This thought by Tom Kirkwood probably makes a good &hopeful closing message #WhyWeAge2020.Should have tweeted it last.I think I tweeted a summary of what I found striking in all talks you can find them under the hastag #WhyWeAge2020, I have more pictures from the slides just ask!

Wow that's quite a thing!

Zepzelca failed confirmatory? Mhm

Guess who's back: @lcbnuoH youtu.be/YVkUvmDQ3HY?si=I3a-…

Sold some $BBIO today (~ 1/3 of position now slightly under 10%)- this has been a great one from 23, and except the earlier HIKMA settlement things went great for them this year too getting out in the 80's would have been better but still

as much as I think GSK overpays here, ultimately not our problem. Lot of capital being recycled back. And say whatever you want about 2026. M&A has NOT been an issue. Best 1H in a really long time.

$CYTK what did Bloom do this time I was away...

Wild actions in European banks!

This is an insane paper and I love it arxiv.org/abs/2605.31514

Can't leave this thing (the XBI) alone for half a day looked much better when I left earlier ha. @pawcio2009 fishing this weekend?

What are more recent equivalent books to MacLane’s “Mathematics Form And Function”? eg js Lawvere’s “Conceptual Mathematics” of a similar vein?

This is the funniest chart. It took Micron around 20 years to recover from the Dot Com Bubble- And then they climbed by 900% in the last year.

❤️ Do biologics differ in MACE risk in PsA? YES - JAKi highest risk after 2.5yrs vs TNFi/IL-17i/IL-23i in a retrospective cohort of 22,246pts. OP0181 #EULAR2026 @RheumNow

🚨 Stage III unresectable NSCLC has quietly become one of the biggest precision oncology success stories. A decade ago: ➡️ Median OS ~2 years Today: ✅ PACIFIC: mOS 47.5 months ✅ LAURA: mPFS 39.1 months in EGFR-mutant disease ✅ 5-year OS approaching 43% The key lesson? Every patient needs upfront molecular testing. 🧬 EGFR-mutant → Osimertinib 🛡️ EGFR wild-type → Durvalumab The era of “all stage III patients get the same treatment” is over. Full paper in comment below ⬇️ #OncoTwitter #MedTwitter #LungCancer @OncoAlert @myesmo @esmo_open @JCO_ASCO

If the S&P closes up this week, it would be 10 up weeks in a row. The last time this happened was in 1985. So a real rarity.

$MLYS Really is a strange name. Looked at their deal (buying back ~7% royalty, maybe a bit more, it´s tiered 5-10%) for 200m upfront 100m in m/s (lets call it 225m NPV). One hand, they buy like 14-20% (roughly?) of the economics of the drug (assuming a spending-heavy sales force to compete against AZN) for a ~6.9% dilution 75m in debt which is pretty damn great if you´re a bull. OTOH Tanabe sells its royalty for an implied "drug value" of ~1.1-1.6B (225/14-20%) against today´s market cap of 2.2B / EV 1.7B. All of that against AZN expectations they´ll be doing 5-10B with Baxfendy and MLYS possibly taking 15-30% of market? If they manage to sell the Co for a decent price, this will have been a very strong deal.

Horror! - Doppelknall in der deutschen Pharmabranche. Lilly halbiert Milliardenprojekt in Deutschland. Die Investitionskrise erreicht nun auch die Pharmabranche. Wenige Stunden zuvor hatte bereits Boehringer Ingelheim öffentlich bekannt gegeben, 900 Millionen Euro nicht in Deutschland zu investieren. Dir Ursache läge in Berlin. #Pharma #Investitionen #EliLilly #Industrie Standort: Der US Pharmakonzern Eli Lilly plante ursprünglich Investitionen von 2,5 Milliarden US Dollar in Alzey. Am 3. Juni 2026 teilte das Unternehmen mit, dass der noch ausstehende Projektumfang um 50 Prozent reduziert werden soll. Ursachen: Konzernchef Dave Ricks begründete die Entscheidung mit den gesundheitspolitischen Rahmenbedingungen in Deutschland. Im Handelsblatt erklärte er, Deutschland werde bei der Unterstützung der Branche auf den letzten Platz der europäischen Märkte zurückfallen. Die Entscheidung ist ein weiteres Signal dafür, dass internationale Konzerne ihre Investitionen in Deutschland zunehmend kritisch bewerten. Vielen Dank für den wichtigen Hinweis! Quelle: boerse.de / dpa AFX boerse.de/nachrichten/Eli-Li…

Over Al the $ABVX drama what happened to the $LNTH rumour(?)

Because of the Drexler Smalley debate? And because Smalley kinda "won" even though all the 90's kids how there is a secret way!

Main reasons why LLMs are bad at multi-layered abstraction-heavy domains like Langlands program: > LLMs can come up with 1-2 tricks, they can't theory build in any significant way, and this kind of domain usually requires new methods/settings to begin with. > Combinatorics and similar domains are definition-light, which makes it fully about the tricks. Arguments are usually short. Definition-heavy domains require juggling many layers of abstract concepts, and going from global (macro level of where we are in the proof, e.g. we want to prove an equivalence of categories) to local computations (e.g. computing cohomology groups of some ad hoc defined schemes) many times over, back and forth. This is a bit like Go in terms of making local plays that are useful globally, but here the board is basically infinite. I've been thinking about how to overcome this with auto-research methods and it looks genuinely hard. You could solve it sometimes by throwing way more compute (context, Monte Carlo brute forcing attempts, AVO-style evolution, etc, etc), but this is definitely not the right solution. We need some new understanding here of what local-global, multi-layer problems involve.

$BBIO now below where it traded last November when longer Pfizer patent was on the cards but considering all the ve updates