it would probably be a good learning tool I think, and you could tutorialize by having the player copy some pre-saved animations where the game walks the player through principles. wrap that with some nice dialog and different movement styles and yeah you've got your onboarding system too.

huh yeah. doable, I guess rotor?

now I'm not someone with a team under me, but that's how I'd turn your game-as-a-concept as you're thinking of it into something that could be presented as a brief that you could then put in front of a team and say "okay, go build that" and know its possible with some fine tuning along the way.

I probably went off into the wild blue yonder but yeah this is me thinking "oh I've seen pacific rim"

I wish I could do this as a living. its like the big missing link in game design as a field, because a lot of game designers let their programmers decide how to execute on concepts and stressed out coders just go "ok, let's make it <half-assed implementation of existing genre> then"

edit:
inb4 "I just wanted beat saber but with two players lol what the gently caress is all this"

Expo70 fucked around with this message at 10:39 on Apr 6, 2023

LanceHunter
Nov 12, 2016

Beautiful People Club


bob dobbs is dead posted:

but you will agree that recommending hand-armor or softer hammers as treatment is worthless, just like excision for syphilis sores instead of antibiotics is basically worthless

I don't quite get where you're going with the metaphor here. If you're talking about stuff like keto and other weird non-calorie-based restrictive diets, then yeah I agree that this really puts the lie to their claims when you see how effective a basic calories-in/calories-out diet can be (when the person doing that diet doesn't feel insanely hungry all of the time and thus isn't set up for inevitable self-sabotage).

the panacea posted:

does it work with weight gain related to anti-depressants etc?

I don't know as I don't really have any experience in that area. I do think that if there are other things that could prevent someone from being able to stick to a diet while on Wegovy then it might be a bad idea to waste a bunch of money when you won't get the results. You can definitely still over-eat while on the drug, it's not like one of those stomach surgeries where it can be physically painful if you put down too much food. It's a lot harder to over-eat, since you'll feel very full, but it's not impossible. And if you eat the exact same number of calories after starting a semaglutide as you did before, then you won't really see the benefits. (Also, I absolutely dread the constipation you'll be dealing with.)

So I imagine if there are other psychological issues at play that can complicate things then it will probably be worthwhile checking with the relevant healthcare providers for their take.

bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost

LanceHunter posted:

I don't quite get where you're going with the metaphor here. If you're talking about stuff like keto and other weird non-calorie-based restrictive diets, then yeah I agree that this really puts the lie to their claims when you see how effective a basic calories-in/calories-out diet can be (when the person doing that diet doesn't feel insanely hungry all of the time and thus isn't set up for inevitable self-sabotage).

I don't know as I don't really have any experience in that area. I do think that if there are other things that could prevent someone from being able to stick to a diet while on Wegovy then it might be a bad idea to waste a bunch of money when you won't get the results. You can definitely still over-eat while on the drug, it's not like one of those stomach surgeries where it can be physically painful if you put down too much food. It's a lot harder to over-eat, since you'll feel very full, but it's not impossible. And if you eat the exact same number of calories after starting a semaglutide as you did before, then you won't really see the benefits. (Also, I absolutely dread the constipation you'll be dealing with.)

So I imagine if there are other psychological issues at play that can complicate things then it will probably be worthwhile checking with the relevant healthcare providers for their take.

calories in and calories out is true but worthless, because the cause is a bad lipostat. like saying that you need to open your windows because heat in and heat out when your thermostat's broken and your heater's jammed to on. all dietary interventions have failed at population level. no polity has successfully treated obesity en masse or even stopped the rate of increase. given an actual effective cause and actual mass practical treatment, this may change.

bob dobbs is dead fucked around with this message at 09:28 on Apr 6, 2023

Cybernetic Vermin
Apr 18, 2005

going "calories in calories out" about weight is like going "simply write 100% correct code" about security flaws.

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

bob dobbs is dead posted:

calories in and calories out is true but worthless, because the cause is a bad lipostat. like saying that you need to open your windows because heat in and heat out when your thermostat's broken and your heater's jammed to on. all dietary interventions have failed at population level. no polity has successfully treated obesity en masse or even stopped the rate of increase. given an actual effective cause and actual mass practical treatment, this may change.

according to our world in data's data set, only 40% of country-years have a positive 2nd derivative of obesity rate. 33% of country years had a negative second derivative of obesity rate

bob dobbs is dead
Oct 8, 2017

I love peeps
Nap Ghost
first derivative zero, not second derivative negative

there is venezuela and cuba after the embargoes but the way they did it is by becoming radically poor real fast. north korea and sudan and ethiopia multiple times also were really poor the entire time and had famines. industrialized rich nations (aka, the ones with mass obesity to treat), bupkis

bob dobbs is dead fucked around with this message at 10:31 on Apr 6, 2023

Powerful Two-Hander
Mar 10, 2004

Mods please change my name to "Tooter Skeleton" TIA.


fart simpson posted:

what happens when you stop taking semaglutide

bob dobbs is dead posted:

you get fat again lol

there's an absolutely insane book called (iirc) 21st century blues which features a fad for "food that isn't there" so you can eat as much as you like and not get fat but then it goes wrong and everyone gets super fat overnight or something

I think later on it involves octopus like aliens impersonating humans to have sex with them on live tv

Powerful Two-Hander
Mar 10, 2004

Mods please change my name to "Tooter Skeleton" TIA.


Expo70 posted:


*Words about controllers*

counterpoint, steel battalion:

Shame Boy
Mar 2, 2010

Expo70 posted:

its like the big missing link in game design as a field, because a lot of game designers let their programmers decide how to execute on concepts and stressed out coders just go "ok, let's make it <half-assed implementation of existing genre> then"

that's how all programming works not just game programming fyi

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

Shame Boy posted:

that's how all programming works not just game programming fyi

my understanding as someone who just dinks around in unreal engine is this:

the entire field of programming is a castle made of sand that's both also rotting and on fire some how

a lot of people with money want to "make it all better", and "make it different"

what they're actually doing is trying to invent rain to put the fire out

what they don't understand is the sand is made of triethylaluminium and rain will not put the fire out, but make cool pretty and very toxic explosions

they will then claim the explosions were "on purpose" and then try to sell that some how calling it a "pivot"

they will then be shocked when their kingdom is washed away by the rain 5 to 10 years later

every 10 years, another new status condition (brittle, melty, etc) gets added to the castle made of sand


does this sound right to you?

Expo70 fucked around with this message at 10:57 on Apr 6, 2023

AnoHito
May 8, 2014

Expo70 posted:

my understanding as someone who just dinks around in unreal engine is this:

the entire field of programming is a castle made of sand that's both also rotting and on fire some how

a lot of people with money want to "make it all better", and "make it different"

what they're actually doing is trying to invent rain to put the fire out

what they don't understand is the sand is made of triethylaluminium and rain will not put the fire out, but make cool pretty and very toxic explosions

they will then claim the explosions were "on purpose" and then try to sell that some how calling it a "pivot"

they will then be shocked when their kingdom is washed away by the rain

every 10 years, another new status condition (brittle, melty, etc) gets added to the castle made of sand


does this sound right to you?


Shame Boy posted:

that's how all human endeavors since the beginning of recorded history works not just programming fyi

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff



bruh

no wonder y'all act like monks in hoods filled with forbidden knowledge and regret

its like tech-workers are ants and work is just aluminium casting people into profitable shapes for rich idiots to guffaw over

this of course kills the ants

this aluminium vibe sucks, let's go back to happy posting instead

Expo70 fucked around with this message at 11:00 on Apr 6, 2023

Ruffian Price
Sep 17, 2016

Internet Janitor posted:

probably not the best way of phrasing it, admittedly

it's not so much that i think this approach will immediately peter out, but it's a kind of an ontological dead-end. llms are the most general and accessible machine learning tools to ever exist, with no objective way of measuring how effective or reliable they are. GPT-3.5 is generally considered better than 3 and worse than 4, but it's all vibes-based. problems are not reproducible, improvements are not falsifiable, and everything depends on irreplaceable datasets of dubious origin
Of course there's an objective measure

https://vicuna.lmsys.org

Ruffian Price
Sep 17, 2016

Sapozhnik posted:

21st century smartphones are not a particularly revolutionary technology and i'm tired of pretending that they were.

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

Expo70 posted:

my understanding as someone who just dinks around in unreal engine is this:

the entire field of programming is a castle made of sand that's both also rotting and on fire some how

a lot of people with money want to "make it all better", and "make it different"

what they're actually doing is trying to invent rain to put the fire out

what they don't understand is the sand is made of triethylaluminium and rain will not put the fire out, but make cool pretty and very toxic explosions

they will then claim the explosions were "on purpose" and then try to sell that some how calling it a "pivot"

they will then be shocked when their kingdom is washed away by the rain 5 to 10 years later

every 10 years, another new status condition (brittle, melty, etc) gets added to the castle made of sand


does this sound right to you?

what does this mean

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

Ruffian Price posted:

Of course there's an objective measure

https://vicuna.lmsys.org


one thing i see consistently is these things never actually test for non-semiotic reasoning

I've never met an langtran that can handle scenarios where the fail state is more common than the success state in the data-set and comes with zero explanation or equivalence

...

AI can't triforce, basically but i'm not sure if that's down to their character handling or if they genuinely can't and we have no way of knowing



Expo70 fucked around with this message at 11:16 on Apr 6, 2023

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

Powerful Two-Hander posted:

there's an absolutely insane book called (iirc) 21st century blues which features a fad for "food that isn't there" so you can eat as much as you like and not get fat but then it goes wrong and everyone gets super fat overnight or something

I think later on it involves octopus like aliens impersonating humans to have sex with them on live tv

you already can eat as much as you want and 40% of people do and get super fat. this happens today

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

fart simpson posted:

what does this mean

metaphor for software development

- every attempt to improve SD makes it worse by someone who only understands a small part of the total problem

- the worseness is sold as the product because it can be twisted by idiots into being profitable

- every idea to improve SD further fragments those who could improve it

- the total problem is indescribable without direct experience

- those with direct experience are the least able to effect meaningful change

- the rate of decay for the usefulness of direct experience is very fast

- the rate of new direct experience is slower than the rate of loss

- the main things that permit its continued existence are quality of life changes, or starting from scratch with new approaches

- this has a lot of inertia and tends not to work

for another analogy...

assuming we lived in a universe with a big crunch, you survive the universe by collapsing it un-evenly to make a singularity and then make another universe inside it and start over -- and then you continue to abstract endlessly because the root problem is inherently unsolvable from a practical standpoint

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

Powerful Two-Hander posted:

there's an absolutely insane book called (iirc) 21st century blues which features a fad for "food that isn't there" so you can eat as much as you like and not get fat but then it goes wrong and everyone gets super fat overnight or something

I think later on it involves octopus like aliens impersonating humans to have sex with them on live tv

you're doing it, peter!

i absolutely need to read this book. it sounds like a riot.

Shame Boy
Mar 2, 2010

Expo70 posted:

one thing i see consistently is these things never actually test for non-semiotic reasoning

I've never met an langtran that can handle scenarios where the fail state is more common than the success state in the data-set and comes with zero explanation or equivalence

...

AI can't triforce, basically but i'm not sure if that's down to their character handling or if they genuinely can't and we have no way of knowing





that's almost certainly due to how the tokenization works and not an actual failing of the underlying thing

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

Expo70 posted:

metaphor for software development

i dont really agree with a lot of that. software development seems like its absolutely easier in a lot of ways than it was in the olden dayes. theres been a lot of stuff thats improved it, like theres a reason people arent using gotos much anymore. some of it is probably that as development gets generally easier then people are trying to build more and more complex systems

i dont think computer programming is all that special either. we talk about it and complain about it here because this is the funy computer subforum but pretty much everything is a lot more complex than it appears from the outside to someone with little experience.

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

altho i guess it depends what you think the "problem" is. maybe i dont know what you mean by "the total problem"

Shame Boy
Mar 2, 2010

the problem is computers

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

Shame Boy posted:

the problem is computers

Powerful Two-Hander posted:

counterpoint, steel battalion:



counter-counterpoint: it lost money

i still think a twinstick with thumbsticks with the right software design mentality could out-do this within a few months.



fart simpson posted:

i dont really agree with a lot of that. software development seems like its absolutely easier in a lot of ways than it was in the olden dayes. theres been a lot of stuff thats improved it, like theres a reason people arent using gotos much anymore. some of it is probably that as development gets generally easier then people are trying to build more and more complex systems

i dont think computer programming is all that special either. we talk about it and complain about it here because this is the funy computer subforum but pretty much everything is a lot more complex than it appears from the outside to someone with little experience.

yeah but the field's crowning achievement is its ability to squander the vast improvements in hardware we've had and to cause cascading problems from short-sightedness which aren't trivial to fix

Shame Boy posted:

that's almost certainly due to how the tokenization works and not an actual failing of the underlying thing

awwwh i got so excited too. it still sucks at anything that's not immediately semiotic but the moment they plug models into other models which are mainly trained on non-semiotics that starts to become a non-issue... the idea of a langtran with aesthetic sensibilities somewhat concerns me.

i still think lemoine-syndrome and techbro religionation of ai and of course laundering are gonna be like, the big threat rather than ai itself. its not gonna be skynet that kills you but the guys who convince idiots to put skynet in places where skynet shouldn't be. ai is just radioactive dickpills again.

gonna take a break from posting for a bit cuz i got that ADD itch coming on strong

Expo70 fucked around with this message at 11:36 on Apr 6, 2023

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

Expo70 posted:

yeah but the field's crowning achievement is its ability to squander the vast improvements in hardware we've had and to cause cascading problems from short-sightedness which aren't trivial to fix

i mean, ive seen vast improvements in the convenience and workflow of my daily life. just in the past 10 years. thanks to computer and computer programming

Adbot
ADBOT LOVES YOU

Expo70
Nov 15, 2021

Can't talk now, doing
Hot Girl Stuff

fart simpson posted:

i mean, ive seen vast improvements in the convenience and workflow of my daily life. just in the past 10 years. thanks to computer and computer programming

okay... i'm talking about like twitter, facebook, etc

y'know profitable poo poo that runs in the massive gobshite that is a browser or the game engine

discord does not need to be anywhere near as hungry as it is, that's just how things happened

game engine programming is a chicken/egg race with gpu drivers trying to make you jump through hoops to fake fps counts

some how dinguses are like "the future of computing is combining these two things packed with technical rot where the features of each weaken the other"

its really dumb stuff

dang it i wanted to take a break from posting lol

Expo70 fucked around with this message at 11:40 on Apr 6, 2023

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply