DeskJockey - Awesome. Yes. More to follow on this but I think that's exactly right.
But first, putting to rest the 'differential risk' piece (flossing a dead horse...) - everybody's right. We're talking past each other. It doesn't affect the paper's core argument about killing=combat.
[Warning: Statistics & Probability Theory follows. Feel free to skip.]
Conditional probability is the probability of x given y. Absolute probability is the probability of x. Differential probability is the increase in probability of x given y.
- The risk claim is that bounding current conflicts, for the sake of argument 2006-present, there is no significant difference in differential probability of general enemy-action-related risk given combat when comparing manned to RPA.
- To rephrase Hacker's argument, if we changed the bounds to include the opening phases of OIF, or a hypothetical opening phase of Iran, then this would not be true. This is very likely the case, as Risk(Intact IADS) > Risk(Creech Attack.) However, the institution continues to provide routine costly signals (combat pay, medals, prestige, etc. - albeit at a decreased rate) after the IADS has been INOP for more than half a decade. Absolutely agree with the pucker factor point. But this logic does not justify the current policy.
- On the core discussion here's what I mean, bounding to current combat zones:
1) Probability of Harm given Manned Aircraft > Probability of Harm given RPA
[Risk from CFIT, Aviation Hazards] > [Risk from crashing due to exhaustion]
2) Probability of Harm given a Manned Aircraft and Combat still > Probability of Harm given an RPA and Combat
[Risk from CFIT, Aviation Hazards + MANPADS/SMARMS, CONUS General Terror Attack] > [Risk from crashing due to exhaustion + Creech-specific Terror Attack.]
3) But Increase in Probability of Harm for Manned Aircraft given Combat ~ increase in Probability of Harm for RPA given Combat
[Risk for MANPADS/SMARMS, CONUS General Terror Attack] ~ [Risk for Creech-specific Terror Attack.]
- I believe most of the controversy is about claim 2, where my argument is about claim 3. Hence talking past each other.
- My point with this was to destabilize the conventional wisdom about 'combat risk' - that the reason RPA should not consider themselves in combat is because they do not experience a comparable increase of risk due to combat in current circumstances. This assumes a difference in baseline risk between RPA and manned aircraft. Therefore I need to make a weak but plausible claim. Given that our adversaries have expressed that RPA is their number one threat, and they have had more success with CONUS-based major attacks than downrange attacks, compared to the difficulties of acquiring and achieving a kill with MANPADS/SMARMS, it is plausible that the differential combat risk for being in the population of RPAs is higher than the differential combat risk for being in the population of manned aircraft.
- Unless I'm forgetting something, there are no data points for 10k+ aircraft crewmen in this timeframe lost to enemy fire, and no data points for a Creech attack. Therefore we fall back on weak Bayesian priors for risk in both cases, which are insufficient for a strong inference of difference. Accordingly, another definition is required. The fact that there are no data points is in both cases a tremendous credit to OSI (and SF.)
- As previously noted, this was unclear in the article, and regrettably detracted from the core argument. I agree with this critique.
[End stats.]
On the point brought up about losing friends - These are my friends as well and I have mourned them. The only way I know to honor them is to fight all the harder for the things they sacrificed for, and from what little I know, that is best done by fighting smarter and thinking harder about how to do things better.
Building a sustainable Pred community provides the tactically proficient operators that will finish dismembering what is left of our enemy's network in this war. It also builds the expertise we will need to ensure that whenever a future manned striker punches through the FEBA, it will do so with a flight of drones alongside - a swarm of robots willing to take a missile to save the crews' lives will bring more of our comrades home. Incentives matter because culture matters. Culture matters because performance matters.
Medals may or may not be the right answer. But the status quo with 60% to O-4 at Creech and minimal RPA DOs picked up for command certainly is not. Whether this is due to prior adverse selection or due to present conditions is irrelevant - in the first case we shortchanged the program from the outset, in the latter case we are shortchanging it at present - either way we're behind. For all of the deep problems that remain a decade and a half into this project, its in all of our interests to fix this. If it falls apart, it won't go away - the Joint community and the DoD absolutely won't let that happen, even if it means we bleed out the pilot corps again. Far more importantly, letting RPA flounder because of tribalism (i.e. the 'new navs' comment) lets the mission down and kneecaps capabilities that will ultimately come back around to helping out the manned community.