https://xitter.cf/t.co/YfcrnyV3jJ #unjournal (https://xitter.cf/t.co/UX8Sn5SYGZ): see more on Bluesky. Research tools. #openscience, #econtwitter, effective philanthropy
Massachusetts, USA
Joined August 2017
- Tweets4.6K
- Following677
- Followers627
- Likes3.6K
π³ππππ πππππππππ retweeted
This interested me! and GiveWell (blog.givewell.org/2024/06/26β¦)! But I was also skeptical. Thankfully I'm an evaluation manager at The Unjournal (unjournal.pubpub.org), where we do open evaluations of impactful research, so we evaluated it.
π³ππππ πππππππππ retweeted
I've noticed when submitting peer reviews that some journals now have a box you have to check that pledges that you didn't use AI when doing the review. This seems dumb to me. I do use AI when I do peer reviews. I do the following:
1. Write my review as I always have,
2. Upload the paper to a frontier model and ask it to do a review as well. (I have settings set up so this data is not retained e.g. for training),
3. I compare the LLM review against mine, and I add to mine things that the LLM caught that I did not if such things exist.
I think this produces a strictly better peer review.
π³ππππ πππππππππ retweeted
GiveWell literally funded research by a Nobel prize winning development economist on a question GW cared about. They then flagged it for an open peer review process (unjournal.pubpub.org/pub/evaβ¦). And yet Leif writes this π
In March I published a critique of Effective Altruism in Wired. After reading the responses, especially from young EAs, I've written a long open letter to young EAs that I hope they'll find useful.
docs.google.com/document/d/1β¦
π³ππππ πππππππππ retweeted
Two takes on the same paper. Different messages (one sounds a lot more positive than the other) but both are technically correct summaries.
π³ππππ πππππππππ retweeted
We are launching the "Allegedly Does Not Replicate" Podcast to discuss these questions (and many more)!
First episode is now live in Spotify and YouTube
podcasters.spotify.com/pod/sβ¦
π³ππππ πππππππππ retweeted
If I ever stop being a professor it will probably be because I can no longer withstand how many people doing research do not deeply care about trying to get the right answer.
π³ππππ πππππππππ retweeted
A thread about how science should work, ways to cheaply save lives, peer review, and diarrhea π
π³ππππ πππππππππ retweeted
Nice thread about an open review of a v important paper (Kremer et al's meta analysis on water chlorination and child mortality).
Great to see the Unjournal (unjournal.org) reviewing this type of important work and publicly sharing reviews - kudos @GivingTools!
π³ππππ πππππππππ retweeted
Researchers: Submit your papers if you'd like to be considered for evaluation at the Unjournal! This is the best way to get your research out, and we're constantly innovating to ensure Unjournal is making a meaningful contribution.
@leightjessica @Prof_Nishith_P @KaranBabbar19
π³ππππ πππππππππ retweeted
I've been part of a novel and open peer-review initiative aptly called the #Unjournal (unjournal.org/), founded by the super David Reinstein (@GivingTools). Posting a thread on some of the evaluations I've (co)managed, and encourage researchers to submit work to us 1/n
π³ππππ πππππππππ retweeted
Ridiculously high-impact applied research job! #altac
LEEP is hiring a Director of Research and Strategy! Take a look to find out more and to apply. leadelimination.org/jobs/
π³ππππ πππππππππ retweeted
βReplicability & Generalisabilityβ: Applying a discount factor to cost-effectiveness estimates.
statmodeling.stat.columbia.eβ¦
π³ππππ πππππππππ retweeted
An Oxford-led paper exploring impacts of #wildlife #trade bans has won first place in The #Unjournalβs 'Impactful Research Prizes'. The study provides evidence that trade bans on some threatened species may increase demand for other non-banned species. See bit.ly/4bcUGC9
π³ππππ πππππππππ retweeted
As a reviewer, I do not recommend acceptance until I have access to the data and materials. I put this statement in all my reviews (copied from another reviewer). If more reviewers do this, maybe journals will eventually change their policies (or stop asking us to review...)
π³ππππ πππππππππ retweeted
In my experience, reproducible manuscripts have probably saved me time. After learning how (lakens.github.io/statisticalβ¦) I have benefitted immensely from recompiling a manuscript after correcting an inclusion error - all tests and figures just update! Would recommend it to anyone.
π³ππππ πππππππππ retweeted
This take is the most important take on Gino's defamation lawsuit.
I hope the academic community supports the datacolada group.
π³ππππ πππππππππ retweeted
This is a tragedy. I am less proud of the UK as a result of this.
π³ππππ πππππππππ retweeted
#1
Two candidates, with same number of publications (~10) from the same journals and very similar qualifications.
A: PhD from Ivy-league University
B: PhD from non Ivy-league
Would: Ivy > Non Ivy.
Should: Non Ivy > Ivy.
π³ππππ πππππππππ retweeted
There are ai citation tools that extract the paragraph in which the citation occurs. What you suggest could be automated by then applying classification tools to fhe mentioning paragraph. Scale it up and create an alternative or add-on to Google scholar.
