jonny

jonny, 4 months ago to bluesky

Compare figure 3 here in the #atproto / #bluesky paper
https://bsky.social/about/bluesky-and-the-at-protocol-usable-decentralized-social-media-martin-kleppmann.pdf
To the diagram here:
https://bsky.social/about/blog/5-5-2023-federation-architecture

The paper figure is a lot cuter, but by linearizing it and presenting it as two parallel tracks they have obscured the most salient feature of the network: the big relay in the middle. Beyond "centralization bad," that pins down most of the undesirable and dangerous features of the protocol, and makes it seem like theres a lot more choice than there is.

Since the design purposefully hides the architecture: you dont know where your feed generators are drawing from, or those used by your friends. So you cant know what the effect of choosing a different relay would be, aka the main relay is always indispensable. Importantly the relays subscribe to you, you dont push to the relay, and since you arent really supposed to operate your own data store, you can be dropped from the network without knowing - the relay serves as an unaccountable point of moderation.

jonny, 3 months ago to random

In general I think that all "measuring disruptiveness of science using citation metrics" papers are hot garbage, but this one is hotter than usual.

someone said "the entire result is just a seaborn bug," which looks right
https://pubpeer.com/publications/E728CA80B5E267FA3F6C6B318BEEDA

but also if you grab the data & code, there isn't any code for actually computing the "CD5" value at all, and for the data that is present (eg. the normalized WoS table) there is a huge amount of quantization indicating they have very little data available for most papers.

if you plot just the mean number of papers whose value is -1, 0, or 1 per year (a cheap way of checking for 0 or few data points per paper) then you can basically replicate the main result of the paper too with different scaling lmao

re: https://mastodon.social/@BorisBarbour/111994672499652558

jonny, 5 months ago to random

Wondering if all this fiddling with ZFS and big box of hard drives is worth it, went to compare it to AWS elastic file storage and whelp...

I've downloaded 100TB (more, but 100 is round number) of data so far. To store that in EFS near me, that would cost $0.33/GB/month to store, $0.03/GB to read, and $0.07/GB to write. Downloading the data does a 100% write (obvi) and read (checksumming). So merely to get the data it would cost $10,000, and to park it for ONE MONTH would cost $33k.

That's more than the whole NAS for ONE MONTH. and the NAS has 500TB. Even with 25% storage efficiency from mirroring and backups, we could afford to buy a new NAS every month instead of using AWS. That's a hell of a scam.

jonny, 3 months ago (edited 3 months ago) to random

For the labs youve been in, who writes the grants for the lab? (not grants for individual people, the ones for the whole lab)

• PI only: PI does all the substantive work with figures and maybe some copy editing from lab
• PI mostly: PI does most of the substantive work but will labmates will draft significant text and figures too
• Shared within lab: PI structures and leads, but work of writing is shared with labmates as equal partners

jonny, 9 months ago to random

what if peer review wasn't something we dreaded, but a treasured chance for our friends to tell us we're wrong?

jonny, 3 months ago (edited 3 months ago) to python

I'm looking for reviewers for two packages at the moment:

Automata (@pyOpenSci )
Review: https://github.com/pyOpenSci/software-submission/issues/152
Repo: https://github.com/caleb531/automata
A #Python library for simulating finite #automata, pushdown automata, and Turing machines.

Kirstine.jl
( @joss )
Review: https://github.com/openjournals/joss-reviews/issues/6193
Repo: https://sr.ht/~lsandig/Kirstine.jl
A #Julia package for Bayesian optimal experimental design with nonlinear regression models.

You'll be working with another reviewer to read and run the code, make sure it fills a basic checklist which usually only takes a few hours, and beyond that whatever youd like to focus on. Both of these are collaborative review processes where the goal is to help these packages be usable, well documented, and maintainable for the overall health of free scientific software.

Its fun, I promise! Happy to answer questions and boosts welcome.

Edit: feel free to volunteer as a reply here, DM me, or commenting on those issues! Anyone is welcome! Some experience with the language required, but other than that I can coach you through the rest.

#PeerReview #OpenReview #CodeReview #FiniteAutomata #TuringMachines #Bayesian #Regression #Statistics

jonny, 8 months ago to random

#UAW keeps winning. I fucking love the slow drip strategy of "dont fuck with us or we'll close another shop"

https://uaw.org/uaw-wins-just-transition-general-motors/

"We were about to shut down GM’s largest money maker, in Arlington Texas. The company knew those members were ready to walk immediately. Just that threat provided a transformative win. GM has now agreed in writing to place their electric battery manufacturing work under our national master agreement.

What this will mean for our membership cannot be understated. The plan was to draw down engine and transmission plants, and permanently replace them with low-wage battery jobs. We had a different plan. And our plan is winning at GM. And we expect it to win at Ford and Stellantis as well."

jonny, 3 months ago to random

The thing about that "the papers will disappear if the journals do" article is that they wont and the only reason is piracy.

jonny, 7 months ago to random

the thing about the reliability of scientific literature is that most scientists I know a) write their own analysis code and b) do not know what software tests are.

jonny, 4 months ago to random

Everyone raise your hand if you have been screaming about how building a single API endpoint into literally every piece of software is an unimaginably dangerous surveillance vector this whole time
https://arstechnica.com/security/2024/01/ars-reader-reports-chatgpt-is-sending-him-conversations-from-unrelated-ai-users/

jonny, 5 months ago (edited 5 months ago) to random

Dear neuroscientist and programmer friends, would anyone like to review a python package for procedural generation of visual stimuli @pyOpenSci with me and @nicholdav ? You are particularly welcome if you have never done an open, generative, post-pub review in the same vein as JOSS, or dont think of yourself as a "software person" but would like to be more involved with FOSS and get some exposure to best practices in packaging, docs, and tests. I will help you out through the whole process, which is actually fun and good and not the burden normal reviews are.

Edit: ofc software ppl are also welcome, I especially welcome ppl who dont think of themselves as being in the software in-crowd bc this is a great way to learn and see how other ppl do things. Subject matter expertise is welcome but not required at all.

Lmk! The tracking issue is here:
https://github.com/pyOpenSci/software-submission/issues/150
And the reviewer guide is here:
https://www.pyopensci.org/software-peer-review/how-to/reviewer-guide.html

jonny, 4 months ago to random

why* would the us government sell its helium reserves. Thats some shit you literally cant get back because it literally goes into space, and we need it for uh lots of things.
https://www.nbcnews.com/health/health-news/us-just-sold-helium-stockpile-s-medical-world-worried-rcna134785

*of course it was some shady backroom deal

jonny, 10 months ago to random

The wildest #UniversityIT bullshit I've heard yet:

At a union meeting, talking with someone from a different department. IT revoked superuser privileges from all their work computers so they have to submit a ticket EVERY TIME they want to sudo. They said they have an outstanding ticket to install adobe reader, and in the meantime they CAN'T OPEN PDFs.

usually when I try to talk about how digital #infrastructure impacts every part of our working conditions I'm talking about larger problems like how the deinfrastructuring of our communication systems makes it so publishers become the only venue for our work, which structures our incentive systems and patterns of work from top to bottom. This is way beyond that tho, the very basic ability for information workers to interact with information at all.

jonny, 4 months ago to random

Never have I seen a Wikipedia list page in greater need of more entries
https://en.m.wikipedia.org/wiki/List_of_allegedly_cursed_objects

jonny, 3 months ago to monsterdon

what up #monsterdon, @monsterdon has got u covered if you need to stream but don't want to see any ads. i went and mirrored the colorized archive.org version which according to the comments is an actually good colorization: https://tube.me.jon-e.net/w/8EqDJYZDtSeZzA8L4d88vf

jonny, 6 months ago to random

The fact that we call any collection of documents whose location is their title a "wiki" is plain evidence of how MS Word and now Google Docs foreclosed our ability to imagine collective information systems so profoundly it left a scar over the place in our vocabulary where we would describe them

jonny, 11 months ago to fediverse

has any masto instance set up a parallel #Lemmy instance where you can share your masto account and basically have two views on the same posts? I think that would be super interesting and basically how fedi should work anyway imo. I would love to set that up here, but don't have the time to dig into the internals to figure out how to set it up ATM. #FediDev #FediAdmin

boosts welcome, or links to previous conversations bc I'm sure other instances are thinking similarly

jonny, 9 months ago to random

Alright my dogs what is the best way to serialize multidimensional array data for cross-language use? Is there any kinda standard or am I naïve.

I want to be able to translate between arrays stored in hdf5, relational databases, and binary blobs in content hash graphs. Is that too much to ask?

jonny, 3 months ago (edited 3 months ago) to random

One thing that sucks about #PeerReview being so broken and a vector of domination rather that cooperation is that, in the best case, they can be skillshares as much as anything else. In some code reviews I have given and received, I have taught and learned how to do things that I or the other person wished they knew how to do, but didnt.

That literally cant happen in the traditional model of review, where reviews are strict, terse, and noninteractive. Traditional review also happens way too late, when all the projected work is done. Collaborative, open, early review literally inverts the dreaded "damn reviewers want us to do infinity more experiments" dynamic. Instead, wouldnt it be lovely if during or even before you do an experiment, having a designated person to be like "hey have you thought about doing it this way? If not i can show you how"

The adversarial system forces you into a position where you have to defend your approach as The Correct One and any change in your Genius Tier experimental design must be only to validate the basic findings of the original design. Reviewers cannot be considered as collaborators, and thus have little incentive to review with any other spirit than "gatekeeper of science."

If instead we adopted some lessons from open source and thought of some parts of reviews as "pull requests" - where fixing a bug is somewhat the responsibility of the person who thinks it should be done differently, but then they also get credit for that work in the same way that the original authors do, we could
a) share techniques and knowledge between labs in a more systematic way,
b) have better outcomes from moving beyond the sole genius model of science,
c) avoid a ton of experimental waste from either unnecessary extra experiments or improperly done original experiments,
d) build a system of reviewing that actually rewards reviewers for being collegial and cooperative

edit: to be super clear here i know i am not saying anything new, just reflecting on it as i am doing an open review

jonny, 18 days ago to random

This makes total sense given microsofts surveillance strengths. At first I was thinking "but why images then, why not just a key and mouse logger and patch into the core apps like office and browser that probably dominate most of the economic behavior they'd be modeling," but then as with LLMs being used for all text tasks no matter whether natural language is relevant, I had a laugh thinking the tech for image parsing is similarly overdeveloped, and how the contemporary "AI" approach is basically to recapitulate human sensory modalities as the janky ass interfaces to the world that they are rather than directly use symbolic logic or adapt to specific domains.

I think thats sort of cute, in a "human are space orcs" kind of way. Instead of teaching the computers to think, we taught them to see and talk, and whatever happens in the middle there is less relevant.

https://aus.social/@KathyReid/112484952397296289

jonny, 3 days ago to random

The cool thing about cracked adobe products is that blocking their access to the internet is usually part of the crack. Piracy and privacy are besties

jonny, 18 days ago to DuckDuckGo

Wow idk when they did this... I went to search (with google!?!?) to see if #duckduckgo had a blog or something for news about their outage, and ofc they just have a twitter page. But if you look at it from the browser while not logged in you dont see their most recent posts, but posts ranked mostly by number of likes but also in random order.

https://x.com/DuckDuckGo

I know we are all saying this but how does the internet keep getting worse every day

jonny, 21 days ago to monsterdon

Root post for #monsterdon u know tha deal

here's an ad-free stream courtesy of @monsterdon
https://tube.me.jon-e.net/w/tEMLRMtXrXKRDbJvVP7uaw