Associate Prof @VU_Law • Faculty Affiliate @CodeXStanford • Founder Stanford #ComputationalAntitrust project | @NetworkLawRev | Scaling Theory podcast • 🎾 🏃
Joined November 2012
- Tweets 12,300
- Following 808
- Followers 10,907
- Likes 26,408
4,192 Photos and videos
Pinned Tweet
18 Mar 2024
🎙️ NEW PODCAST 🎙️
Where should I start? I am launching a new academic podcast called “Scaling Theory”. I am of course incredibly excited, but more importantly, I hope you’ll find it interesting.
I want to explore the power laws behind the growth of businesses, technologies, legal systems, and living systems. The podcast will feature scholarly discussions with select guests (I’ve already recorded a couple, to be published very soon!!) and deep dives into the academic literature. Everything is explained in the introductory episode.
➝ Spotify: open.spotify.com/show/7o5GqR…
➝ Apple: podcasts.apple.com/us/podcas…
➝ YouTube: youtube.com/channel/UCoxJ0e-…
3
24
77
15,296
US blocks Fable 5 in Europe.
Europe: “See? This is why the EU must build its own AI.”
Solving “a government controls our tech” with “a different government controls our tech.” Visionary.
1
2
163
As W. Brian Arthur put it, technology is what separates us from the Middle Ages.
1
118
Thibault Schrepel retweeted
In memoriam: Marius Schwartz (@Georgetown)
In tribute to Marius, we are republishing the article today. We extend our deepest condolences to his family, to Yongmin Chen, and to his many colleagues and friends.
networklawreview.org/default…
2
270
Delighted to host @PZilgalvis next week!
Human oversight, fundamental rights, judicial decision-making.
What changes with AI?
Judge @PZilgalvis (@EUCourtPress) discusses 👇
🗓 June 16, 12:00 to 13:00
Register: alti.amsterdam/event/peteris…
1
243
Strong IP. Slow innovation.
Weak IP. Faster innovation.
That’s the unsettling lesson Jason Potts (@profjasonpotts) draws from new Chinese innovation systems. He’ll present the evidence.
Join this exceptional talk👇
alti.amsterdam/event/jason-p…. ONLY at @alti_VU.
1
2
180
Subtitle: “Now that we have the lead, let’s pause.”
Tennis equivalent: “I was one set down, I now lead two sets to one, so instead of playing another set, let’s just stop here.” Ridiculous.
1
1
3
449
AI is entering courts. The hard question is under what conditions.
Honored to host Judge @PZilgalvis (@EUCourtPress) on June 16 to discuss how the CJEU approaches it.
Join us 👇
alti.amsterdam/event/peteris…
2
122
Thibault Schrepel retweeted
1/ From the @FTC to Saudi Arabia, antitrust agencies are building computational tools. Bid-rigging detectors. Contract scrapers. LLMs for spotting illegal information exchanges. Antitrust 3.0 is here. The question is what gets left behind. #computationalantitrust
1
1
5
434
1/ From the @FTC to Saudi Arabia, antitrust agencies are building computational tools. Bid-rigging detectors. Contract scrapers. LLMs for spotting illegal information exchanges. Antitrust 3.0 is here. The question is what gets left behind. #computationalantitrust
1
1
5
434
2/ Here is the worry. Every tool so far is built for what counts cleanly. Prices, quantities, document volume. Innovation does not count cleanly. There is no economic theory of it. So the machines get sharper while the blind spot grows. Or does it?
1
2
238
3/ Our answer is contrarian. Computation is not the enemy of innovation competition. Done right, it gives innovation a more robust role than the old case-by-case approach ever did. Nothing is automatic. The challenge is ours to get wrong. Read it 👇
papers.ssrn.com/sol3/papers.…
3
195
1/ On Monday, I gave a talk to the UK Competition and Markets Authority @CMAgovUK on AI agents and antitrust.
The claim is simple. Behavioral economics is the intuitive lens for these cases. It is an insufficient one. A new author now sits between the consumer and the market. The consumer no longer navigates the choice environment. A chatbot authors it.
Paper: papers.ssrn.com/sol3/papers.…
1
1
534
3/ First, belief. An AI agent does not only recommend. It retrains. Each round narrows the next, and a small skew compounds. Or the opposite.
Second, preference. Recommendations do not satisfy preferences. They constitute them. So the counterfactual is not undistorted choice. It is a different architecture.
Third, equilibrium. Markets with increasing returns do not clear. They tip. The question is which way.
1
269
4/ For an agency the payoff is measurement. Run synthetic users through the chatbot API. Score output diversity. Entropy that stays low marks a self-reinforcing attractor.
The framework is symmetric. It supports clearance as readily as intervention.
papers.ssrn.com/sol3/papers.…
108