Dan Goodman

https://neuromatch.social/@neuralreckoning

I'm a computational neuroscientist and science reformer. I'm based at Imperial College London. I like to build things and organisations, including the Brian spiking neural network simulator, Neuromatch and the SNUFA spiking neural network community.

Website: https://neural-reckoning.org
Youtube: https://youtube.com/@neuralreckoning
Brian: https://briansimulator.org
Neuromatch: https://neuromatch.io
SNUFA: https://snufa.net
Photography: https://adobe.ly/3On5B29

Public Key

npub14c5eddu9ddulgvgh8zrk9mz954tw32zymd4h88aj27f0g8lddy3sa2rfvf

NIP-05 Address

neuralreckoning_at_neuromatch.social@momostr.pink

Profile Code

nprofile1qqs2u2vkk7zkk705xytn3pmza3z624hg4pzdk6mnn7e90yh5rlkkjgcpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsg8r7rk

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub14c5eddu9ddulgvgh8zrk9mz954tw32zymd4h88aj27f0g8lddy3sa2rfvf

Show more details

Last Notes

2026-03-13 12:35:04 UTC

- reply

recently I started doing the first slide of my talk titled "TL;DR" where I try to tell the essential points of the whole talk in one slide, and telling the audience that they're welcome to go to sleep after this one slide if they want to. Quite a few people have told me they really like it, both those who did subsequently go to sleep, and those who found it useful for framing the rest of the talk. Doing social media threads has actually been really helpful for me, it's shown me that you can usually get the essential message over in very few words.

2026-02-26 16:07:33 UTC

London #neuroscience people you may like this. We're hosting a series of talks at Imperial & Crick on how to get experiment and theory working together better. Each session will have a talk around this and extended networking / group discussion on the questions raised. Plus, free food!
I think the time is right for us to do this seriously. The growth of "NeuroAI", large scale experimental projects and a steady stream of papers showing that earlier papers in both experiment and theory made overly strong conclusions because they didn't do this right.
https://www.eventbrite.co.uk/e/understanding-the-brain-in-theory-and-practice-tickets-1983856488160?aff=oddtdtcreator
https://media.neuromatch.social/media_attachments/files/116/137/422/648/447/513/original/43f8c5e5ed1c31e6.png

2026-02-04 09:40:06 UTC

Nice article but the social dynamics of science mean that peer review is de facto a "throughout problem". LLM usage in PR will increase because it reduces pressures on overstretched researchers. We can't stop LLM PR without addressing those pressures.
https://www.timeshighereducation.com/opinion/ai-not-peer-so-it-cant-do-peer-review
Career pressures on scientists mean we don't have time to engage with PR but feel an obligation to editors (either personally or because we're worried they'll reject our own submissions if we don't). Banning LLMs but leaving the pressures in place will just lead to concealment.
I agree with the article that the good thing about PR is not that it's an error correcting mechanism. It's actually very bad at this. Instead, it's a mechanism to get peers to engage deeply with each others' work. We can get this benefit and reduce pressure by making PR optional and post publication.
As a footnote, it's fascinating and telling how the staunchest "defenders" of peer review (journals) show the most contempt for its actual value by their willingness to flirt with the normalisation of LLM "peer" review. It's almost as if they had some other motivation. 🤷‍♂️

2026-01-29 17:35:33 UTC

- reply

it has to do with degree of decision making power. This is somewhat but not entirely correlated with age.

2026-01-29 16:51:27 UTC

- reply

everyone who has decision making power over someone else's salary or future career prospects does it, even if they don't admit that to themselves.

2026-01-29 14:14:14 UTC

- reply

maybe the thing that is unclear is the word "directing research". Nobody should be doing that. You should do research, or advise others who are doing research, but not tell others what research to do.

2026-01-29 14:09:21 UTC

- reply

I'm not trying to kick people out of any community. I said that. I just don't want any one class of scientists deciding what work should be done.

2026-01-29 10:51:08 UTC

- reply

I think people should continue to be paid to work if they want to and they are able. I do think it'd important that we end the system of senior researchers directing research and instead move to one where they're considered valuable advisors to autonomous researchers. As long as the advice doesn't have to be followed. At the moment, the most senior scientists are the ones I'd be least likely to go to for advice because if I had a good research idea they'd be the ones in the best possible position to put their team on it and publish first, and with the most incentive to do so. This is the opposite of how it should be.

2026-01-28 20:12:31 UTC

Interested in a fellowship in London? Come and join me in my department. The department is EE but research topics can be very wide: happy to host people with any overlapping interests in NeuroAI/SNNs/etc. More info on available fellowships and timelines here: https://www.imperial.ac.uk/electrical-engineering/research/fellowships/

2026-01-26 16:11:33 UTC

- reply

he's not wrong about the bloated websites full of ads, although that's coming to ChatGPT now too as I understand it.

2026-01-26 10:31:31 UTC

RE: https://wetdry.world/@lucydev/115912618642468067
I love this. When I compare the world my children are growing up in to the one I grew up in, the really striking thing is that essentially all knowledge is close to freely available to them and pretty easy to find. A teenager today could use freely available tools to do cutting edge research, for example. Big monopolistic AI companies would love to put an enclosure around that freedom, control it and make us pay for it. Let's not let them. (This is not a rant against AI btw, it's a rant against capitalism.)

2026-01-11 15:13:05 UTC

Story about a site trying to poison LLMs. Framed as by activist industry insiders. But, if it was run by one of the LLM makers to poison their competition, that would give them a big advantage. Maybe all the LLM companies are doing this to each other already?
https://www.theregister.com/2026/01/11/industry_insiders_seek_to_poison/

2025-11-27 15:48:53 UTC

- reply

yes I'm so glad to see it! So far most eLife papers just looked like normal journal papers, but this shows what the new model can do. Brilliant stuff.

2025-11-27 15:30:50 UTC

- reply

follow this thread because the commentary is almost as interesting as the paper (which also looks very cool btw):
https://bsky.app/profile/behrenstimb.bsky.social/post/3m6i6v3ydf22n

2025-11-27 13:34:55 UTC

From my bsky feed - two consecutive posts. Nature Sci Rep publishes incoherent AI slop. eLife publishes a paper which the reviewers didn't agree with, making all the comments and responses public with thoughtful commentary. One of these journals got delisted by Web of Science for quality concerns from not doing peer review. Guess which one?
https://media.neuromatch.social/media_attachments/files/115/621/914/689/580/164/original/e3794a800978ac41.png

2025-11-12 09:37:40 UTC

- reply

oh wow that's much higher than I remembered!

2025-11-12 09:21:26 UTC

- reply

I don't know how it's calculated but I wonder if there is a time scale difference too. I'd guess a lot of science's payoff comes many years after the work is done.

2025-11-12 09:09:28 UTC

- reply

that's more than the return on science isn't it? 🤯

2025-11-11 10:09:58 UTC

- reply

I'd love to believe it but what's the evidence?

2025-09-04 15:05:55 UTC

Is anarchist science possible?
As an experiment, we got together a large group of computational neuroscientists from around the world to work on a single project without top down direction. Read on to find out what happened.
The project started as a tutorial on a new technique at the @npub1arn…chmc 2022. We realised that the technique was easy and cheap for anyone to use with a lot of low hanging fruit.
https://neural-reckoning.github.io/cosyne-tutorial-2022/
At the tutorial we announced a 1-2 year open research project that anyone could join, starting from the materials of the tutorial, and a few starting questions we found interesting, but with no other constraints. We were inspired by the Polymath Project in mathematics.
https://en.wikipedia.org/wiki/Polymath_Project
31 people contributed to the project, joining for monthly meetings to discuss progress. All code was publicly available throughout, and when we started writing up the work in progress was also fully public. You can see that version here: https://comob-project.github.io/snn-sound-localization/paper
Not everyone who was involved made it to the paper (didn't respond or couldn't find contact details), and not all are on Mastodon, but authors include: @npub14qw…hy8f @npub13y3…qzgx
@npub17cj…40f5 @npub1zjy…vapl
We used GitHub and Jupyter notebooks to coordinate development, with a website showing everyone's current code and results to make collaboration easier. We used @npub1v2y…q6qn and GitHub actions to automate this.
https://comob-project.github.io/snn-sound-localization/
So how did it work out? Well, some things went well and others not so well. We published a paper with our results and reflections on the process. If you're interested in spiking neural networks, sound localisation, or anarchist science, check it out:
https://www.eneuro.org/content/12/7/ENEURO.0383-24.2025
Generally, the infrastructure we built worked well, as did the monthly meetings. Starting from the tutorial was a good decision because it gave everyone a common reference and meant they could easily get started.
However, the lack of direction meant that we didn't achieve very coherent results in the end. We don't think this is a catastrophic problem, but when we try again, this is something we'd like to address. If you have thoughts or would like to be involved, get in touch!
Ultimately, we didn't achieve a scientific breakthrough in this project, but we did show that without top down direction or any specific funding, we could organise a large group of scientists to work together and publish their research in a good journal. We think that's a hopeful sign for the future!
#neuroscience #computationalneuroscience #compneuro #anarchism #science
https://media.neuromatch.social/media_attachments/files/115/146/646/292/439/753/original/3c29d7fcc94d4c9f.jpg

2025-08-05 13:24:17 UTC

I hear Mastodon likes rats, so here's one that tried to steal my cake while I was sitting in a park waiting for a train in Paris.
https://media.neuromatch.social/media_attachments/files/114/976/372/503/289/004/original/c0334bd96683ea4c.jpeg

2025-07-30 14:09:15 UTC

- reply

it would be nice if we hadn't let our universities be run by malign incompetents. Ah well.

2025-07-30 14:08:12 UTC

- reply

I'm really sorry you're personally affected by this utter nonsense. What an absolute joke to tie salary to REF. It's almost a case study in the wrong thing to do given the low test retest reliability of review scores.

2025-07-25 13:58:35 UTC

We are hiring a postdoc. It's a broadly scoped position but I think it would be of interest to someone in #Neuromorphic or #SpikingNeuralNetworks. See ad below. Inquiries to Dan Akarca because I will be on holiday.
Note, the application deadline is very soon! (Unavoidable admin issues.)
https://www.imperial.ac.uk/jobs/search-jobs/description/index.php?jobId=24879&jobTitle=Research+Associate*+in+NeuroAI%2C+Neuromorphic+Systems%2C+Hardware-Software+Co-Design

2025-07-22 09:44:29 UTC

Yeah, let's give an American company access to loads of government data and pay them a load of money to run the government for us using software that systematically fabricates false information and has nasty racial biases. That'll surely stimulate the UK economy.
https://www.bbc.co.uk/news/articles/czdv68gejm7o

2025-07-04 15:28:30 UTC

Arrgghhhh using Word to edit a file that's in a OneDrive folder is now virtually unusable. It occasionally just randomly selects a section of text and declares it conflicts with another version, or worse makes an entire paragraph you've written disappear with no way to get it back.

2025-06-04 09:27:39 UTC

The tipping point for me deciding that we should abolish scientific journals was my first experiences as an editor. It made me feel both scientifically and morally uncomfortable to be making what I felt were necessarily ill informed decisions that affected scientific progress and people's careers.
I don't think it's possible to be knowledgeable enough to quickly read a paper and understand its potential significance. But the entire system rests on this foundation.
I could also see that if I kept doing it, I'd gain confidence and it would start to feel normal, even though it was still wrong.
So instead I decided to quit those roles, and campaign for an end to scientific journals and help build new systems without the many flaws of journals.
I do believe in the value of scientific expertise, of course, but I don't think we can predict what will turn out to be important or valuable in the future. Too much of scientific culture and our institutions are based on what seems to me the palpably false assumption that we can.

2025-05-17 23:34:11 UTC

Almost summer... #photography
https://media.neuromatch.social/media_attachments/files/114/525/784/619/703/502/original/302fa7e510066d45.jpg
https://media.neuromatch.social/media_attachments/files/114/525/784/715/870/819/original/c843f9af4d3fff9e.jpg
https://media.neuromatch.social/media_attachments/files/114/525/784/735/830/602/original/4f35bd2563fc5d6c.jpg
https://media.neuromatch.social/media_attachments/files/114/525/784/750/210/176/original/ab6f1ab69c585b1a.jpg

2025-05-12 14:29:48 UTC

- reply

difficult to imagine right now!

2025-05-12 14:26:29 UTC

- reply

oh fascinating, I'd never heard of that! Puts an entirely different spin on the usual story we're told.

2025-05-11 16:57:10 UTC

#SilentSunday
https://media.neuromatch.social/media_attachments/files/114/490/252/417/430/223/original/99fe886bb43a8675.jpg

2025-04-19 22:55:59 UTC

#SilentSunday
https://media.neuromatch.social/media_attachments/files/114/367/087/224/284/830/original/44c85cdb7e46c2ca.jpeg

2025-04-14 19:00:38 UTC

It's worth reading the letter of demands that the US government sent to Harvard. Really chilling. Fortunately, Harvard has said no. Hopefully other universities will join them. Any university president that agrees to this should never be allowed power at a university
https://www.harvard.edu/research-funding/wp-content/uploads/sites/16/2025/04/Letter-Sent-to-Harvard-2025-04-11.pdf

2025-03-30 17:23:52 UTC

#SilentSunday
https://media.neuromatch.social/media_attachments/files/114/252/540/124/496/515/original/a911c4a51722db88.jpeg

2025-03-29 13:45:53 UTC

Paris railway station would fit well in a future Game of Thrones. "The iron way of the north."
https://media.neuromatch.social/media_attachments/files/114/246/020/339/600/895/original/2fd54b83a1c95e7b.jpeg

2025-03-25 21:10:29 UTC

What is the point of a scientific society that won't stand up against an attack on science by an incipient fascist?
https://www.theguardian.com/technology/2025/mar/25/royal-society-decides-not-to-take-disciplinary-action-against-elon-musk

2025-03-19 10:30:06 UTC

My institution is now blocking access to the Exchange server that hosts my work calendars to any app except Outlook. This is an organisational nightmare given that I also have Google calendars and I want to see events in the same place. Anyone know a workaround without sharing access to your data?
Currently thinking about running an emulated android tablet because apps on Android can put their calendars into a global one. Seems slightly insane but can't think of another way.

2025-03-02 14:13:19 UTC

#SilentSunday
https://media.neuromatch.social/media_attachments/files/114/093/240/350/835/817/original/ead26de59df52e7c.jpeg

2025-02-20 17:23:20 UTC

I wrote a thing for @npub17rv…w9lk. The attack on scientific infrastructure happening in the US shows that relying on any one country is not a good option for science. We need to start supporting and building international, decentralised infrastructure for science.
https://www.thetransmitter.org/policy/science-must-step-away-from-nationally-managed-infrastructure/
#science

2025-02-18 00:34:26 UTC

We scientists would love to be left alone to do our work, but powerful people won't give us that option. Science, as the pursuit of truth, however imperfect, is fundamentally challenging to power. If we want to keep the option to do the science we want, we have to work at making science resilient.
Yes it's uncomfortable for many to think about science as political. Yes, the thought of building and maintaining our own infrastructure for data, papers and code might be mind numbingly tedious and scarily outside our skill set, but nobody we can trust in the long term will do it for us.
We have to stop wishing that we could go back to how things were before because that way of doing things is what got us in this mess in the first place. Let's look to the future and build our own way of doing science.
#science #politics

2025-02-12 10:20:58 UTC

UK friends. Reform is now polling consistently ahead of Labour. Membership is growing rapidly at the same time Labour membership is falling, and may overtake it within months. If we don't do something to get the Labour party to change course, in four years we'll be in the same place as the US.
If you have any pull in the Labour party please urge those higher up not to continue austerity policies and Reform-lite anti immigrant posturing. It's a disaster in the making. The rest of us need to make it clear we won't vote for a Labour party that continues like this.
#politics #ukpol #ukpolitics

2025-01-29 10:07:53 UTC

From what I've seen DeepSeek is more efficient though engineering tricks rather than better ML? Doesn't this mean it's not as disruptive as thought? Big players can just use the same tricks now and still have advantage from bigger compute?

2025-01-27 22:32:36 UTC

- reply

right but that's the technology development I mentioned in the OP, right? We can only do pangenomes because of rapid sequencing tech?

2025-01-27 18:08:14 UTC

Thinking about big science, the human genome project looms large. I'm not an expert so: did this project fulfil its potential? From what I've seen, it seems the major achievement was development of sequencing tech and a focus on open data. Otherwise, people still talking in terms of potential value?
Apologies in advance if this comes across as ignorant, I really don't know much about genetics.

2025-01-20 01:30:17 UTC

Wow! The much touted FrontierMath dataset was secretly funded by OpenAI who had privileged access to it.
https://techcrunch.com/2025/01/19/ai-benchmarking-organization-criticized-for-waiting-to-disclose-funding-from-openai/

2025-01-14 13:11:50 UTC

Mastodon etiquette question, what can you do for alt text for an image like in the post below which is showing data from loads of graphs? I don't think I can do it justice in the alt text other than just repeating the summary in the main text, but this still seems unfair in a way? https://neuromatch.social/@neuralreckoning/113826859929010069

2024-12-18 14:55:31 UTC

There's a movement in neuroscience suggesting we should be pursuing bigger bets with larger teams. I think there's a case for doing a bit of this, but I think it's a bad idea to prioritise it for two reasons, and a good case for saying we should be moving in the exact opposite direction.
Bigger bets generally means less money available for the smaller bets. This means fewer ideas being pursued. Similarly, larger teams means fewer people in leadership roles with ownership over their research direction. Again, fewer ideas and less diversity.
This is bad for neuro because most ideas have turned out to be wrong in the sense that they haven't moved us closer to a global understanding of how our brain's work. (By this I don't mean that they were bad ideas, or that they weren't worth pursuing!) This is maybe controversial so let me explain.
When I read papers from 50 or 100 years ago speculating about how the brain might work, I don't see a huge difference with how we talk about it now. Few (none?) of the big debates have been settled: innate versus learned, spikes versus rates, behaviourism versus cognitivism, ...
The second reason big bets/teams shouldn't be our focus is that it limits opportunities for the autonomy and development of less senior researchers. Let's be honest, if we start pursuing 'big bets' those bets will be the bets of the most senior scientists, not the junior ones.
Junior scientists will - to an even greater extent than currently - be relegated to implementing the ideas of senior scientists. That's bad for diversity of ideas, but it's also bad for developing a new generation of thinkers who might be able to break us out of our (proven unsuccessful) patterns.
That's why I think we should go in the opposite direction. Empower junior scientists. Don't make junior PIs beg for grant funding in competitions decided by the opinions of senior scientists. Don't make them postdocs subordinate to their supervisors.
Instead, fund junior scientists directly and independently. Let senior scientists compete with each other to work with these talented younger scientists. Give senior scientists a non-binding advisory role to help develop the talents of the junior scientists.
If we think larger teams might be important, let's find ways to empower groups of secure and independent scientists to work together positively and voluntarily - because it makes their work more impactful - rather than negatively and through necessity - because they need to get a job.
#neuroscience

2024-11-24 01:21:54 UTC

My statement on Bluesky. I'm there but I'll also be staying here. I get something very different from the two places. Bluesky is indeed like old Twitter, which is both the good and the bad thing about it. There's an infectious energy about the place, but also I can see already that it has the same hierarchical top down energy that old Twitter had. I've also already started seeing the sort of engagement farming outrage/controversy posting that made Twitter pretty tedious at times, which Mastodon feels much freer from. Mastodon also feels younger and more egalitarian
That makes Bluesky more useful for career reasons, and Mastodon just a nicer place to hang out.

2024-10-11 19:03:36 UTC

- reply

this is the right way!

2024-10-09 18:28:22 UTC

The problem with the Nobel prize is not just that they've again awarded not a single woman, although they should certainly be ashamed of themselves. The problem is our deference to these fossilised patriarchal elitist institutions, reflecting broader issues in scientific culture.
It's been depressing this week too see all the excitement and respect given to this prize. It shouldn't be a surprise to us that the bequest of a wealthy weapons manufacturer, managed by a monarchic institution, makes decisions that are out of touch with our values.
Scientists should be beacons of independent and critical thought, not have their heads turned by the interests of those with wealth or status.

2024-10-03 08:36:46 UTC

I'm writing a detailed guide to how to run a successful online workshop at no cost and low effort, with some software to automate stuff. What's the best way to make it available so that it's widely seen? Blog? Paper? Which journal? Something else? Want to encourage people to switch to doing more online and less in person conferences by making it easy to do.

2024-09-14 09:47:50 UTC

Some academics intent on running to Bluesky when there's a technically better option owned and run by their own community. Makes me think publishing reform is doomed: we'd just end up paying a fortune to the first company that comes along to sell us something we already have.
#PreachingToTheChoir

2024-09-06 05:50:57 UTC

Hopefully that's it for Macron and his party. It was always obvious that it was a Thatcherite attempt to destroy the left and the idea of solidarity in France. The fact that they're willing to ally with the fascists to keep out a democratically elected left should be an action they can never recover from.

2024-06-30 18:39:33 UTC

- reply

this is my problem with traditional peer review, even if done well. It doesn't tell you that the result is safe to build on. Firstly, studies have shown that peer review misses a lot of problems, so a paper that has passed peer review is only a little less likely to have problems than an unreviewed one. Second, often it wouldn't have been possible to realise the problems with a study at the time of review. I know of papers where everyone "in the know" is aware of the issues that only became known once people tried to build on it, and they warn their trainees about it but there's no trace of that in the literature and people go on building on it and wasting time and resources. Third, I think you only have a chance of getting a thorough review from people with skin in the game (ie not disinterested) because a lot of issues are subtle and require a lot of knowledge that would only have gained by people who work on those problems and therefore likely have a favoured theory.

2024-06-29 22:21:51 UTC

- reply

thanks for that, it's very clear now. 👍
So what's your reasoning for saying that disinterested evaluation is better than comments from someone with an interest? That runs counter to my intuition. I can see that from the point of view of an editor who wants safe content for their product it might make sense, but from the point of view as a scientist who wants to be challenged, I want to hear precisely the views of those who are passionate about it. Probably the most interesting paper that will actually change things is going to be one that provokes intense disagreement from readers with some hating it and some loving it. Such a paper is unlikely to pass a filter of a disinterested observer.
And I want all this highly interested discussion to be open so that I can actually read the reasons, and also so that I can decide for myself whose views and reviews I trust, and who to turn to for certain sorts of commentary. And I don't want any of this codified into metrics that determine people's career success chances because that distorts everything and encourages evasion and even fraud.
I don't think what I'm suggesting has been tried, so it's an experiment. But actually, I don't think what you'd like has been tried either, or at least not much and not until recently. We should do both. Actually that's all I really want, a level playing field for different ways of doing science, and not a monoculture led by commercial interests that in my view seriously distorts science and slows progress.

2024-06-29 19:46:01 UTC

- reply

wait I'm confused. You're not arguing for peer review as we do it now where the outcome is only binary? What is true peer review? Is it just peer review where the reviews are published? What makes this the true thing that is so important for science, given that it's a recent innovation and not in widespread use?

2024-06-29 19:11:54 UTC

- reply

so the main function of peer review for you is to give a yes/no answer to whether or not the work is correct for people outside of the field? And this information is encoded by the fact that it's been published in a journal?

2024-06-29 15:53:25 UTC

- reply

it seems we have a very different idea of what peer review is good for.

2024-06-29 13:44:23 UTC

- reply

pubpeer works rather well I think? Especially given it's a third party system layered over an uncooperative journal system. If comments made on pubpeer or another such system were prominently displayed whenever accessing an article, usage would go up hugely. That's the key. If there's no value in providing commentary because nobody looks at it, of course people don't do it. I suspect that's a big part of why there's a crisis in the current system. Reviewers don't find it a valuable way to do peer review. Here's my suggestion: give authors an easy way to recognise valuable peer reviews that improved a paper by giving them the option to add them as authors. There are probably some kinks to work out in that idea but I'd guess it would lead to much more active, engaged and constructive peer review.

2024-06-29 12:31:16 UTC

- reply

it should be less work, not more.

2024-06-29 12:24:39 UTC

- reply

as far as I'm concerned this only works if review is post publication and not carried out by journals but by actual readers. Your problem shouldn't arise in such a system.