The Worm That No Computer Scientist Can Crack

[-]

TheLordB@reddit

I work in bioinformatics.

We constantly get computer scientist offering their services to save us.

Then you point them at the relevant data to what they want to work on and you never hear from them again.

Biology is hard. For most problems it requires knowledge of both biology, chemistry and computer science to even attempt to solve anything beyond toy problems. The most successful scientific research is done by specialists in all areas who have deep domain knowledge and the lab/organization has the ability to run wet lab experiments to generate additional data and test the algorithm results.

There have been some cases where the thing needed was simple enough for a computer scientist to make significant impact such as the various algorithms to align dna sequences to a reference’s or each other. But they are few and far between.

[-]

ecnecn@reddit

Seriously, they read Bioinformatics and believe the short prefix "bio-" is the easy part.

[-]

safdwark4729@reddit

I'm in a somewhat adjacent field. My experience is uh, slightly different than yours. For better or worse, everything biology and chemistry just pays worse than CS/SWE/CE. Lots of engineers come in and see these kinds of projects and think "Awesome I could help with that!" Thinking the importance of something in biochem correlates to compensation. Then they find that they get 1/6th their current wage working on something or nothing at all. This means that very few programming experts actually cross over to aid in biology, even with things that could be improved a million times over from how it's currently being done, even fewer who take the plunge decide to stay and help.

Then you have biochem math aversion (especially anything calculus, diffeq), and sheer ignorance of what is even possible in computer science. A lot of times they don't even know how to ask or articulate things in a way for something to be solvable, so it doesn't get solved even if they can consult with actual software engineers.

In contrast we don't have the same problem for physicists, who are forced to learn how to program in school even if they don't have classes for it and must understand complex math to even graduate. The problem with some physics sub disciplines is they have a hard time writing anything intelligible (horrible overloaded self conflicting equation notation that conflict with other disciplines as well), where as I can look at lots of bio papers and know what is going on with minimal background knowledge, or know how to figure it out with minimal effort, they rely way less one "code golf you are forced to read" and are more results oriented.

[-]

SimplyUnknown@reddit

Relevant XKCD

[-]

TheLordB@reddit

Randall lives near (in?) Cambridge MA which has a huge concentration of bioinformatics people (broad institute, pharma companies, and a ton of universities).

It is clear he is friends with at least one compbio/bioinformatics person because there have been quite a few compbio/bioinformatics specific comics.

I always wonder a bit when I go out to bioinformatics meetings if he will randomly be there or at least if one of these days I will meet whoever the friend is that he gets the ideas from.

(I wouldn’t fully discount the possibility that Randall has an interest in it himself and the ideas aren’t just coming from friends, but I think it is more likely the ideas come from friends who are actively employed in the field)

[-]

howtoDeleteThis@reddit

There's always one

[-]

istarian@reddit

Finding people who (a) want to cross specialize and (b) are reasonably good at both is disciplines is probably a lot harder than finding an expert in just a single field...

[-]

voronaam@reddit

At least 60 GB of free space on your machine and at least 2GB of RAM.

I understand the RAM, but what takes so much disk space for the simulation?

[-]

CrownLikeAGravestone@reddit

The data included in OpenWorm is (as near as possible) every neuron, every synapse, every ion channel, every muscle connection, and that's just the neurological parts. The entire worm's body is modeled down to the cellular level through every part of its lifecycle and they're working on adding data on how it ages as an adult too.

It's a MASSIVE amount of data. 60GB is honestly kinda impressive.

[-]

gpacaci@reddit

Transform to JSON. 6TB.

[-]

Valuable-Benefit-524@reddit

That’s bizarrely small. I don’t care if the entire thing is absolute garbage; just the spatial optimization is impressive. I had to double take.

[-]

LoweringPass@reddit

Does this include data for the rocket launcher?

[-]

NotFloppyDisck@reddit

Wild when you think of it. To be able to quantize a living thing.

[-]

CrownLikeAGravestone@reddit

It is, isn't it? They've already embodied the connectome of OpenWorm in a lego robot and it slid around doing... worm things. Makes you wonder what further things could be achieved - are we finding evidence for a computational theory of mind?

[-]

JJJSchmidt_etAl@reddit

That's like one bush in modern CoD

[-]

stult@reddit

They need more elegant C to model C. elegans efficiently

[-]

Xp_12@reddit

lmao

[-]

MAValphaWasTaken@reddit

At a certain point, maybe it's worth switching to C++. elegans?

[-]

Quexth@reddit

Total guess, disk is being used as swap space as this project sounds old and it is not and would not have been common to have RAM anywhere near 64 GB.

[-]

voronaam@reddit

That is a good explanation. It is just... with a few LLMs downloaded locally I find myself with 62Gb or RAM and only 61Gb of disk space available. This projects sounds really cool though. I might still try it out.

[-]

ILikeLiftingMachines@reddit

They're screwed because they can't simulate chemical reactions at scale and in enough detail.

[-]

Akaino@reddit

Yet

[-]

ILikeLiftingMachines@reddit

:)

Some of those scale at O(n^7)...

[-]

pleaseNoMoreFish@reddit

oh ez, it's just polynomial time.

[-]

DuckDatum@reddit

Not to be that guy, but what do you say for 50-100 years from now when I’d imagine quantum computing is more operational?

[-]

dontyougetsoupedyet@reddit

Quantum computing requires a very large number of numbers when you are trying to simulate something like a worm nervous system. Far, far too many numbers. We could have a quantum computer impossibly large, the size of planets, and never get close to simulating an organs function. When we simulate quantum things clasically we create a grid of some size, cutting out information.

I’m not convinced we will ever simulate something like a worm accurately.

[-]

CreationBlues@reddit

The worm isn’t a single coherent quantum system though.

Realistically speaking, the thing you use quantum computing for is simulating a couple thousand atoms, and using that atomic lens to scan through and understand all of the chemical processes and components that make up The Worm.

Sure, it takes millions or billions of qubits to simulate it fast with high precision, but we do that with regular electronics it’s probably doable within an order of magnitude or something to that scale.

Then you use your ability to efficiently simulate large quantum molecules to figure out a big transition matrix for all the different parts of the system and you turn the work into linear algebra.

[-]

Spare-Plum@reddit

What makes quantum computers powerful is that they can hold essentially an infinite amount of data in a single bit, qbits aren't discrete and instead cover all real numbers when they are superimposed

As a result, operations on them could theoretically process an infinite amount of data in one operation. This is why you can get crazy low runtime complexity like shor's algorithm which can do prime factorization in O(log(n)^3) compared to the much larger runtime and space required on turing machines

The problem is how precise our existing systems are, it's not that you would need a planet sized quantum computer with more qbits.

However with a theoretically perfectly accurate and precise quantum computer it could be viable to do these simulations with the right algorithm.

[-]

Kwantuum@reddit

Fiction is fiction.

[-]

ILikeLiftingMachines@reddit

:)

Some of those scale at O(n^7)...

[-]

alangcarter@reddit

Hmm... So in the near future we might have a creature (however simple) that does not know it is, and is living in, a simulation running in some (to it) cosmic horror's incomprehensibly vast computer.

[-]

Schmittfried@reddit

The horror part is your outside, anthropomorphic interpretation.

But more importantly, it’s an open question whether this project will ever be successful.

[-]

Xyzzyzzyzzy@reddit

This particular project may never be successful, true.

But if you're talking more broadly, about whether any project to faithfully simulate C. elegans can be successful - it would take significant new science to explain a "no" answer. As far as we know, matter does not contain an ineffable 'special sauce' whose properties are not computable.

[-]

Schmittfried@reddit

No, it doesn’t. We don’t even have to go into the special sauce territory to get problems in physics or even cognition that are uncomputable or so resource intensive as to be considered practically uncomputable.

Then again, the burden of proof is on the one who makes the claim. The proof that life and especially consciousness is computable is yet to be seen. Until now we‘ve only ever observed falsifying results.

I‘m not saying it isn’t possible, just that I‘m not holding my breath and that I think the current wave of optimism that AGI etc. is just around the corner is yet another wave of hubris.

As far as we know, matter does not contain an ineffable 'special sauce' whose properties are not computable.

As far as we know, awareness is pretty darn ineffable. So much that philosophers and scientists alike banged their heads against a brick wall for millennia.

The collapse of the wave function is also something physicists like to hand-wave away.

[-]

Xyzzyzzyzzy@reddit

For my extraordinary claim that the laws of physics are the same inside the human brain as they are everywhere else, that there is no known physical mechanism for the human brain to produce art that is not subject to the laws of physics, and that mathematics works the same in the presence of the Mona Lisa as it does everywhere, I submit literally every scientific and mathematical paper ever.

I consider my burden of proof met.

[-]

Successful-Money4995@reddit

Perhaps you are that creature, vastly simpler than the God who made you.

[-]

currentscurrents@reddit

Or perhaps you are the vastly complex output of a very simple optimization process.

[-]

Full-Spectral@reddit

You take the blue bacteria, the story ends, you wake up tomorrow in the garden and believe whatever you want to believe.

[-]

wiredmagazine@reddit (OP)

Stephen Larson is a cofounder of OpenWorm, an open source software effort that has been trying, since 2011, to build a computer simulation of a microscopic nematode called Caenorhabditis elegans. His goal is nothing less than a digital twin of the real worm, accurate down to the molecule. If OpenWorm can manage this, it would be the first virtual animal: the “holy grail,” as OpenWorm puts it, of systems biology.

Unfortunately, they haven’t managed it, even though scientists have been studying C. elegans for decades (in fact, no fewer than four Nobel Prizes have been awarded for work on the worm).

So why keep trying? What is it about this little worm that pulls generations of scientists towards its challenge? Well, it’s an opportunity. Understanding C. elegans is a stepping stone toward understanding more complex nervous systems and eventually, someday, the human mind.

Read the full story: https://www.wired.com/story/openworm-worm-simulator-biology-code/

[-]

opened_just_a_crack@reddit

Okay Mr pay to read

[-]

HomsarWasRight@reddit

IMHO I think subscriptions are far more ethical than modern web ads.

[-]

FullPoet@reddit

Is it ethical to pay a subscription and they still harvest your data?

Thats how the argument always is - subscription vs enormous harvesting.

The result just both, you pay them and they sell your data.

[-]

Scared_Astronaut9377@reddit

Who sells your data? The company OP cited? Not true as you can see by reading terms of service.

[-]

FullPoet@reddit

Of course they do - Wired doesnt sell articles, they sell views, which are in turn ads.

How do you think they make money?

[-]

CaptainShaky@reddit

This is such a meaningless complaint. This isn't a mobile app with too many permissions or a mailing client. It's a website where you read articles. What data exactly would they harvest ?

[-]

Exepony@reddit

By accusing every website of "data harvesting" you're making the term meaningless. It's alarmism.

I'm afraid that battle has been lost a long time ago. I'd be surprised if a single person crying "dAtA hARvEsTiNG" would be able to coherently describe what data exactly they believe is being "harvested", how it's being used, and why that is a problem.

[-]

FullPoet@reddit

Oh its easy to concretely tell you what they harvest. Its here:

https://www.facebook.com/terms/ (not included the 10 or so links to the privacy policies) https://policies.google.com/terms?hl=en-US (not included the 10 or so links to the privacy policies) https://redditinc.com/policies/user-agreement

etc. etc. Now maybe you could tell me which parts of what they collect is fine with you?

!remindme 10 years

For when you're done trawling through the inevitable legalese

[-]

RemindMeBot@reddit

I will be messaging you in 10 years on 2035-03-27 13:38:46 UTC to remind you of this link

CLICK THIS LINK to send a PM to also be reminded and to reduce spam.

^(Parent commenter can ) ^(delete this message to hide from others.)

^(Info)	^(Custom)	^(Your Reminders)	^(Feedback)

[-]

RisKQuay@reddit

I know this is a bit off topic to the OP, but for example: DuckDuckGo tracking blockers for a single app, on a single day, of one single ad company

I mean, why does Adobe need to gather all this information about me when I'm shopping for groceries? And it's not even to do with app permissions - the app has zero user granted permissions.

Why is it a problem? Because why should a company get access to all this about me? It's not even going to Tesco - it's going to a random third party. But I'm paying for my groceries with real money. Why should I also have to pay with my data?

I don't want every company on the planet to know I bought toothpaste with my bananas and lube.

[-]

FullPoet@reddit

What data exactly would they harvest ?

Everything they can get their hands on, or their ad partners. There are companies whos sole business model is matching up advertising data to effectively deanonymise people.

Its all one huge scam.

[-]

threesidedfries@reddit

Not that data harvesting isn't a problem, but isn't the argument subscription vs ads

[-]

frenchtoaster@reddit

The main concern people have with web ads these days is tracking and data harvesting. Viruses and disruptive/popup ads is less a thing these days, especially on a site like Wired.

And the point that FullPost is raising is that subscription doesn't actually escape those factors. It might be even worse tracking problem when they have your name and credit info for selling / tracking.

Seems directly relevant to whether subscription vs ads and which is worse?

[-]

FullPoet@reddit

Yes, that was my point thank you.

[-]

HomsarWasRight@reddit

All that is totally fair.

However, I was originally responding to someone complaining about “pay to read”. Since they took issue with the principle of having to pay to access content on the web I wanted to address it that way.

There’s nothing in subscription pricing that requires data tracking. But embedding ads from a third-party provider requires it by definition.

It’s absolutely true, though, that most things placed behind subscriptions (by large companies at least) do so in an immoral way by being just as bad.

I generally like to support smaller creators who I have some confidence are not monetizing their data (can’t be sure that the services they use like Patreon aren’t, though).

[-]

Randolpho@reddit

“More”? Yes. “Far more?” No.

And web ads are a really low bar, too

[-]

campbellm@reddit

Have you never heard of archive.is?

https://archive.is/jdkRz

[-]

opened_just_a_crack@reddit

No that’s very cool

[-]

stonerism@reddit

I'd like to see direct links to their articles.

[-]

tofu_ink@reddit

lol, more then likely you will have to pay the fee to get those articles. Even science articles get blocked even depending on how they got money for their article (even if it should be in the community/public for free) https://www.vox.com/the-highlight/2019/6/3/18271538/open-access-elsevier-california-sci-hub-academic-paywalls

[-]

ammonium_bot@reddit

lol, more then likely

Hi, did you mean to say "more than"?
Explanation: If you didn't mean 'more than' you might have forgotten a comma.
Sorry if I made a mistake! Please let me know if I did. Have a great day!
Statistics
^^I'm ^^a ^^bot ^^that ^^corrects ^^grammar/spelling ^^mistakes. ^^PM ^^me ^^if ^^I'm ^^wrong ^^or ^^if ^^you ^^have ^^any ^^suggestions.
^^Github
^^Reply ^^STOP ^^to ^^this ^^comment ^^to ^^stop ^^receiving ^^corrections.

[-]

Derproid@reddit

Why did we lose all the good bots but shit like this is still around?

[-]

ammonium_bot@reddit

Thank you!
Good bot count: 1200
Bad bot count: 441

[-]

vplatt@reddit

https://github.com/openworm/openworm

[-]

dismantlemars@reddit

The article isn't very science heavy, it's more snippets of interviews with people working on the project discussing the challenges, scope, and direction. It does link to this paper which proposes an optogenetic approach to mapping the behaviour of biological C. Elegans.

There's also this page on the project site that lists publications based on the OpenWorm project.

[-]

amroamroamro@reddit

scihub is your friend

[-]

amped-row@reddit

how exactly are journalists supposed to make money? Everyone is using adblock

[-]

TheDevilsAdvokaat@reddit

I recently got rid of chrome because they delisted ublock origin from their extenstions.

After a month I got curious, resinatklled chrome without the extentsion and tried it.

MY GOD it's bad.

Clickjacking, animated ads, pages that pop up advertising games or porn or crypto...

After five minutes I gave up and went back to firefox.

[-]

self@reddit

When the extension went away, I switched to uBlock Origin Lite (also by Raymond Hill). It supports filter lists, but you can't pick sections of a page to block.

[-]

TheDevilsAdvokaat@reddit

How's it working for you? Is it doing a decent job?

[-]

self@reddit

It uses the same filter lists, and I don't see ads. I haven't run into a site where I had to disable the extension to view content.

[-]

TheDevilsAdvokaat@reddit

Thanks, sounds good.

[-]

rookie-mistake@reddit

you can keep using it on chrome, you just have to check some boxes to acknowledge that its unsupported now and that you still want it active

[-]

TheDevilsAdvokaat@reddit

Didn't downvote you, sorry.

But you can't download it any more.

So as I got rid of chrome already, I can't get it.

[-]

opened_just_a_crack@reddit

All I’m saying is don’t make a Reddit post about it when the link you provide is expecting me to pay for it

[-]

gage117@reddit

If it was just a link with no summary or info provided, I'd be saying the same thing. But there's a pretty good article summary provided for free. If that's not enough, and you're interested in reading more, then pay for the service of it. That's how the news has always worked. Interesting headline, you can read the front page, but if you wanna break that paper open and read more then you have to buy the paper.

[-]

opened_just_a_crack@reddit

Information and education should not be something reserved for only those that can afford it. Unless it’s deemed as entertainment which I would gladly pay for.

But I would hope people are not posting entertainment to a programming subreddit where people are here to share and discus information regarding computer science topics.

[-]

WaitForItTheMongols@reddit

Why can't computer science be entertainment?

[-]

opened_just_a_crack@reddit

Information and education should not be something reserved for only those that can afford it. Unless it’s deemed as entertainment which I would gladly pay for.

[-]

ammonium_bot@reddit

read more then you

Hi, did you mean to say "more than"?
Explanation: If you didn't mean 'more than' you might have forgotten a comma.
Sorry if I made a mistake! Please let me know if I did. Have a great day!
Statistics
^^I'm ^^a ^^bot ^^that ^^corrects ^^grammar/spelling ^^mistakes. ^^PM ^^me ^^if ^^I'm ^^wrong ^^or ^^if ^^you ^^have ^^any ^^suggestions.
^^Github
^^Reply ^^STOP ^^to ^^this ^^comment ^^to ^^stop ^^receiving ^^corrections.

[-]

-fno-stack-protector@reddit

it's funny, i spent so many years in my 20's with almost no money, to the point i knew my pay came through at 7:17pm every tuesday, to the minute. now i'm older with a degree and i make a bagga cash, but i still find myself stuck in my old stingy ways

all of this is to say, i finally bought a subscription

(which I've tossed up for years, i know wired has a lot of their 90's content on their site)

[-]

drsjsmith@reddit

“Mother Earth Mother Board” is a classic still worth reading.

[-]

fabalaboombitch@reddit

Ads went full retard and everyone installed adblock. Ads shouldn't have gone full retard.

[-]

GuyWithLag@reddit

I don't know whether you're sarcastic or not, so apologies for the mansplaining - scientists don't see a dime of the money the journals collect.

[-]

tofu_ink@reddit

Because the smart sites only give you a blurb about the article, you need to login to download the full article. The bad sites use javascript to hide article data after the page loads

[-]

vplatt@reddit

https://github.com/openworm/openworm

[-]

TheoreticalDumbass@reddit

Okay Mr write for free

[-]

opened_just_a_crack@reddit

yeah no you are right, things that are made with the idea of profit in mind are intrinsically ethical. You right.

[-]

FairlyGoodGuy@reddit

Tip 1: Firefox's "Reader View" gets around many soft paywalls you run into -- including this one.

Tip 2: If you don't see the full article after you open it in Reader View, refresh the page.

[-]

Tjessx@reddit

Makes me think of the great tv show “devs”.

[-]

napoleon_wang@reddit

And Pantheon

[-]

bravopapa99@reddit

That went from average to dark to f* awesome by the end. I was spellbound.

[-]

safrax@reddit

The real comparison is https://en.m.wikipedia.org/wiki/Pantheon_(TV_series)

[-]

vplatt@reddit

https://github.com/openworm/openworm

[-]

Leverkaas2516@reddit

Read the full story:

If you want me to read the full story, you're going to have to post the full syotory here.

[-]

bonerfleximus@reddit

In the title

[-]

Whackles@reddit

Yes cause you work for free ?

[-]

Schmittfried@reddit

Not relevant. People are entitled to demand money for their content. Likewise, people are entitled to not consume it.

[-]

Whackles@reddit

Ah but he is not saying he won't consume it. He's asking to get it for free.

[-]

Schmittfried@reddit

They are entitled to state their conditions for consuming said content as well. Of course it’s up to OP whether they give in to the demand or keep the pay wall.

[-]

Ecstatic_Potential67@reddit

so, why still calling it a worm. call it a beetle, fgs.

[-]

tomasartuso@reddit

This article is fascinating—OpenWorm feels like the perfect intersection between biology and programming. It’s wild to think that something as “simple” as a worm still can't be fully simulated, even with all the computational power and talent we have.

It really puts into perspective how complex even the smallest forms of life are. Makes me wonder: is it just a matter of time and resources, or is there something fundamentally missing in how we model behavior?

Also curious—anyone here ever contributed to OpenWorm or a similar project?

[-]

currentscurrents@reddit

It is a matter of time and resources. There is irreducible complexity that can’t be abstracted away and must be simulated.

Just look at how hard it is to simulate folding a single protein - and a worm is made up of trillions of proteins.

[-]

red75prime@reddit

AlphaFold has found heuristics that are efficiently computable on a classical computer. It makes me think that evolution might have selected for structures that are robust to quantum noise because it's just too hard to make quantum weirdness useful (as we try to do in quantum computers).

[-]

Tura63@reddit

What an insufferable writing style

[-]

Umustbecrazy@reddit

It's wired, so I'm not expecting pulitzer level prose.

[-]

Different_Fun9763@reddit

The only two options.

[-]

ConnectomicAGI@reddit

As a co-founder of the OpenWorm project, there has been many interesting and fun off shoots of the project. I was instrumental in cataloging the nervous system and left the project back in 2014 to explore higher level animals. My worm emulation work has been written up in many places including Wire, as well as, replicated around the world and applied to many different robots. My point is don't discount their work. It has led to a number of discoveries despite the ultimate goal not having been achieved.

https://youtu.be/YWQnzylhgHc?si=z9AJ774ZrahS785t

[-]

TheDevilsAdvokaat@reddit

Yes it would be amazing. One day we will do it..and then of course make our way up the evolutionary ladder step by step.

But that first step is a doozy.