[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@varma_ashwin97 "@dwarkesh_sp @jacobkimmel @jackscn X continued. We have to do more tests and more observations per patient driving up cost per patient. Moreover the trials are larger. Go look at trial sizes in the 1980s compared to today. The average trial size is much larger higher total costs"
@varma_ashwin97 on X 2025-07-19 01:38:18 UTC 2064 followers, XXX engagements

"@thekaransinghal @OpenAI @PendaHealth Excellent work More real world testing is needed now that LLMs have largely saturated MC and curated case environments in medicine. Did you randomize who received the LLM copilot and who did not"
@varma_ashwin97 on X 2025-07-22 21:54:35 UTC 2063 followers, XXX engagements

"@arohan Hm if they used the basic reasoning model that seems perhaps different than what OpenAI did which appears to be based on some highly experimental techniques"
@varma_ashwin97 on X 2025-07-22 04:12:08 UTC 2063 followers, XX engagements

"So Eroom's Law is absolutely not indicative of a failure of the biopharmaceutical industry. It could very well be indicative of its success"
@varma_ashwin97 on X 2025-07-20 10:53:35 UTC 2063 followers, XXX engagements

"@dwarkesh_sp @jacobkimmel @jackscn X. You should be *** extremely *** skeptical of this trend changing. It is a trend that has resisted an explosion of new knowledge and new techniques. Sequencing Crystallography HTS Biologics Recombinant Tools nothing has changed this"
@varma_ashwin97 on X 2025-07-19 01:57:41 UTC 2063 followers, XXX engagements

"New biology and new tools expand the pool of possible projects and makes it more likely you find superstar drugs (e.g. Solvaldi). If so the rational response by a planner is to advance more candidates and tolerate a higher failure rate (and thus lower NMEs/$)"
@varma_ashwin97 on X 2025-07-20 10:52:34 UTC 2064 followers, XXX engagements

"@dwarkesh_sp @jacobkimmel @jackscn X. Oh yes one more thing productivity maximizing = welfare maximizing Sometimes you want to sacrifice productivity for more shots on goal"
@varma_ashwin97 on X 2025-07-19 23:32:03 UTC 2063 followers, XXX engagements

"A random point. If you model "advancing technology" (new biology new drug discovery techniques) not as a scaling of a research productivity parameter but as increasing the variance in outcomes from projects then it's easy to write down a model that replicates Eroom's"
@varma_ashwin97 on X 2025-07-20 10:52:34 UTC 2064 followers, 2188 engagements