Simon L

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!

@AaronRosenblum5 Thanks!

When the headline said $MAZE meets its own expectations I could have sensed something amiss and it's down quite a bit-what do people specifically dislike

Sugar has interesting local marked dynamics especially in the EU

$tenx they going to 10x the amount of shares outstanding

I just saw this letter by Spassky to President Bush, wow!

Oh my! (thread)

$GLPG is impacted too they are the partner

Breaking: $GILD to acquire Ouro Medicines for up to $2.18 billion @bloomberg “Under terms of the deal, Gilead will pay $1.68 billion in cash up front to buy Ouro, plus as much as $500 million more contingent on meeting certain milestones. The deal would provide Gilead access to Ouro’s antibody drug gamgertamig, which received so-called “Fast Track” designation from the US Food and Drug Administration in January. It’s in early-stage trials for autoimmune hemolytic anemia, immune thrombocytopenia and other autoimmune conditions.” Editor: And interesting re: $GLPG..

@Joshthe3rd_ Thanks 🙏 you made your daily random internet critic happy haha!

@SnupSnus My bad, I have no idea how PCYC slipped through, here is an updated chart for you. It shows announcements only.

$XBI $IBB $BBC 2014-2025 All 8 Billion+ public biotech M&A transactions charted (UPDATED)

@Sentient_Atoms @maxmarchione Interesting thank! Good luck with it too!

My NO system is genetically bottlenecked at every level, so when the veins/arteries constrict, I don't natively get the normal endothelial pushback that buffers it. Plus I have multiple 100% coronary blockages maintained by collaterals at their flow ceiling, so narrowing the pipes is not ideal. Even before I developed ASCVD, it was always genetically a bad idea (though I had no clue back then), and I was a smoker for about 10 years. I do have to keep graft durability in mind personally, but, an otherwise healthy person with impaired eNOS/BH4/cGMP genetics could supplement their way into narrowing the gap enough for limited nicotine use. I'd venture to say most people have never researched their nitric oxide related genetics to know if they are even putting themselves at risk.

Just about every >150 iq person I know uses nicotine. Nicotine is underrated and misunderstood

@bingbingbom I can spoiler you for WH: He has an earing! Back in the day😱 Don't you dare spoiler me for Hoppers I actually have to see that one on Wednesday!

can’t believe I’m watching Hoppers right now it was either this Wuthering Heights

@bingbingbom Hey your writing used to be less Schizo .

$COGT. Serious post. Read until end. Bezu for GIST only. part 1 – known material nontiming factors part 2 – known material timing factors part 3 – scenarios analysis based on known material nontiming + timing factors A brief look into unexhaustive factors I find important or interesting. Theres few typos. Part 1.0: factor of concern: brief overview of drug activity : (1) the drug caused tryptase to decline alot, consistent W/ mast cell reduction, and this physio signal is the foundation for credibility. (2) in pbo, n1 showed that tryp drop—0 hit—so the on-trt effect looks clean. (3) even if pts guessed tx arm from skin sx, you can’t fake a blood read; under worst-case imputing (wk24 missers = nonresp), the ~87 vs 0 gap still leaves ample margin over pbo -> efficay concern decreased. (4) prespec labs + tissues line up: tryp, KIT-vaf, bm mast cells all stat-sig and consistent—plausible biol, no red flgs. (5) net in a pbo-controlled frame: the drug shifts disease in blood/tissue by a big margin; surface read = credble, enuf to lean long so far. (6) earlier GIST reads lack rndmzd ctrl, so no formal comp adv yet, but the ~19.4-mo mPFS in heavily pre-tx pts is a real actv signal for the combo—true on its face. (7) post-imatinib resis is common; bezu hits KIT exons 9/11/17/18 and suni/sub hits 9/11/13/14—spectrum coverage looks logical. Ok looking good. part 1.1 factor of concern: prior efficacy signal + plausible effect size: (1) ph1 shows mPFS 10.2mo (95% ci 7.4–19.4) overall w/ orr 27.5%, and in the 1-prior-tki bin mPFS 19.4mo (95% ci 1.0–ne) w/ orr ~33.3%. (2) vs a 2l suni/sun benchmark ~8–9mo mPFS, this leans towards a feasible p3 ranomized effect, hr ~0.67–0.80 (low-teens combo medain) not some extreme ~0.50 (16–18mo), which is rare in randomized 2l settings. (3) the single-arm “floor” ~10mo despite ~⅔ being suni-expected provides basis for directional plausiblity of separation once the p3 pop is cleaned to randomized 2l suni-naive pts. (4) net: looks like a real signal, expect decent delta, not moonshot, well, anything goes. But looking fine part 1.2 factor of concern: control arm: (1) a credible prediction needs a credible control anchor; the expected second-line suni median pfs of 8- 9 months (conservative control range to bracket variability, and to avoid over attrit to any cal delay to drug effect, basically upper bound strees test, not base case, though yes 7-8 may be more reasonable; and if control goes to ~9–10 mos and the combo underdelivers (dose-intensity loss, operational drag), you can land at ~9.3 v~9.3 , hr ~ 1.0 - consistent with timing? less so, with q8–12-week scans and typical bicr/lock, a >13-month lpi to topline gap is directionally inconsistent with hr ~1.0 unless operations are unusually slow + a true null usually hits the event trigger roughly on schedule) serves that anchor. (2) if ph3 reproduces a ctrl mPFS in that ~8–9mo band, an hr ~0.70–0.80 implies a low-teens combo median—quant-consistent w/ part1 actvty and the evt-math. (3) conversely, a ctrl inflatn toward ~12mo compresses eff sz and raises non-sig risk; this is watched by checking stratif vars (lot, mut mix) and keeping censorng/exponsure symmtry across arms. (4) net: credible anchor = feasible effect. Looking good. part 1.3 factor of concern: crossover + os interpretability: (1) prespec crossover implies the first intention-to-treat os analysis is expected to be neutral even when pfs is positive. (2) W/ x-over baked in, the 1st itt os readout likely nets ~null despite a +pfs; model-based adjstmts (eg rpsft/ipcw) can be directionally supportive but not decisve for a near-term “success” call. (3) so, a neutral itt os at the initial cut remains compatble w/ a real pfs benefit under allowed xover, esp if censorng/expousre stay symmetric. (4) net: no need to focus too much on early os; pfs wins stand on their own in this crossover design. Neutral. part 1.4 factor of concern: safety, dosing, and exposure: (1) dose reductions and interruptions in the mid-20% range (approx 24 to 29%) represent a realistic pathway to exposure loss. (2) if these cluster in the combo arm, 2 things follow: 1 lower bio efficacy and 2 more missed scans/censorng, which can artifactually tug the KM shape; net effect = damped signal + funky km steps. (3) a credible prog-free-surv read needs broadly similar dose intensity across arms (approx W/in ~10%) and tiny early-window censor deltas (approx <=2–3% over initial assess windows), ie keep dosintensiy and early censr stable arm-to-arm. (4) these quant thresholds address whether observed sep’n is real biology vs ops-driven censoring; if di/censor gaps stay tight, the speration looks bio not operational. part 1.5 factor of concern: case mix + biology: (1) diffs p1 p3 in molecular subsets (for ex, the distribution of resistant biology such as KIT exon 9), prior therapy exposure, and suni-naive purity materially influence effect size. (2) in p1 vs p3, shifts in mut-mix (eg kit ex9 load), prior tx exp, and suni-naiv % all move eff sz; p3’s rndmzd 2l suni-naive design strips a confound that p1 had (approx⅔ suni-exp), so expct clr’r sep if ctrl sits ~8–9mo. (3) all-else-equal, the p3 purif’d pop should bias the hr down (better sep), whereas p1’s 2/3 suni-exp made the bar higher—ie more resis bio, less headroom; hence the delta vs ctrl was probly damped there. (4) net: cleaner randmzd 2l, suni-naive frame -> clearer seperation; any miss vs ~8–9mo ctrl would more likely be ops/assay noise than true lack of actvty. part 1.6 factor of concern: endpoint coherence: (1) W/ 6 -12–wk imaging, the segmentation of hazard produces stepwise km patterns. (2) under q6–12wk scans the haz gets “chunked,” so km curves step; robustness is better when pfs dirctnality is mirrored by actvty readouts like orr and, if avail, dor/ttr. (3) part1’s orr 27.5% overall and 33.3% in the 1-prior-tki bin sets a quant prior; in ph3, any orr uplift vs suni—even nominal if alpha is spent—so long as it’s consistent w/ pfs, strenghtns the bio infernce. (4) net: stepwise km is expected from assess windows; the key is pfs + orr/dor/ttr lining up, if so then signal looks biologic, not just schedulng artifacts. It’s fine. part 1.8 factor of concern: ddi risk: (1) a mech concern is cytochrome p450 3a4 induction that lowers suni exposure, W/ reported steady-state area under the curve reductions on the order of -26 to -35 % in relevant combos. (2) if present at scale, such a ddi would accel progress (earlier evnts) and blunt efficay; net = attenuated signal. (3) in the current framwrk, a protracted path to maturty is directionally inconsstent w/ a dominnt negative ddi, and there’ve been no protocol level ddi amendmnts disclosed—ie no smoking-gun ops change. (4) absent cntrevdnce, ddi is unlikely the prim drv of any aparent effcacy delta. So pretty good. part 1.9 factor of concern: external comparators + floor/ceiling effects: (1) in later treatment lines, regorafenib’s approx 4-month median pfs frames a lower-bound reference; in second line, suni’s approx 8 to 9 months frames the control expectation, while part 1’s roughly 10 month overall median (W/ a suni experienced majority ) and roughly 19.4 month 1-prior-tki subset signal set a realistic ceiling and indicate that a hazard ratio in the ~0.67–0.80 range is consistent W/ historical patterns for targeted add-ons in second line; a hazard ratio near ~0.50 would exceed typical randomized experience. (2) in later lines, rego’s ~4mo mPFS = low-bd ref; in 2l, suni’s ~8–9mo is the ctrl bar; p1’s ~10mo overall (w/ suni-exp maj) + ~19.4mo 1-prior-TKI bin give a real “ceiling” (3) so an hr ~0.67–0.80 tracks hist pattns for tgt add-ons in 2l—ie expect low-teens combo med—whereas an ~0.50 hr (16–18mo vibe) would be outlier vs rndmzd exp; not the base-case, more like unicorn. (4) net: floor = rego ~4mo, ctrl = suni ~8–9mo, ceil = p1 ~10/19.4; the plausible band is hr ~0.7–0.8, not ~0.5. Very good. part 1.1a factor of concern: sample size, maturity, power: (1) an 412 pt design targeting roughly 290 evts at approx 75% maturity concentrates power at this boundary. (2) in numbers: n=412 w/ ~290 evts (~75% maturty) stacks powr right at the info edge, so the test bites when the curve seperation actually shows. (3) power sensitivty to hazard ratio mags in the 0.67–0.80 matches p1 priors + the ctrl anchor; ie the band we care about is prespec’d by p1 and the suni ctrl, not some fantasy. (4) net: samp-sz + maturty + prior sigs -> pfs-pos expec w/out needing an extreme eff sz; hr ~0.7–0.8 is enough, no need for ~0.5 moonshots. It’s fine. Part 1 summary: nontiming factors for efficacy leaning bullish: (1) taken together—part 1 efficacy (median pfs 10 .2 months overall; 19.4 months in the 1 prior tki subset; orr 27.5 and 33.3%), a credible second-line suni control anchor of approx 8 to 8 months, crossover dampening of early os, acceptable safety + exposure symmetry thresholds (dose intensity W/in approx 10%; early censoring W/in approx 2 to 3%), and the low likelihood that a strong negative ddi dominates—the non-timing evidence points toward a stat sig pfs in p3 W/ a hazard ratio around 0.67 to 0.80 and a low-teens combo media (2) in p3 terms: expect pfs+, hr ~0.67–0.80, low-teens combo mPFS; 1 key 2% could hit if alpha was preserved, otherwise 2%s land as descrptive but dir-aligned w/ pfs (3) os at the first itt cut is ~neutral under x-over; don’t over-read early os—this is baked into the design, not a red flag (4) net: these conclsuons rest on NON-timing evidence (exp symmetry, early-censr bal, ept cohernce) and allign W/—rather than depend on—the separate timing-based inference (maturty/evt accrual) part 2.0 timing: brief overview of timing on efficacy: we want to gauge, before topline, what p3 results are most likely; the trial is evt-driven on pfs, W/ a primary analysis once a prespec no. of “progression” evts are confirmed by a blinded indep central review, and bc imaging happens on a schedule (not continuously) the cal time from last-pt-in to topline carries info about how quickly evts accrue—longer intervals are directionally consistent W/ slower evt accrual (supportive of efficacy on the experimental arm), shorter intervals fit faster accrual (weaker efficacy); this timing signal is directional, not evidentiary (3) safety + data modulate meaning: dose reducs/interrupts/discons -> exposure loss; missed/delayed scans increase early censr; if 1 arm misses more early scans (eg AEs), the KM can look better w/out true biol—ie artfctl steps; so any timing hint must be read alongside exp symmtry and early-window censr balance. (4) rule-of-thumb: if dose-intensity W/in ~10% and early-interval censor deltas <=2–3% and epts cohere (pfs <-> orr/dor/ttr), the dirctnl timing read is credbl; if not—eg higher AE-driven early censors on combo—apparent sep may be ops-artifact even if nominally sig; net: interpret timing w/ exp+censr context or risk a mis-call. something to follow. part 2.1 timing: evt driving pfs trial mechanics: (1) how an evt-driven pfs trial behaves in mechanics is straightforward: progressions are only observed at scheduled imaging visits, many true progressions occur between visits and are recorded at the next scan, early after randomization few participants have reached their 1st assessments so evt accrual begins slowly and then ramps, and central review plus database lock add calendar lag W/out changing the underlying biology; the primary trigger is an evt count at a target maturity, and bc accrual is staggered the early months after last-pt-in are enriched for participants too early in follow-up to contribute many evts, so stepwise KM drops accumulate later as more participants reach the at-risk windows (2) in short: evts are “seen” only at scans; w/ q6–12wk imaging, true prog’s land btwn visits and get stamped at nxt scan, so early post-rndmz accrul is slow then ramps as more pts hit assess windows; bicr + db-lock add ops lag (uploads, dates, queries, lock) = time stretch, not biol change (3) the triggr is an evt cnt at target maturty (eg ~75%: 290/412); if late, either evts slower (tx eff and/or scan cadence) or ops slower—or both; under an ~expo-like proc the instnt haz can be ~const, yet observed evt cnt over calendar time is discretized by the visit grid, so equal biol effs can yield diff cal lags under q6 vs q12 imaging (4) net takeaway: the LPI->topline gap blends biology (slower prog on exp arm) and ops (bicr + db-lock); while not dispositve, a longer gap lines up more w/ slower prog on exp vs null/neg, provided ops tempo is “typical”—ie no odd site slowness or censor blips skewing the clock part 2.2 timing: expontial approx: (1) a standard approx for time‑to‑evt under roughly exp1ntial hazards is: mPFS_combo approx mPFS_control / hr. if control mPFS approx 8–9 months and hr approx 0.80, combo mPFS approx 10–11.25 months. if hr approx 0.70, combo mPFS approx 11.4–12.9 months. If hr approx 0.50, combo mPFS approx 16–18 months (uncommon for randomized 2L designs (2) : If LPI occurred ~09/03/2024 and >13 months have elapsed W/out reaching ~75% evts, simulations W/ control approx 8 months indicate combo mPFS must be ≳11 months; otherwise, the 75% trigger would likely have been reached by ~September - you do not need hr to be approx 0.50 (approx 18‑month median) to explain several extra months of cal delay; hr approx 0.67–0.80 (low‑teens combo mPFS) can do so once scan timing and bicr/lock lag are included. the longer this interval extends (W/in customary central‑review timelines), the more it tilts toward hr approx 0.7–0.8 rather than a null; conversely, inferring hr approx 0.5 would require an extreme delay well beyond typical operational lag and is historically unlikely in this line (3) evts do not accrue linearly bc assessment windows separate the hazard. under q8–12‑wk imaging W/ bicr, a ~4‑month extension from LPI to topline can be fully explained by hr in the 0.7–0.8 range W/out invoking outsized efficacy. do, a low‑teens combo mPFS is consistent W/ the observed calndar signal, while ~18‑month median would over‑explain the delay part 2.3 timing: ddi and why it matters for efficacy and timing: if the investigational drug induces cytochrome p450 3a4 and lowers suni exposure (for ex decreases area under the curve), combo efficacy could be attenuated and adverse evts could increase from the added agent; in short, if cyp3a4 increase and suni AUC decreas, combo effcy dips and AEs tick up, and a neg ddi that cuts suni exposre should accel evnt accrual (earlier progs), shortening—not lengthening—the lpi->topline gap, so a longer calendar gap is directionally inconsstent w/ a strong neg ddi (2) operational things: a pharmk or ddi substudy W/ pk objectives typically does not gate the primary pfs topline unless it forces a protocol changes such as a dose adjment for exposure; in short, pk/ddi substudy =/= gate unless there’s a prot changes—std designs tie topline to evt maturty + bicr proc, not pk paperwork; tl;dr: timing follows evts + bicr/lock, not pk (3) safety linkage: exposure is the bridge between safety and efficacy, so if the combo arm sees more dose reductions or interruptions, exposure falls and missed scans can rise, which both lowers true efficacy and inflates censoring that can artifactually flatten or separate KM curves; in short: lower DI on combo -> less biol punch + more early censr, so km can look “prettier” w/out real bio; thus, if calndr delay coexists w/ balanced dose-intensity (approx W/in ~10%) and similar early-window censor deltas (approx <=2–3%), the delay aligns w/ true effcy, but if delay pairs w/ asym dose mods + higher early censr on combo, the lag may be ops not biol (4) additional material nuances: prior observations in similar settings that co-administered can reduce suni steady state areaunder curve by roughly 25 35 %, if present here at scale, to bring evts so1r; put simply, ~25–35% suni AUC hit should pull evts forward, so today’s later-than-linear lpi->topline pattern leans away from a dominant ddi-attenuation story; net: longer gap + exp symmetry + tidy early-censr = bio wins, not pk drag. Part 2 summary: final thoughts on material timing factors leaning bullish: structure of evt-driven pfs, the segmentation imposed by 6-to-12–wk imaging, and the greater-than-13-month last-pt-in to topline interval together support a forward expectation of a pfs–positive outcome W/ a hazard ratio in the roughly 0.67-0.80 range and a combo median in the low teens; in abbrev: evt-drvn pfs + q6–12wk grid + >13mo LPI->TL => expect pfs+ w/ hr ~0.67–0.80 and combo mPFS low-teens—no need for ~0.50; the calndr delay is explained by modest–mod hazard cut once bicr/lock lags are in (2) os framing: under typ x-over, first intention-to-treat os likely reads ~neutral regardless of pfs; i.e., xover dampens early os even when pfs is pos—don’t over-worry the 1st os pass (3) safety gate: the calendar-based inf’rnce stays credbl if dose intensity btwn arms is broadly similar (approx W/in ~10%) and early-interval censor deltas are small (approx <=2–3% across the first 2 windows); big early d/cs, materially lower exposre, or asym early censr on combo would weaken the biol read of any pfs edge. (4) net: absent those safety/exposure asym’s, the dirctnl timing signal aligns w/ a biol-grounded pfs improvmt and a neutral os at the initial readout; if DI/censr symmetry holds, call leans true effcay, not ops artifact. Part 3.0 scenario A 50% - PFS win; OS nonsig (crossover allowed); at least 1 hard secondary endpoint (orr or dor/ttr): interval exceeding 13 months from last‑pt‑in (approx 09/03/2024) to the present, W/ a planned primary analysis at ~75% maturity (illustratively 290 evts among 412 randomized pfs), is directionally consistent W/ slower‑than‑expected accumulation of centrally confirmed evts. Under the expintial approx mPFS_combo approx mPFS_control / hr, if control mPFS is approx 8–9 months and the true hr lies around 0.70–0.75, the projected combo mPFS sits in the low‑teens (approx 11–13 months). That hr range accounts for several months of cal delay W/out requiring an extreme hr approx 0.5 (approx 16–18 months). non‑timing anchors increase confidence: Part 1 reported mPFS 10.2 months (95% CI 7.4–19.4) overall W/ orr 27.5%, and in the 1‑prior‑TKI subgroup mPFS 19.4 months (95% CI 1.0–NE) W/ orr 33.3%. this signals define a feasible direction of effect that, if reproduced in a randomized 2L suni‑naive pop, supports a stat sig PFS (95% CI for hr entirely <1). if the multiplicity strat preserves alpha beyond the primary, 1 of orr, dor, or ttr can also reach sig; an orr uplift would be consistent W/ the Part 1 orr (27.5% overall; 33.3% in 1‑prior‑TKI). re safety and exposure, this scenario presumes dose intensity diffs bt arms remain W/in ~10% and early interval censoring diffs W/in ~2–3% across the first 2 KM intervals; larger early AE‑driven discont on the combo would incrs risk that censoring rather than biology contributes to KM separation. W/ prespec crossover, itt os is expected to be non‑sig at the initial cut bc post‑progression targeted therapy compresses survival diffs; adjment methods (e.g rpsft ipcw) may show directionally favorable effects but are typically not evidentiary. The non‑timing evidence strengthens Scenario A beyond timing... Part 1’s 10.2‑month overall mPFS despite a ⅔ suni‑experrienced mix, the 19.4‑month mPFS signal in 1‑prior‑TKI, and orr 27.5%/33.3% increase the plausibility that, in a purely 2L suni‑naive randomized setting, a low‑teens mPFS W/ hr ~0.7–0.75 is reproducible and at least 1 alpha‑protected secondary can clear sig if alpha was reserved Part 3.1 scenario B 25% - PFS win; OS nonsig: same >13‑month LPI ->topline timing aligns W/ a real PFS effect even if all alpha is allocated to the primary endpoint. W/ control mPFS approx 8–9 months and hr around 0.70–0.80, combo mPFS lies in the low‑teens W/out invoking extreme effect sizes. 2ndary endpts (orr/dor/ttr) may show only nominal improvements aligned W/ prior activity (orr 27.5% overall; 33.3% in 1‑prior‑TKI) but lack evidential weight. safety is the gatekeeper: confidence that PFS reflects biology rises if dose‑intensity is balanced (W/in ~10%), early discont is not concentrated in 1 arm, and early censoring imbalances are modest (more than ~2–3% across initial intervals). given prespec crossover and limited death evts at the PFS analysis, OS is again expected to be non‑sig on iittt at the first look. Part 3.2 scenario C – 15% PFS trend only (CL crosses 1) or fail: short LPI->topline interval would fit faster evt accrual and weaker effect, though operations can confound. if hr >= ~0.85 or the 95% CI includes 1.0, the dataset is non‑confirmatory. contributors include early‑interval censoring imbalance (e.g missed scans or ae‑related drop‑outs concentrated in the combo arm). prior ddi observations in analogous settings describe suni AUCss reductions on the order of −26% to −35%; such attenuation would be expected to accelerate evts (shorten time to maturity), not delay them, and therefore cannot explain a protracted LPI->topline interval. safety wise, higher early discont or larger dose‑intensity gaps in the combo arm both diminish true efficacy and inflate censoring, weakens KM read even if nominal separation appears. Part 3.3 scenario D – 10% PFS win but operational artifacts fuck up credibility: nominal PFS positivity could be driven by nonbi o artifacts: early censoring deltas favoring the combo by several % points in first assessment windows; stepwise KM drops aligned W/ batched visits (q6–12‑wk imaging) rather than continuous hazard; or internal inconsistency where PFS improves but orr/dor/ttr do not move consistently despite adequate exposure. this scenario is less consistent W/ a >13‑month LPI->topline unless the delay is dominated by operations (eg., prolonged bicr queue aetc etc leaning bullish given the rationales mentined above. check references. too lazy to attach. this is a brief look and not exhaustive, so do your own diligence. btw, I took a jerk break after part 1.6. swear on grandma’s life. Thought you should know...

@Sentient_Atoms @maxmarchione Why contraindicated for you forgive the curiosity

@maxmarchione If it weren't vasoconstrictive (contraindicated in my physiology), it would be in my regimen.

@chaotropy Have a follow for building and posting this! (I don't know the tenax story so can't comment on that sadly)

Perhaps the most striking result from my PK simulation of $TENX's oral levosimendan (TNX-103): at modeled steady state, the lowest point in the oral dosing cycle may never drop below the IV peak. Throughout the entire final dosing week, the model suggests oral TID could maintain free effective concentrations above the IV weekly maximum. Looking forward to your feedback, especially from @A_May_MD!

@cagujjubhai @KobeissiLetter Salami has been killed last year already

This is the fundamental problem with US policy on Iran for 45 years. The decision-making structure is deliberately opaque. Supreme Leader Khamenei sets the red lines, but the IRGC Quds Force and Supreme National Security Council execute independently within those lines. Araghchi is foreign minister, not a decision maker on military strategy. Trump calling him a 'fax machine' actually shows some understanding. The real power sits with IRGC Commander Salami, Quds Force head Qaani, and the defense establishment. None of them will pick up the phone. Iran's whole deterrence model is built on strategic ambiguity about who calls the shots. That is by design.

BREAKING: President Trump is trying to figure out who "actually makes decisions" in Iran and how to get in touch with them, according to US officials. Araghchi has been the primary intermediary in the past, but Trump advisers see him as a "fax machine" in talks, officials say.

🦔 Jeff Bezos is reportedly trying to raise $100 billion to buy up old manufacturing companies in aerospace, chipmaking, and defense, then modernize them with AI through his startup Project Prometheus. He's been traveling to Singapore and the Middle East looking for investors. Prometheus launched with $6.2 billion and is focused on AI models for manufacturing and engineering. My Take The guy is worth $200 billion and he's flying around asking other people for $100 billion. That's how you stay rich. You use other people's money for the big bets and keep yours safe. If it works, he takes the credit. If it doesn't, the investors take the loss. The actual manufacturing question is harder. AI can help with design, logistics, supply chain optimization. It can't weld or dig tunnels. Most of the rust belt exists because labor moved overseas, not because someone forgot to install a chatbot. I'm curious what Prometheus actually does that justifies buying entire companies to feed it. Hedgie🤗

@masterlongevity @PTBwrites SF or nationwide?

@PTBwrites 12% of mothers are >40

Nearly one in 10 babies born in San Francisco were born via in vitro fertilization (nationally, the share is around 2%)

I somehow feel that this should be a bigger story in the WSJ…

@IPublius89 @lithos_graphein You must be older than me because my default is 90nm😭

@lithos_graphein Elon will bring back 157nm

I might make some new friends tomorrow.

Tobriand Islander seashells! And a fat stack of "new ¥"...