Reposted by Matt CollinMatt Ford @mford.bsky.social · 24/09/2026Ezra Klein asked Nvidia CEO Jensen Huang about AI making kids worse at math and reading. Huang’s reply: “Does it matter?” Then he said he doesn’t know his own address or phone number. “We’re going to lose some fine intellectual dexterity.” www.nytimes.com/2026/09/23/o... 58755101512
Matt Collin @mattcollin.bsky.social · 23/09/2026I'm a development econ guy and I don't think I'd be that happy losing my job to the AIs! 011
Reposted by Matt CollinBrexit Bin 🇪🇺 🇬🇧 🇩🇪 @brexitbin.bsky.social · 21/09/2026So true !!! 💻 Check out this great Norwegian video about how shitty the internet has become 😅 310968
Matt Collin @mattcollin.bsky.social · 10/09/2026One thing I've found that Claude Code is good at is very quickly showing me visually something my 42 year old brain strugles to understand from its sometimes challenging prose 010
Matt Collin @mattcollin.bsky.social · 07/09/2026That means that we've entrusted companies who many not have the best track record of which clients they work with, with safeguarding the validity of the Companies House register! 000
Matt Collin @mattcollin.bsky.social · 07/09/2026When we rank ACSPs by their "riskiness" (what share of affiliated companies got struck off or moved to CH's default address, we find a pretty extreme concentration in ID verification behavior. The top decile of risky ACSPs have performed roughly 40% of verifications of companies with foreign-BOs!! 120
Matt Collin @mattcollin.bsky.social · 07/09/2026Now, ACSPs are AML regulated, so in theory they should be doing a good job But not all of them have a history of doing a good job In fact many have formed, or worked with companies that have been involuntarily struck off or caught by Companies House for lying about their place of business 100
Matt Collin @mattcollin.bsky.social · 07/09/2026Companies have 2 ways they can verify the identity of their officers: either directly with CH (free, but requires biometric ID) or through a third part "Authorized Corporate Service Provider" With the latter - another company tells CH that they checked your documents and everything looks ok 100
Matt Collin @mattcollin.bsky.social · 07/09/2026Those companies will have to come into compliance by November of *this* year, but if you are worried that some companies are engaging in money laundering or fraud or serious tax evasion, then those companies bought themselves a full year of freedom from scrutinystatic.klipy.comWizard of Oz: Pay No Attention to the Man Behind the CurtainALT: Wizard of Oz: Pay No Attention to the Man Behind the Curtain 110
Matt Collin @mattcollin.bsky.social · 07/09/2026We find that many companies that would have had to undergo ID verification because their filing window falled on or after Nov 18, just file a confirmation statement early, effectively resetting the clock for a year Depending on the method used, between 5-13% of companies close to the policy window 100
Matt Collin @mattcollin.bsky.social · 07/09/2026But we also find that many companies used a loophole in the company filing process which allowed them to push their deadline back for nearly a year Companies normally file their confirmation statement on a yearly basis But they can find it earlier than their filing window, even without new info 100
Matt Collin @mattcollin.bsky.social · 07/09/2026We find significant compliance effects: beneficial owner-directors of companies 'caught' by the policy were around 46% more likely to have verified their identity roughly six months later 100
Matt Collin @mattcollin.bsky.social · 07/09/2026Directors (inc beneficial owners who were directors) had to verify their ID if the company's confirmation statement was due on or after Nov 18, 2025 So we use the confirmation statement date - measured *prior* to the announcement of the policy data to implement a regression discontinuity design 100
Matt Collin @mattcollin.bsky.social · 07/09/2026New research out with @jeannebomare.bsky.social and Eva Nappert We examine the impact of Companies Houses's new ID verification rollout, which intends to require all directors and beneficial owners of UK companies to verify their identity either directly with CH or through a third party provider 151
Matt Collin @mattcollin.bsky.social · 04/09/2026Look, I am in favour, but I really don't think firms engaged in international trade give a shit about how big a country looks on a map 010
Reposted by Matt CollinKetan Joshi @ketanjoshi.co · 02/09/2026Who would like to guess why UK, US and European media outlets are not publishing 3,000 articles per minute about this www.dukechronicle.com/article/duke...dukechronicle.com‘Insane data’: Duke professor Dan Ariely accused of additional research fraudNew evidence released Monday shows that Dan Ariely, professor of business administration in the Fuqua School of Business, likely tampered with or fabricated data in an influential research paper from ... 502488755
Reposted by Matt CollinBelinda Archibong @belindaarch.bsky.social · 31/08/2026ICYMI, the new “America First Global Health Strategy” has turned the US approach to global health into a series of transactional, bilateral agreements with potentially devastating consequences for global health security as we write in our Brookings piece here: www.brookings.edu/articles/the...brookings.eduThe new 'America First Global Health Strategy' could erode years of progress under PEPFAR | BrookingsThe authors unpack the impacts the new "America First Global Health Strategy" will have when it comes to HIV/AIDS mitigation in African countries. 141
Reposted by Matt CollinBrandon Friedman @brandonfriedman.bsky.social · 11/08/2026Axios writer: I melted my brain with AI. Not only can I no longer read, but I don't even understand the purpose of reading anymore. Here's how you can also do this. www.axios.com/2026/08/11/r... 49372151666
Reposted by Matt CollinStephen Abbott Pugh @stephenabbottpugh.bsky.social · 10/08/2026EU Commission may open new cases against states lagging on beneficial ownership registry access www.amlintelligence.com/2026/08/news... News story by @cgboffa6.bsky.social for AML Intelligence #beneficialownership #legitimateinterest #6AMLDamlintelligence.comNEWS: EU Commission may open new cases against states lagging on registry access - AML IntelligenceThe European Commission said it may open a case against countries that have not yet completed a framework to grant journalists and civil society organisations access to national beneficial ownership r... 034
Reposted by Matt CollinStephen Abbott Pugh @stephenabbottpugh.bsky.social · 31/07/2026Beneficial ownership information on 480,360 German entities is missing from the country’s national Transparenzregister, according to a risk assessment released by the country’s ministries of finance and justice www.bundesfinanzministerium.de/Content/DE/D... #beneficialownership #UBO #AMLR #6AMLD 154
Reposted by Matt CollinG. Willow Wilson @gwillow.me · 27/07/2026Not sure I’d shit on something that helped me sell half a million copies of my book, personally 6065836
Reposted by Matt CollinChristian Odendahl @codendahl.bsky.social · 24/07/2026The winning design. 🇪🇺💶 311603506
Matt Collin @mattcollin.bsky.social · 16/07/2026Tip to money launderers: if you want to buy a property, maybe don't buy a property that has already been frozen in the past on suspicion of money laundering 042
Matt Collin @mattcollin.bsky.social · 30/06/2026Europeans: Beneficial ownership data is a massive privacy risk! We have to make it hard to access UK: We have an public streaming API that will let you see ownership changes in real time 052
Reposted by Matt CollinStephen Abbott Pugh @stephenabbottpugh.bsky.social · 30/06/2026Watch UK beneficial ownership information change in real time as open, standardised data bods.stream #beneficialownership #openstandards #opendata #UBO #datastandards #interoperability #opensource 173
Matt Collin @mattcollin.bsky.social · 25/06/2026I mean, good lord, it's 104 degrees outside right now. No one wants to be outside, air conditioning or not. The worry that occasional air con use is going to destroy the fabric of society is really very silly 050
Matt Collin @mattcollin.bsky.social · 25/06/2026Every morning my 18 month old gets up and demands that I put on a vinyl record (by pointing to her ears) She only accepts vinyl music And for the past two weeks, she only accepts this record by Ghanian musician Gyedu-Blay Ambolley. Before that it was a Bowie obsession. 5110
Matt Collin @mattcollin.bsky.social · 24/06/2026This looks awful. Klara and the Sun is a haunting, strange and very beautiful book which absolutely did not deserve the Taika Waititi twee treatment youtu.be/wixzainceAE?...youtu.beKLARA AND THE SUN – Official Trailer (4K)YouTube video by Sony Pictures Entertainment 200
Matt Collin @mattcollin.bsky.social · 24/06/2026Got to love the French university position of "We didn't invest in climate control, so please don't come to work and die of heatstroke, we much prefer that you die of heatstroke at home" 070
Matt Collin @mattcollin.bsky.social · 19/06/2026At what effective sample size would you be satisfied? 000
Matt Collin @mattcollin.bsky.social · 19/06/2026The estimated effect is more or less the difference in where the end of that local polynomial intercepts the cutoff on one side vs the other side 100
Matt Collin @mattcollin.bsky.social · 19/06/2026It is an automatic bandwidth selection procedure, which choose a bandwidth (in this case days/weeks away from the cutoff) that balances bias with power, more or less). Then it runs a linear regression through the remaining sample on either side.... 100
Matt Collin @mattcollin.bsky.social · 19/06/2026We discussed this two times before, so maybe third time is a charm: These kinds of RDs aren't just a comparison of people on either side of the boundary. That's the "concept" of an RD, but not how they are actually estimated these days. 100
Matt Collin @mattcollin.bsky.social · 19/06/2026I recall very well, you mentioned many problems! But your take on the sample size mischaracterized how these RDs work: after bandwidth selection, the regression sample used for the RD had about 100,000 people in it, which seems pretty well powered 200
Matt Collin @mattcollin.bsky.social · 19/06/2026I thought you didn't accept RDs where age was a running variable? But you accept the ones that show a null effect? ;-) 110
Reposted by Matt CollinZachary Tvarozna @shadyshores.bsky.social · 12/06/202618 months in the making! My conclusion forever will be Russia's A7 was never entirely a cryptocurrency network and relied on captured states. Thanks 2 everyone who supported and hugest s/o to the OSC for turning this hypothesis into an investigation. www.opensourcecentre.org/research/the...opensourcecentre.orgThe Big ShorOSC and TRM Labs reveal how the Kremlin-backed A7 Network used the international correspondent banking system and Kyrgyzstan to likely move billions of dollars of Russian payments through the global f... 144
Matt Collin @mattcollin.bsky.social · 16/06/2026Not judging it! I've only been to Vietnam, where I don't remember seeing it. Just found it funny, since there was a 2 month period when I was in middle school where it was a status marker to have one, and then I've basically not seen them since. 100
Matt Collin @mattcollin.bsky.social · 15/06/2026"We believe these people are circumventing sanctions" and "We are not suggesting any wrongdoing here" in the exact same paragraph is a hell of a way to start a report 010
Matt Collin @mattcollin.bsky.social · 15/06/2026I really don't know how the EAs keep convincing other EAs to give them money to... start blogs? 010