What happened yesterday is easy
Can we prevent bad things from happening tomorrow?
Intro
I’m writing this on the train on the way to meet combat reservists. I volunteer in the MiluimTek program. We drink coffee and I listen. We talk about tomorrow.
Today, I want to talk about how medicine and tech are fundamentally different and how we can do a better job for patients by doing better work in life science R&D.
Today I write about factual and counterfactual reasoning.
Let’s get started.
Factual reasoning
Factual Reasoning works with what is - analyzing present data, and observable conditions to draw conclusions.
Drug and device companies and their investors are very good at factual reasoning.
They reason about present and past using protocols, clinical data and regulatory guidance.
Drug companies write protocols based on previous studies to model safety, toxicity and efficacy.
Device companies do hazard analysis and manage QA/RA.
They both submit to the FDA, report serious adverse events, and perform post-market safety surveillance.
These are examples of factual reasoning that asks:
What bad things happened in the past, and how do we prevent them in the future?
Drug and device industries spent decades perfecting this kind of reasoning.
Highly bureaucratic — and well suited to checklists, reviews, and regulatory submissions.
The hard part—the one that quietly destroys enormous amounts of value for medical innovation, has almost nothing to do with safety failures or regulatory compliance gaps.
It has to do with paths taken today that result in failed futures. To change the future, we need counterfactual reasoning.
Counterfactual reasoning
Counterfactual Reasoning explores what could be under different conditions - imagining alternative scenarios to understand causal relationships and hidden risks.
Tech founders and investors are very good at counterfactual thinking.
Venture capitalists live in counterfactuals:
If the team were different, would this work?
If the timing were better, would this be fundable?
If we pivoted from consumer to enterprise, what would change?
If the market re-rated, would this outcome look obvious in hindsight?
This kind of thinking is core to venture investing. It’s imaginative, flexible, and scenario-driven. It’s a big reason why venture works so well in software and consumer tech, where bets are big, pivots are cheap and reversibility is high. In tech, we assume we can get off the wrong path and pivot to product-market fit.
It’s part of the magic of TechBio for investors since TechBio companies can generate revenue quickly with platform licensing with tech-style mindsets in a $1.5TN industry.
But what happens for a life science company that travels a 7 year regulated path to market?
What happens when they discover that they are irreversibly incapable of commercializing their invention?
How can a life science founder think about counterfactuals ?
Can they think about alternative paths with abject failures, walking dead and XXL exits?
Time is not in your favor
In life sciences, decisions about the path to market in 5-7 years are taken in the present.
The present is a perfect time to make irreversible decisions because you wanted to please investors with optical progress.
Consider the choice of a regulatory pathway.
Do you choose 510(k) (Premarket Notification) because you found a predicate device or do a PMA (Premarket Approval) which requires clinical trials and data to prove safety and efficacy? Did you consider the value of clinical data as a core IP asset in a future exit transaction?
Consider powering for safety.
Maybe your pivotal trial is underpowered for safety because you reduced the number of subjects to save money.
With over 15,000 clinical trials performed every year, and only 5% of US clinics recruiting, it would be hard to claim to your investors that taking the best path for a future exit was an unknown unknown.
Venture is optimized to price risk under uncertainty:
outcomes vary
variance is acceptable
options remain open
reversibility is assumed until very late
But life science decisions are often path-dependent, not risk-distributed.
Once a path is chosen, you might reduce your risk today at the expense of missing a $100M exit in 7 years.
You can choose a path, reduce near-term risk — and silently eliminate the possibility of a meaningful exit
The issue isn’t that VCs fail to imagine alternative futures. They’re often excellent at that.
The issue is that they struggle to convert those counterfactuals into calibrated, therapeutic-specific paths with irreversible consequences.
In software, you can pivot and with AI-augmented coding the cost might be less than a month.
In life science, you may not be able to afford a pivot given the cost is 5 years of operations and clinical trials.
Most of the analytical machinery in life science and pharma —regulatory, clinical, financial—is built to reason about the present path.
That’s why so many post-mortems sound the same:
the science worked
the product was safe
the team was strong
the timing was unlucky
What’s rarely articulated is that the company took a short-term tradeoff and went down a path which crippled the long-term revenue capacity; often in ways that made success impossible even if the science worked.
Why This Matters
If we want fewer dead-end products, fewer stalled companies, and fewer “how did this fail?” stories, the answer isn’t more imagination or better storytelling.
It’s better reasoning about paths, not just risks.
That means learning to ask a different question alongside the familiar ones:
Not just “What could go wrong if we proceed?”
But “What becomes impossible if we do?”
Safety frameworks protect patients but they don’t protect company futures.
And in life science, it’s the futures we casually eliminate that end up costing the most.
The tragedy isn’t that these decisions are irrational — it’s that they’re rational under the wrong model of time.
Outro
In next week’s Friday essay, I’ll walk through three real cases where factual reasoning was perfect but counterfactual blindness destroyed the outcome - and share the framework I use to stress-test regulatory and commercial paths before they become irreversible.
Subscribe to get it when it drops.
This Week on Life Sciences Today
My guest this week was David Bates, CEO of Linus Health.
I talked with David Bates, scientist, investor, and CEO/founder of Linus Health, about transforming brain health from late-stage “sick care” to proactive prevention.
Linus Health’s AI-driven brain health platform helps payers detect cognitive impairment early and monitor members over time.
David traces his path from signal detection and multimodal sensing at Georgia Tech, through investing in built-environment technologies, to the insight that behavior is the primary observable output of the brain. Linus Health uses rapid, tablet-based assessments to capture thousands of digital biomarkers and analyze the process of how patients perform tasks, enabling objective, early detection of cognitive change and replacing hours of traditional neuropsychological testing.
David explains why primary care and health systems are their core customers, how the platform fits fee-for-service and value-based care, and why diverse longitudinal data is their core moat.
We close by exploring a future where clinically integrated, consumer tools make brain health monitoring continuous—like a “blood pressure cuff for the brain.”
Visit Linus Health here
You can see the episode here - Transforming brain health from late-stage “sick care” to proactive prevention.
About me
I’m a writer, host of Life Sciences Today podcast, ex-pharmatech founder, father of 4.
I’ve been building in tech, cyber, privacy, and clinical data for 25+ years across Israeli medical device startups, Verily, Amgen, and the Fortune 1 company. I work at the intersection of engineering, AI and clinical data.
If you love my writing — share it
If you want more like this — subscribe to Clear Thinking