iiinova — where your digital self keeps evolving

Reposted by Brad LeVeck

Jorge Peña 🇵🇸 @jorgeapenas.bsky.social · 08/10/2026

Contrition matters too. The contrite discriminator cooperates to make amends when in bad standing. When errors are rare, it is the only strategy that keeps cooperation stable in every dilemma, provided its norm is consistent. It even invades the standard discriminator in some snowdrift games. 4/6

Zoom on the red cluster of the previous figure: the 1,024 consistent discriminating systems colored by strategy, with cooperation index from 0.88 to 0.94 and stability index from 0.83 to 0.95. The 256 systems using the contrite discriminator (CDISC, pink) form the two highest groups, with L2 at the top, near 0.95. Systems using the other three discriminators (DISC, which includes L6; ODISC; and PDISC) sit lower, between 0.83 and 0.87.

111

Reposted by Brad LeVeck

Jorge Peña 🇵🇸 @jorgeapenas.bsky.social · 08/10/2026

New preprint with Julian Garcia & @tobyhandfield.bsky.social 🧵 We tested 1,048,576 moral systems of indirect reciprocity, each pairing a strategy (how to act) with a social norm (how to judge others), across all prisoner's dilemmas, snowdrift games and stag hunts. arxiv.org/abs/2610.03832 1/6

Model diagram. Two players, each with a good or bad reputation, play a two-by-two game with payoffs R, S, T and P. Each player cooperates or defects according to a strategy table that reads both players' reputations. An observer watches and assigns both players new reputations according to a social norm table that reads both reputations and both actions. Side panels show example strategies (ALLC, ALLD, DISC, CDISC, ODISC, PDISC) and norms (Scoring, Stern Judging, Simple Standing, Standing, Consistent Standing, Generous Standing) as colored grids.

33510

Brad LeVeck @bradleveck.bsky.social · 20h

One way to counter AI generated misinformation is to rely more on established news sources with a reputation to maintain, but this is an interesting workaround.

110

Brad LeVeck @bradleveck.bsky.social · 08/10/2026

So, would you say the post had high surprisal?

110

Brad LeVeck @bradleveck.bsky.social · 04/10/2026

The obvious answer is one of a few topics in a research design course, but it’s funnier to me if I pretend it’s on Self-Confirming Equilibrium.

110

Reposted by Brad LeVeck

Stefan Müller @stefanmueller.bsky.social · 29/09/2026

Our School of Politics and International Relations at @ucddublin.bsky.social is hiring two permanent (!) Assistant Professors in any field of political science, international relations, political theory, or development. Deadline: 9 November Start date: 30 August 2027 my.corehr.com/pls/coreport...

Lecturer/Assistant Professor in Politics and International Relations (x2), UCD School of Politics and International Relations, Permanent
Applications are invited for a permanent post of a Lecturer/Assistant Professor in Politics and International Relations (x2) within UCD School of Politics and International Relations.

The UCD School of Politics & International Relations (SPIRe) is a globally ranked, research-active school with teaching and research expertise across the fields of political science, international relations, normative political theory, international development, and other areas. It is committed to the highest standards in undergraduate and graduate teaching and research supervision. The school is home to a number of research centres, and further information can be found at: https://www.ucd.ie/spire/about/

95 Lecturer/Asst Professor Above The Bar (7943) Salary Scale: €66,055 - €104,600 per annum

Appointment will be made on scale and in accordance with the Department of Finance guidelines.

Closing date: 12:00 noon (local Irish time) on 9th November 2026.

Applications must be submitted by the closing date and time specified. Any applications which are still in progress at the closing time of 12:00 noon (Local Irish Time) on the specified closing date will be cancelled automatically by the system. UCD are unable to accept late applications.

UCD do not require assistance from Recruitment Agencies. Any CV's submitted by Recruitment Agencies will be returned.

15546

Reposted by Brad LeVeck

Timothy Gowers @wtgowers.bsky.social · 30/09/2026

The advisory group on mathematics and artificial intelligence, of which I am a member, has just published a set of recommendations concerning the responsible release of mathematical results generated by AI companies using internal models. 1/3 agmai.org/general-sep29/

agmai.org

general-sep29

Responsible Release of AI-Generated Mathematics September 29, 2026 Back to main page Download PDF At present, some frontier AI labs are testing advanced mathematical problems on proprietary models …

23813

Reposted by Brad LeVeck

Bob Elliott @bobeunlimited.bsky.social · 28/09/2026

Its 2036 and you are waxing poetic to some juniors that there was the time you could lock in 3.25% real returns for 30 years. Except of course you couldn't bring yourself to do it at the time b/c you were too greedy chasing stocks at all-time highs.

1213

Brad LeVeck @bradleveck.bsky.social · 27/09/2026

Honestly, pretty understandable. A little depressing how the OP hasn't done anything to correct it a day later though.

010

Brad LeVeck @bradleveck.bsky.social · 24/09/2026

... So, I asked "What's a restricted breathing task?". As they explained it to me, I could feel myself fuming at the fact that the social science IRB had held up my own survey for 2 weeks because there wasn't an explicit "decline to say" response on a question about family income.

020

Brad LeVeck @bradleveck.bsky.social · 24/09/2026

Even within the same university, it can differ enormously. Right after submitting my first IRB, I was talking to an outside committee member with a med school appointment. They described to me a new study where they tested whether a restricted breathing task changed political survey responses...

121

Brad LeVeck @bradleveck.bsky.social · 24/09/2026

Yeah. I'm more thinking of some of the criminology results where theoretically you should be able to replace something like a low detection rate with an extreme punishment, or, as you suggest, make liability very strict. But, empirically, that doesn't always do a great job of deterring people.

000

Brad LeVeck @bradleveck.bsky.social · 24/09/2026

I like this because it focuses us on the right questions: Can we construct laws and enforcement mechanisms where the detection/attribution rate * punishment is high enough to deter reckless AI development. However, I'm genuinely uncertain about whether this is possible or how best to do it.

110

Reposted by Brad LeVeck

Hans Hassell @hjghassell.bsky.social · 22/09/2026

The answer is, as you might expect given the question, Aldrich. What strikes me, though, is the similarity between these two arguments about parties. The subtle differences are important, but they are also very subtle.

131

Reposted by Brad LeVeck

Hans Hassell @hjghassell.bsky.social · 22/09/2026

"The consequence is that office seekers and officeholders are more constrained to the extent they depend on policy-motivated activists for access to office." Who wrote this, John Aldrich in Why Parties or The Party Decides?

131

Reposted by Brad LeVeck

Aditya Dasgupta @adasgupta.bsky.social · 22/09/2026

Brief mention of our paper on bureaucrat assignments as instruments of regulatory corruption in India in the FT: www.ft.com/content/ba07...

062

Brad LeVeck @bradleveck.bsky.social · 22/09/2026

Yeah. I'm curious if this is something the protocol even allows currently. I.e. we have custom algorithmic feeds, but I've never seen a way to give feedback to those algorithms manually.

010

Brad LeVeck @bradleveck.bsky.social · 22/09/2026

Back in undergrad, Jerome Feldman (CS) and George Lakoff (Cogs) taught a course, Computation and the Brain. I.e. what it does it mean in a technical sense to say the brain "computes" and how is it different from standard CPUs? Anyhow, more than 20y later, it's paying off in ways I didn't expect.

070

Brad LeVeck @bradleveck.bsky.social · 22/09/2026

Yeah. That’s a great example.

080

Brad LeVeck @bradleveck.bsky.social · 22/09/2026

But I’m open to the idea that it’s not that unique and I just don’t engage with the far right very much.

030

Brad LeVeck @bradleveck.bsky.social · 22/09/2026

Does the right have a version of this tactic, deflecting via “you don’t understand“? It seems like the left has this uniquely annoying habit of not wanting to try persuasion or explain their views, replacing it with “you just don’t understand” or “it’s not my job to explain X to you”…

1140

Brad LeVeck @bradleveck.bsky.social · 21/09/2026

Anyhow, the answer is a little more complicated than "It's AI's fault," but not much more complicated.

010

Brad LeVeck @bradleveck.bsky.social · 21/09/2026

I think the puzzle is how you were able to buy the PC in July for that price. RAM prices have been high since November of last year and getting worse. The 32gb I bought last September for $90 has been $450 or higher all year.

210

Brad LeVeck @bradleveck.bsky.social · 18/09/2026

Or, to quote Seth bsky.app/profile/smot...

010

Reposted by Brad LeVeck

Aditya Dasgupta @adasgupta.bsky.social · 18/09/2026

In light of this: www.politico.com/news/2026/09... — a good paper on the strategy around an off switch for potentially dangerous AI: arxiv.org/abs/1611.08219

082

Brad LeVeck @bradleveck.bsky.social · 18/09/2026

correction: 69% R in 24. 71% was 2020.

010

Brad LeVeck @bradleveck.bsky.social · 18/09/2026

I mean, the real question is not why Marshall is saying this nonsense. He's in the safest of safe districts (71% R in 24). But a functional party leadership would tell these guys to stfu so maybe a handful of frontline members survive.

210

Reposted by Brad LeVeck

UC Merced Political Science @ucmpolisci.bsky.social · 17/09/2026

Political Science professor Marco Alcocer and his coauthor Megan Erickson's new publication investigates the assassination of government officials by cartels in Mexico. Read the article: academic.oup.com/jpr/advance-...

academic.oup.com

The transformation of Mexican cartels and violence against government officials

Abstract. Since the early 2000s, Mexican cartels have increasingly threatened, attacked, and assassinated government officials. Existing explanations empha

041

Reposted by Brad LeVeck

David Karol @davidkarol.bsky.social · 11/09/2026

Only many years later was the truth about Humphrey's dissent known. By contrast Vance -and others who leaked to Swann and Haberman- have not been frozen out by Trump. Of course, if he did so, he'd have to get rid of most of his senior appointees, which illustrates how remarkable this presidency is.

051

Reposted by Brad LeVeck

David Karol @davidkarol.bsky.social · 11/09/2026

Vance trying to have it both ways on the war is amazing and could only happen under Trump. We now know Humphrey told LBJ not to escalate in Vietnam. But he never dared let this be known in real time, even via leaks as Vance does. Even his private dissent led LBJ to freeze him out for a long time.1)

192

Reposted by Brad LeVeck

Vince Crawford @vincecrawford.bsky.social · 11/09/2026

If they ask you, read up and consider Elsevier's conduct.

041

Reposted by Brad LeVeck

Robin Brooks @robin-j-brooks.bsky.social · 10/09/2026

What's amazing about the global bond sell-off is how many people are saying rising yields are about strong growth. Seriously!? You think rising yields in Japan, France, Italy or UK are about growth? We're in a global debt shock. Not a global growth shock. robinjbrooks.substack.com/p/the-dollar...

44817

Reposted by Brad LeVeck

Kevin Zollman @kevinzollman.com · 09/09/2026

I agree with this. I've use Wikipedia as an analogy for LLMs a few times, and it's clear that people forgot the extreme hostility so many academics had to it in the early 2000s

2212

Brad LeVeck @bradleveck.bsky.social · 09/09/2026

Oh absolutely. I'm too lazy to hand pour, but if I wasn't that's a good one.

020

Brad LeVeck @bradleveck.bsky.social · 09/09/2026

Nice. What'd you go with? Technivorm?

130

Reposted by Brad LeVeck

Dr. Danielle Lupton @proflupton.bsky.social · 09/09/2026

My department is hiring a Tenure Stream Assistant Professor in IR with a focus on human security. It's a great position at a strong university with incredible benefits and good students. Please consider applying and sharing with your networks: apply.interfolio.com/188692

apply.interfolio.com

Apply - Interfolio {{$ctrl.$state.data.pageTitle}} - Apply - Interfolio

043

Reposted by Brad LeVeck

Steven W. Webster @stevenwwebster.bsky.social · 02/09/2026

Open-rank positions in quantitative methods (indiana.peopleadmin.com/postings/34175) and IR (indiana.peopleadmin.com/postings/34176) at Indiana. Deadline for both is 10/1.

173

Reposted by Brad LeVeck

Anne Meng @annemeng.bsky.social · 15/08/2026

🚨Job alert 🚨 UVA is hiring an assistant professor in CP! Apply here 👉🏻 apply.interfolio.com/190673

apply.interfolio.com

Apply - Interfolio {{$ctrl.$state.data.pageTitle}} - Apply - Interfolio

02526

Reposted by Brad LeVeck

Daniel de Kadt @dandekadt.bsky.social · 25/08/2026

Don't sleep on this opportunity to join my former department. The Department of Methodology at LSE is a great place to work, and London is a great place to live.

lse.ac.uk

We're hiring: Assistant Professor in Computational Social Science

02330

Brad LeVeck @bradleveck.bsky.social · 25/08/2026

Thanks! I just cc'ed you because I saw this post through your retweet of it.

000

Brad LeVeck @bradleveck.bsky.social · 25/08/2026

What's the prevailing theory for why these growth rates appear to diverge after the Great Recession? cc @besttrousers.bsky.social.

650

Brad LeVeck @bradleveck.bsky.social · 24/08/2026

What’s your general heuristic between Chat and Claude (assuming those are the other two)?

110

Reposted by Brad LeVeck

Rachel Ryskin @ryskin.bsky.social · 24/08/2026

Very clear demonstration of the remarkable preservation of the language network in the brains of older adults (in contrast to the MD network)! 🧠 👵

1111