THE NEW Miscellaneous Talk That Doesn't Deserve A New Thread Thread Thread (Part 2)

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

When do you think will come the day when linux becomes the #1 most popular PC OS?

[blackjacksike (Weird FNAF cat)]

I think I’ll use GNOME. If I ever need to customize, I guess I’ll just need plugins.

---

NOOO!! I can’t see highlights anymore! What happened?

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

Aren’t they the very first link in that image?

[blackjacksike (Weird FNAF cat)]

You don’t get it! I could see his tweets without logging in Twitter just by googling it. Now, I can’t!

[AnthropocenianAge (AnthropocenianAge Arthropleura wants to give you a hug!)]

And it seems Reddit is broken again for me. I can’t log in via New Reddit or Old Reddit. It is a good thing the Thrive Forum is very reliable.

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

Does BlueSky show the “tweets”?

Edit: Btw I’m in frale’s land now

[50gens]

a certain american individual managed to be even more far right than the afd. i hope nothing bad happens to anyone.

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

Sadly hoping nothing bad happens doesn’t mean it won’t happen.

[AnthropocenianAge (AnthropocenianAge Arthropleura wants to give you a hug!)]

And the converse is also true! Just because you think something good will happen does not mean it will.

[doomlightning (The Invisible Aztec Dodo God. Real Plant Enslaver, Ltd.)]

https://scitechdaily.com/fast-forwarding-evolution-ai-mimics-500-million-years-of-biology/
Can we possible to copy they home work and apply this to Thrive?

[Deathwake (i nuked zenzone and will never let him forget it)]

They’re doing novel protein synthesis without biological evolution. That crosses over with thrive… never. We aren’t simulating individual proteins in any stages. If someone wanted to make an RNA world stage, I’m sure this could be helpful.

[hhyyrylainen]

Yeah, reading that article it becomes very clear that they are just using an AI to help with coming up with new protein structures. This quote is very indicative of how much help it would be in Thrive:

Using a multimodal generative language model called ESM3, Thomas Hayes and colleagues designed and synthesized a novel bright fluorescent protein with a genetic sequence vastly different from any known fluorescent proteins. The researchers note that this achievement is comparable to ESM3 simulating 500 million years of biological evolution.

Emphasis added to the quote by me.

---

Though AI could maybe help in developing Thrive. I’m currently testing out JetBrain’s AI tools to see if they make my workflow better in the IDE.

[Deathwake (i nuked zenzone and will never let him forget it)]

oh whoa yeah I think that makes it real clear. The number they got 500 mullion from was how many years of average evolution could have changed a normal protein into this new one. Evolution isn’t always average and we don’t know what they gave it as a starting position or even how it changes them… huh. Pop-sci annoys me yet again.

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

Yeah, it could’ve taken 500 myrs… or 500 years. Who knows?

[50gens]

So you say evolution can have different speeds (which is obviously true for body shapes) but you don’t want to have any measure of change. I don’t know, therefore it doesn’t exist. Estimation is a sin

@""

We have prompted ESM3 to generate fluorescent proteins. Among the generations that we synthesized, we found a bright fluorescent protein at a far distance (58% sequence identity) from known fluorescent proteins, which we estimate is equivalent to simulating five hundred million years of evolution.

They are trying to boast about how they didn’t just tweaked the already existing protein by a little. If they didn’t use the word 500000000 years, the importance of this research wouldn’t be understood by news agencies and other scientists and engineers. It may take longer or shorter than that. Molecular clock studies exist. Deal with it.

@""

Here we show that language models trained at scale on evolutionary data can generate functional proteins that are far away from known proteins. We present ESM3, a frontier multimodal generative language model that reasons over the sequence, structure, and function of proteins.

Ever heard the phrase “evolution is dumb”? Well, neural networks aren’t. Its intelligent design. I think this is the only viable way to simulate evolution. Especially as the game gets more complicated.

Rather than making small changes to the creatures, an AI should make “good predictions” and only those changes should be tested. This can work both ways, multiple AI’s can be tested, the AI can be mutated and a good AI for Thrive can be created. When people play the game, and defeat the AI, their creations should be downloaded from Thrivepedia, and used as a training data for how to make good creations in such environments

We should have a complicated physics engine. The controls shouldn’t be complicated, press w to go forward, no matter how the legs are placed. But how fast you go, given those legs, should be calculated by the game. The ideal case is, the AI defeats humans all the time, so that people play in the easy mode rather than the hard mode and the AI needs to be dumbed down. Every generation, there are only few guesses by the AI for each species, so that the game doesn’t slow down nor there has to be a waiting screen.

[Deathwake (i nuked zenzone and will never let him forget it)]

Gradient descent is actually dumb too. It utilizes the ability to save it’s work and simulate multiple possible trajectories along with a lot of human tuning to accomplish far faster growth. basically, it’s evolution with a weak time machine and a guiding hand keeping everyone from dying. Analogize a time machine to savegames for when you die and thinking ahead a little, and that’s a thrive game. Gradient descent’s magic is backpropagation, which is basically using math (derivatives, I dunno) to determine what modifications would result in a better outcome. Basically, imagine if you died your life would be replayed with all decent mutations and your offspring would just inherit whichever one of you did best’s genes (I kinda removes the math, i dunno how to put it back, eh). The thing is, while gradient descent is faster than evolution, they weren’t using gradient descent. They train a model on some proteins then told it to make them new ones. While a neural network can implement an optimizing system like evolution or gradient descent, heuristics and simple logical probability gaming are far more common. GPT 4o rarely performs an optimization to find the greatest possible response to “why is my code broken”, it just looks for obvious breaks. So, intelligent design is fully right, better thane evolution is irrelevant.

@50gens

So you say evolution can have different speeds (which is obviously true for body shapes) but you don’t want to have any measure of change. I don’t know, therefore it doesn’t exist. Estimation is a sin

Gotta hand you this one, my anger at 500 million years was just that, an emotional impulse. I hate people treating evolution like a linear thing. It isn’t a course you go down a certain number of steps every million years. I cringed enough at star trek as a kid, pop sci doing it to makes me want to pull my hair out. However, in many cases we can find direct trends between time passed and changes to genes/proteins. That isn’t necessarily an optimizing process, it’s not like waiting a hundred million years gives you 17% more efficient proteins in all mitochondrial organelles (because they have less protection and are purely matralinial (speling), I think it actually is more consistent), you’ll maybe get around 17% difference between proteins doing the same roles on average. I still feel like I’m right, but, you are correct that the argument “measuring in evolution-years is stupid and evil” is fallacious. If I were to redo my argument I’d say “measure the competence of a final product in evolution-years is a nonsensical idea”.

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

New poll

Do you think more stages will be split off or changed into the future of Thrive’s development?

• Yes
• No

[Deathwake (i nuked zenzone and will never let him forget it)]

Probably between early and late space, but aside from that I think every other stage is a logically sized chunk

edit: unless industrial feels small, in which case glue com early space stage to industrial and and call it a day

[aah31415 (The maker of SitF, Radiostrocity, The Lifenote and TGBing; The Second Ascended...; And just maybe a security warning come alive...?)]

Is ascension a full-blown stage or a substage of space?

[doomlightning (The Invisible Aztec Dodo God. Real Plant Enslaver, Ltd.)]

Why not use the Kardashev scale as a division into stages? That would be a little more faithful to what we think of in the space stages or even the industrial stage (we are around 0.75-0.8 on this scale)