AlphaFold3 – why did Nature publish it without its code? | Hacker News Reader

AlphaFold3 – why did Nature publish it without its code?

nature.com

50 pointsluma19 comments

Kind of a...non-statement? Like "well, we strongly prefer when folks include open source, but this was neat and we want to be able to publish cool work, soooo...?"

Read between the lines:

"The private sector funds most global research and development, and many of the results of such work are not published in peer-reviewed journals."

I read this as:

Given the option of publishing pseudocode and not publishing at all, Nature is willing to accept the first.

and

Google would likely not have published if we required full transparency, and having them publish is more beneficial. (and this paper is impactful enough that it is worth making an exemption for.)

And in my opinion:

People are asking too much of Alphabet. Alphabet could easily choose not to publish anything. To be honest, I'm shocked that they do publish.

>People are asking too much of Alphabet. Alphabet could easily choose not to publish anything. To be honest, I'm shocked that they do publish.

Yeah, no. No one would be working at Alphabet if they didn't publish this work. Which means the work would never get done to begin with.

So where else would they work?

In academy for example, including the publicly funded institute mentioned in the article (European Molecular Biology Laboratory’s European Bioinformatics Institute). Also, there is a blog page on the deep mind website where they cite "millions of researcher" around the world using alphafold [1]. Some of those institutions might be happy to develop the new alphafold.

On the other hand, even in academy, scientific articles are not always the best examples of open and reproducible research [2,3]. And in computer science, it is common for articles to be more focused on algorithms than on implementations.

[1]: https://deepmind.google/impact/meet-the-scientists-using-alp...

[2]: https://en.wikipedia.org/wiki/Reproducibility

[3]: https://en.wikipedia.org/wiki/Replication_crisis

Perhaps when Alphafold was just starting, but now they are the undisputed king.

Would you rather publish or work on the cutting edge?

Look at Apple’s ML efforts which are the result of not supporting publishing cutting edge research. Or, better yet, ask Siri and you’ll get the answer.

Just like no one would be working at <innumerable other AI or pharmaceutical companies>. Or the rest of closed source Google? Yeah, no.

I'm a structural biologist and I work at an AI pharmaceutical company. Even we still publish. We publish both methodologies in peer reviewed journals, and patents.

Alphabet has to publish or they would lose a huge amount of their research staff. People in this industry want to publish and will work at the companies that allow it before they work at the companies that restrict it.

They aren't doing it out of the goodness of their heart, they're doing it because it benefits them to. If they want to have those benefits they should be held to the same standards as everyone else.

They publish because they need external validation so people are willing to pay for it and they can retain top researchers. It’s not a gift to Nature or its readers, it’s pure self interest.

Giving Google special rules flies in the face of the scientific process. Nature needs to have one, consistent set of rules, as their stamp of approval is what Google wants and they got it without giving up anything in return.

This is a shameful perversion of the peer review process and the paper should be retracted.

Also no attenpt to justify why google is allowed to blatently disregard the code policy when others are not

Except the editorial explains the policy does allow for this and references several past examples.

It's a statement for sure.

It's a statement that their so-called requirement that "Authors are required to make materials, data, code, and associated protocols promptly available to readers without undue qualifications." is actually just a suggestion.

It's a statement that their requirement that papers pass a peer review stage is just a suggestion, you don't actually have to provide information to the reviewers that would be necessary for them to review the work.

If your pockets are deep enough and/or your work is compelling enough, you can ignore those requirements. In modern society, that seems obvious, but people like to think that academia is ideologically motivated and immune to this kind of thing.

None of this is even remotely surprising to me; DeepMind (specifically Demis) and Nature have had a long-running mutually beneficial relationship.

I guess the way I think about it is, the absolutely most competitive people in a field don't actually need the exact code and data from their competitors- they have enough smart people on staff to reproduce and sometimes even exceed the competitor. That's certainly true for David Baker (Rosetta). It slows the process down, meaning Baker's group might need to spend 3-6 months reading through the paper, reproducing the ideas, seeing what works, what important details were left out (nearly every great paper I've read leaves out some bit of the special sauce), and then releasing their own papers/codes/results.

It's worth seeing this with 200+ year old eyes, for example the competition between England and Germany to be the #1 science country, with massive benefits for not sharing the exact details of technology that was developed locally. One might even speculate that this sort of behavior does globally increase the overall rate of scientific progress by stimulating intense competition instead of complacency.

> The private sector funds most global research and development, and many of the results of such work are not published in peer-reviewed journals. We at Nature think it’s important that journals engage with the private sector and work with its scientists so they can submit their research for peer review and publication … But this goal will not be achieved in a single step. It will require a process. And that requires engagement and dialogue between all stakeholders.

I can understand this line of reasoning if the researchers involved needed convincing to publish their results in peer-reviewed journals at all. But this is Nature we're talking about, and more likely than not I imagine the incentives work the other way around. As it stands, the AlphaFold team gets to use a Nature publication to lend their work the veneer of prestige and credibility, without actually being held to the same standards of openness that are demanded of us ordinary mortals. As far as I can tell the uproar is about this specific double standard, and to the extent that this editorial doesn't really address this, I think it misses the point.

Well apparently you really can bootstrap yourself in science, without building at all on the shoulders of those who came before.

Go figure.

0% chance this would fly if the exact same paper was submitted by Joe Blow & Colleagues at Stony Brook. When I initially saw that the paper was published without code, my assumption was that they'd done it for biosafety reasons. But it sounds like the real reason is "we want to make a shitzillion dollars patenting new drugs because our search business is falling apart."

Reply on news.ycombinator.com