dangerwhale.bsky.social
Exciting new ways in which you're doing it wrong. Editor-in-Chief, Evidence-Based Toxicology. https://linktr.ee/paulwhaley
227 posts
95 followers
133 following
Regular Contributor
Active Commenter
comment in response to
post
A Springer journal should be running Ithenticate or something equivalent, and therefore should already have this information to hand in their editorial workflow. It says here "Diffchecker, Turnitin (if you have access), or Copyleaks" are also good options. www.ithenticate.com
comment in response to
post
I just meant my meeting notes (run of the mill, "here is what we discussed here are the actions" notes), not all meeting notes, and especially not medical notes. I am really not trying to defend LLMs as some magic brilliant thing, all I am saying is evaluate performance against a task.
comment in response to
post
But I wanted a stupid AI generated whale, that was the point. I'm not quite sure why everyone is getting so cross about me saying some specific uses of LLMs are actually quite promising and it is important to differentiate the good from the bad, but I suppose this is the internet. 🤷
comment in response to
post
Dude, it's just a dumb AI generated profile pic. I thought as a Whaley it would be cute to have a stupid AI whale. I used to have my cat in a box.
comment in response to
post
Ian Dunt does not look like that. I've been listening to this podcast for what 3 years and never seen him in my life, but I can tell you that face is all wrong for that voice.
comment in response to
post
Also, if this was a relatively early version (like, maybe a year or so ago, and was in Teams) then Sir, I very much agree with you! :)))
comment in response to
post
Mine is bloody fantastic, though - as I said, some hammers are bad hammers. I've also seen terrible meeting notes from bad products, but I can tell you my tool is better at it than I am. I've had very good experiences with LLM transcription models as well, they are a huge leap forward.
comment in response to
post
That is fair enough. I'm just arguing against blind rejection (or adulation for) tools absent an understanding of what the tool is for. Also, the task can be bad - like hammers are pretty good for bashing in heads, but do we want people to do that? Twitter is good for right-wing propagandising, etc.
comment in response to
post
I'm an editor so don't have any friends, but I found this? howbout.app/blog/making-...
comment in response to
post
P.S. I am very optimistic about what LLMs can do to help editors with certain very specific tasks that are basically impossible for humans to do in a timely way - such as check the appropriateness of cited materials for claims in a manuscript. Fact checking, though? No, but not designed for it.
comment in response to
post
3/2 And because some hammers are not very good hammers, it doesn't mean hammers in general are not useful. Some LLMs just don't do a very good job of even the things they are designed for. This does not mean all LLMs suck. That a hammer cannot trowel does not mean all hammers suck on principle.
comment in response to
post
2/2 - It makes no sense to say a hammer does or does not work. A hammer works for bashing in nails, it is rubbish for trowelling mortar. LLMs are very good at some tasks (e.g. summarising meeting transcripts into notes) and very bad at others (e.g. discerning facts). ok 3/2 coming woops
comment in response to
post
Behind a lot of the nonsense, I think, is a general failure to define the task. LLMs are a tool - like all tools, they do certain types of things better than others, but unless the task is defined the statement "this does / does not work" is meaningless. For example... 1/2
comment in response to
post
Can I say "focused questions with comparable methods"? I'm trying to agree with you, I think... And I'm not trying to defend MA in psych, I have no clue what the d is going on over there. :)))
comment in response to
post
Absolutely, I love a good scoping review, me. 😛 (If I can find one.) As you say, MA will just show convergence on focused questions that have been the subject of several studies, which I believe is a rarity in psychology? It is rare enough in my field.
comment in response to
post
Apropos not much (I just saw this for the first time, hi again Katie btw!) I read recently that what science needs is convergence not consensus (meta-analysis helps measure that?) and consensus is about people's relationships with each other not with data. Struck a chord with me. That is all! :)
comment in response to
post
Also faith policing is very annoying I agree. I didn't really intend that but I'll leave it up as a demo of my occasional fallibility.
comment in response to
post
I agree with you that it's good that Clarivate are highlighting a salient issue (I set up a journal to Achieve Stuff on research integrity) but I guess I generally view the structural issues here as being things actors like Clarivate don't really help with.
comment in response to
post
Christ on a bike. This was just a couple of tweets down from this, and I don't know which is the worst thing I've seen today.
comment in response to
post
Truth also needs to be deployed in good faith.
comment in response to
post
I mean clamping down my arse. Finally noticing, maybe. And if integrity is what they are all about, why delist eLife instead of changing their bloody indexing policies? <deletes swearing because he is a grown-up> I'm just profoundly unconvinced.
comment in response to
post
Except that when you are a publisher or research indexer, (a) point of publishing is the exact and only place you can intervene, (b) focusing upstream as if you are powerless to do anything diverts attention from the very real consequences of your point-of-publishing policies. So bah humbug to this.
comment in response to
post
As a sole trader, I am asking you to JUST LET ME PAY TAX ON PRIOR INCOME
comment in response to
post
It's tax on account with variable income that is a complete nightmare - one never really knows how much to save, and if one is expecting to earn more money in the next period (hurrah!) then the amount of income that has to be set aside to cover payment on account can be very difficult to manage.
comment in response to
post
Hmm, reading the thread below it looks like something like FreeAgent will work pretty well as cheap accounts software (that will handle foreign currency, watch out for that). I use Zoho Books which is fucking horrible but I'm locked in now. :)))
comment in response to
post
I've had an accountant for years even on v low income bc the few hundred quid for a tax return is to me worth it. I've also found that paying for accounts software removes ball-ache. So on balance, yes a PITA in the short-term but it'll work out OK because it ahem incentivises good practice?
comment in response to
post
It's OK, I know where my coat is, I can go get it myself.
comment in response to
post
What I want to know is, does exercise make theory stronger? 🤔
comment in response to
post
It makes me think that, if these are the resources required to publish effectively, then we should have fewer small journals (like mine?) that cannot do this because they don't have the cash flow to resource it - consolidation is important. Smaller journals innovate more easily so balance needed.
comment in response to
post
Copilot is not changing. 🥰
comment in response to
post
It is driving me slightly barmy how, as someone who helps run a small international research collaboration & a distinctly unimpressive consultancy business, even I need to be laser-focused on principle and mission to act strategically... and somehow an entire government cannot manage the same thing.
comment in response to
post
Here are the evaluation reports. We publish these for every submission. zenodo.org/records/1260...
comment in response to
post
I was the handling editor for this. I asked Daniel for the piece because I thought it would be a good fit for a journal introducing RR to a new field. Do you think there was anything wrong with the paper? Or, if you check the reviewer comments and evolution of the preprint, that it was mishandled?
comment in response to
post
You're making some quite large assumptions about how EBT is industry funded (it is funded by T&F with a small contribution from EBTC) and what the journal is trying to achieve (open science policies to make toxicology research work better) and my own motives (I'm formerly NGO turned research).
comment in response to
post
(e) Setting up a journal is difficult, and the opportunities for doing good work can arise in complicated situations. We are doing our best and don't pretend to have perfect solutions. If you want to talk through our journal and your questions, DM me and we can arrange a call. 👍
comment in response to
post
(d) We will be looking to expand and evolve the EB as we grow and get the opportunity to do so. The journal was created because EBTC needed a venue for publishing that is aligned with the mission and values of the Collaboration, and T&F wanted to experiment with open science policies. 4/x
comment in response to
post
(c) Yes there are a lot of interests scattered through the Editorial Board and supporting organisations. We deal with these by balancing them - I, for example, do not have the views expressed in the Nature paper you picked out. As a new journal, many of the EB do indeed know each other. 3/x