Spherically senseless
Expecting the unexpected
There is a great line in Gary Taubes' latest book, a term from physics that describes an idea that makes no sense, however or from whichever direction you look at it: spherically senseless. I loved that concept, imagining the pure sphere of an idea, surrounded by its atmosphere of hypotheses, rationale and beliefs. It's possible that some of those layers have some truth, some insight, but shield an idea that doesn't stand up to any scrutiny.
As I imagine that sphere, I drop in some of our industry's own ideas. Into that perfect space, I can place many, but in the theme of this blog the one that fits best is not just the TPP, but the eNPV it produces. While I've covered the TPP a few times recently, this one will emphasise the role that the eNPV plays in corrupting the value even of that often unwitting fool.
We need a value. For many reasons, especially finite resources, we need to give ourselves a way to estimate opportunity and risk. But, and here is the point, we signally fail to recognise that picking one TPP (or two...) and estimating its value over 10-20 years is an exercise so filled with error that even thinking it is a good way to estimate opportunity and risk is senseless. Remember, we're not taking a target opportunity profile and working back ('7 billion people need a vaccine and it needs to be 'x effective' with 'y safety' and cost 'z dollars' and we know that we're the only ones looking at that opportunity and we can deliver it today) - we are taking a molecule about which we know almost nothing, and projecting the slings and arrows of fortune of one almost-randomly chosen path, and a market in 5-10 years about which we know literally nothing. We'd think ourselves foolish to project the outcome of a snakes and ladders game, but somehow we trust a 'positive eNPV' in phase I.
Think back to the '1000 prototypes between each race' of last week's post: all built and tested against a model of the future in a particular circuit at a particular time a couple of weeks hence, and the ones to come - with numerical, physical values to be gauged against. They still make mistakes. Which is what happens even when you build 30,000 prototypes per season. But pharma? No, one is fine for us, thank you. The estimate we cram into the 'e' of the eNPV has to do all the heavy lifting: industry averages, therapeutic area averages, someone's guesswork, someone else's best-evidence guesswork and more, all mushed together into an opaque 'expected' value.
Except we don't expect the 'expected'. Even McKinsey acknowledge pharma forecasting is like a monkey throwing darts (always wrong, and wrong in both directions equally), but then every organisation responds by increasing the number of dart throwers and hoping that somehow it will increase signal and reduce noise... And so, we build our portfolio views armed with numbers we know are wrong (if we're honest), but build our projections from them anyway ("it's all we've got..."), assuming that some finessing or body English from leadership will steer a path through it all...
The TPP and the eNPV cannot work. Physically, mathematically, stochastically cannot work. By design they can't work. In use, we know they don't. Like Ptolemy and his epicycles, the industry invents all kinds of ways to explain away this situation, but it might be easier to accept the truth: the TPP and the eNPV, as applied throughout our industry, are spherically senseless.