Affordably Unmanned: A Cost Comparison of the MQ-9 to the F-16 and A-10, and a Response to Winslow Wheeler's Criticisms of the Drone

Every so often an analysis so bad comes along that I feel compelled to comment. Perhaps I'm moved to defend the honor of those who built and bought the weapon in question, or maybe I'm just appalled that anyone can publish stuff that's so misleading. Today I'm referring to the recent series of articles by Winslow Wheeler on the MQ-9 Reaper drone, recently published on the website of the Project On Government Oversight (POGO), in the journal Common Defense Quarterly, and at Time Magazine's Battleland blog. The title of the CDQ article sets the tone: "MQ-9 Reaper: Not the Revolution in Warfare You've Been Told". For links to all of this, POGO has provided a single page on the topic:

http://www.pogo.org/straus-military-reform-project/20120615-common-defense-quarterly-article-on-drones.html

What’s wrong with the work? Wheeler's study purports to be a cost comparison of the MQ-9 to the F-16 and the A-10, and a highly unfavorable one for the drone, but it fails to follow basic standards of financial analysis. If you bought into this stuff, you'd draw deeply flawed conclusions about the three aircraft, and what they're useful for. So I needed to take some action.

I will not deal with Wheeler's assertion that a Cessna would be less expensive to operate than a Reaper on border patrol duty. His prima facie case seems reasonable. Just note for now that the Reaper’s predecessor, the Predator, got its developmental boost from the desire for casualty-free, constant, medium-altitude aerial reconnaissance over Bosnia. As I will note later on, this is critical in any comparison of the Reaper to a manned aircraft. Somebody needs to think about those pilots. Reapers are meant to provide combat air patrol (CAP), a form of constant overwatch for surveillance, and when warranted, attack. All the while, though, they're supposed to keep the aircrews, such as they are, out of harm's way.

The essence of Wheeler's argument is this: an F-16C costs $4.8 million dollars annually to operate, an A-10C costs $5.5 million, but a flight of four MQ-9s costs $20.4 million. The MQ-9s have a higher loss rate, and require more people to operate. Since they have no meaningful defense capability against other aircraft, only an idiot would buy them.

Unfortunately for the folks at POGO who just hired him, and for the folks at Time who continue to publish him, there's just nothing truthful about this. Indeed, from where I sit, what he has written looks intentionally misleading. A real cost comparison, though, would be very valuable, so I'll construct that in the paragraphs that follow, and demolish Wheeler's work as I go.

Wheeler's work starts out a bit dodgy, when he computes the price of a Reaper as $30.2 million. This is not an historical or current cost of a Reaper, but what the USAF projects as the average price of Reapers over the course of the duration of the decade-long program. As the article seems to be comparing actual USAF aircraft (existing MQ-9s) to actual USAF aircraft (existing F-16s, which I will take as Block 52s), we should use current prices. After all, we're not betting on what may come, when Reapers may fly with yet-more impressive payloads, such as the much-vaunted Gorgon Stare (which Wheeler doesn't much like, but that's another issue). Rather, we're evaluating what the USAF has in inventory. In that context, using current numbers is rather standard managerial accounting practice.

In the entire program so far, the USAF has spent $340.8 million on RDT&E and $2,110.3 million on procurement of the actual aircraft. That's just $21.09 million per airplane—including ground equipment—in capitalized costs. As the Reaper comes with program-specific construction funding, I argue for adding the MILCON funds. Airplanes do need hangars, and for some reason, the USAF doesn't think that all the leftover F-16 facilities are sufficient. These monies, however, seem to be front-loaded for the entire program, so I divide the entire program's expected MILCON funding ($47.2 million) by all 399 projected aircraft for an average figure of $0.39 million each. Overall, that's just $22.48 million per aircraft, and rather less than Wheeler's view.

Next, as the USAF is no longer buying new F-16s, we need to rethink Wheeler's rough figure as the cost of one more serving F-16. He cites a USAF general’s congressional testimony of $55 million, but the backing for this price is not provided. Guessing like this is not necessary. From a recent deal with the Iraqi Air Force, we know that Lockheed Martin has recently been selling F-16 Block 52s (plus some associated ancillary equipment) for just $47 million [1]. These are to be a mix of -C (single-seat) and -D (two-seat) model F-16s, so we might take this as a rough average for Block 52s. It is hard to say exactly what price the USAF could negotiate for its own aircraft, but presumably it wouldn’t be drastically worse. That actually makes Wheeler’s comparison look better, but as the number is more grounded in actual facts, it's the right thing to do.

Thus we find that the purchase price of an MQ-9, including its share of associated ground equipment, at $22.5 million, is less than half that of an F-16 Block 52 ($47 million). But then, as I noted above, Wheeler bizarrely chooses to compare the cost of four MQ-9s to that of a single F-16. He seems to argue, if that's the word, that because the four are grouped to form a single CAP, they are somehow only together as useful as a single F-16. He provides no real rationale for that, and indeed, as I will note below, a single fighter jet cannot even begin to provide a CAP as the USAF describes it. Unfortunately for Wheeler's work, it is this mistake—or obfuscation—that leads directly to his insistence that the MQ-9 is somehow more expensive than the F-16.

It's worth noting as well that Wheeler also compares the cost of four MQ-9s to the cost of a single A-10 Thunderbolt II (Warthog), and again, provides no sound explanation for the multiplication. He cites the US Government Accountability Office's figure for the cost of an A-10A as $18.8 million in today’s (fiscal year 2012) terms. It's important to remember, of course, that this aircraft is not actually available for purchase, as the A-10 hasn't been manufactured in over twenty years. New models aren't an option, so the question is not whether new A-10s would be ordered, but whether new MQ-9s perform a particular mission more cost-effectively than old A-10s. As the USAF has a huge inventory of F-16s (over a thousand operational aircraft), the same could be said for that airplane. The real-world cost comparison, then, would focus on two questions:

1. whether the recurring costs of MQ-9s exceed the recurring costs of either manned type over an expected lifetime (if we are interested in which should be retired first)

2. whether the non-recurring and recurring costs of MQ-9s exceed the simply recurring costs of either manned type over an expected lifetime (if we are interested in whether more MQ-9s should be procured to substitute for either F-16s or A-10s)

Answering these questions requires delving into the cost elements and the cost drivers. Wheeler provides some reasonable base numbers, but he badly misinterprets what they mean.

First, he names the hourly flying costs of the three aircraft as $17,780 for an A-10C, $20,809 for an F-16C, and a mere $3,624 for an MQ-9. He offers to provide the supporting data on request, but he cites no specific documents. After some checking, I found these figures to be reasonable [2], but requiring some adjustment. That's not unusual in financial analysis, but one actually has to do the work.

To begin, the major components of flying costs, by the definition given in the USAF's Selected Acquisition Report on the MQ-9, are repairables, consumables, maintenance contracts, fuel, and contracted manpower. The first four are factors largely variable with flying hours. Military manpower is largely invariant in the short run; contracted manpower is much more variable, though with some dependence on the contract terms. Either way, we'll need to pull manpower costs out, and later treat the whole of the contracted and military costs together, in order to compare fixed costs to fixed costs, and variable costs to variable costs.

For some unknown reason, the USAF also includes a small margin per hour ($66) for something called "continuing system improvements," which really doesn't sound like an operating cost. If we exclude this and the manpower costs from the figuring for now, the USAF's SAR shows that the variable part of the cost per hour is just $2,170 in 2008 terms, or $2,354 in today's terms (again, with that rough-but-reasonable CPI adjustment).

For his part, Wheeler doesn't tell us how many hours each of those aircraft are supposed to fly in a year, but as noted, he did provide (again, without reference) those figures for the annual operating cost of both manned aircraft: $4.8 million for an F-16C, and $5.5 million for an A-10C. With these, and the hourly costs, we can back out each type's approximate expected annual flying hours: 231 for an F-16C, and 309 for an A-10C. Again, those seem reasonable, so hang onto those numbers.

First, though, note how Wheeler asserts that “because each Reaper flies a large number of hours in the air, the math suppresses the per-hour Reaper number.” In effect, he is saying that the annual operating cost is more relevant than the hourly operating cost, because the Reaper flies a lot. This is nonsense. Flying costs are specified as variable costs, so they can't be suppressed by an allocation over a larger denominator. Indeed, the whole point of a patrol aircraft is to fly, and to fly a whole lot. If the Reaper flies more, then the USAF is getting its money's worth from the plane.

Indeed, it would seem that it is: the USAF's objective for a CAP, as described in the SAR, is 7,300 hours per year. As there are only 8,760 hours in 365 days, that's an average of one Reaper from the set of four airborne 83 percent of the time. That has every Reaper in the inventory spending 21 percent of its lifetime airborne. I'll get back to that point, to check whether they're actually attaining this figure, but note for now that it's a rather impressive number.

So indeed, if a CAP is what we seek, then CAPs are what we should compare. How much would constant overwatch cost with either an A-10C or an F-16C/D? I say would because no on tries to do such a thing. As I will show below, the very project would be insane. But let's play along for the moment. We do know that the MQ-9's on-station time is greatly superior to that of either manned aircraft, and largely because it is indeed unmanned. All the weight-bearing options that might have been required for a manned aircraft (cockpit, cockpit armor, ejection seat, radar warning receiver, chaff, flares, speed, etc.) seem to have been left off the MQ-9, producing an airplane optimized for slow-going, fearless patrol. The superior speed of the F-16 will allow it to get to the scene of trouble faster than an MQ-9 could, but this simply argues for a mixed fleet of fighters and drones, not a pure fleet of fighters, as Wheeler seems to prefer.

Moreover, we have two solid reasons for concluding that a CAP of either manned aircraft type would require more than four airplanes. Because the station-keeping time of the F-16C/D and A-10C is inferior to that of the MQ-9, it is logical that more aircraft would be required, as aircraft would need to be transitioning from station to airfield more frequently. It is also notable that both types tend, in USAF service, to fly in patrols of two aircraft at a time. Presumably this is so that should one aircraft go down, another is available to report and provide overwatch until a combat search-and-rescue (CSAR) aircraft arrives. Of course, with a drone, there is no need for CSAR, so no need for a second airplane.

It is true that the manned aircraft can use their better speed to reach their patrol areas more quickly; this, after all, is why the US Navy chose jet-powered P-8s as the replacement for its turboprop-powered P-3s. But even with this advantage, no one will claim that a P-8 can stay on station as long as a P-3, and no one should claim something similar in comparing turboprop drones to fighter jets. But forget all this. Let us give the benefit of doubt to the F-16 and the A-10, and assume that somehow we can manage a CAP with only four jets. Still, in this case, each airplane will need to fly roughly the same number of hours as each MQ-9, because the goal of the CAP is overwatch. This is an unrealistically aggressive assumption for the jets, but the foolishness will accentuate the comparison in the end.

We next need to calculate the labor costs. Americans are suitably expensive, and Wheeler complains that a four-ship CAP of Reapers requires 171 people to operate. This may be strictly true, but it is not helpful to the comparison. According to the very briefing that Wheeler cites, fully 66 of those people do not serve the aircraft, but the intelligence system behind it. If a CAP of F-16s or A-10s is to achieve the same objectives as a CAP of MQ-9s, then presumably the aircraft would be outfitted as makeshift RF-16s or RA-10s. With signals and photo intelligence payloads similar to those on the MQ-9s, we can imagine that another 66 people would be required to process the same data and imagery. So for accurate cost comparison, we must leave those 66 out, and use 105 people as the flying and maintenance staff for an MQ-9 CAP.

So how many are required to operate a four-ship mini-squadron of F-16s or A-10s? It's hard to say, but we can use a recent USAF staffing decision as the source of a ratio. In the reassignment of the 132nd Fighter Squadron of the Iowa Air National Guard from F-16s to MQ-9s, the Air Force Department is cutting the 761 jobs directly associated with the squadron's 21 F-16s, and adding back 269 staff for the drones [3]. That's a net loss of 492 people, but it's the gross loss that's interesting: 761 people are needed for 21 jets. That works out to 36 and a quarter people per airplane, or 145 people per four-plane unit. Given the economies of scale that the USAF cites in grouping its aircraft into comparatively large squadrons and wings, that's probably an understatement of what's required, but again, we'll give the jets the benefit of the doubt. A four-ship CAP of F-16s merely requires 40 more people than a four-ship of MQ-9s.

That's not shocking, as one can imagine that heavier, faster jets might require more maintenance than lighter, slower turboprops. But at roughly $100,000 per person annually [4], the four F-16s will require an additional $4 million in annual direct labor costs. (That's $10.5 million for the MQ-9 staff, and $14.5 million for the F-16C staff.) While I don't have a recent figure for A-10Cs, it's probably not far off.

The labor figures are also useful for estimating the variable costs of flying the manned aircraft. Labor costs of $14.5 million for a four-ship CAP work out to $3.63 million per aircraft. That leaves variable costs of $1.18 million for an F-16C, and $1.87 million for an A-10C. Dividing through by the average annual flight hours yields variable flying costs of about $5,108 for a Falcon and $6,050 for a Warthog. These costs for the manned jets are roughly twice those for the unmanned turboprop, and as a rough ratio, that passes the taste test. They're jets, after all.

Losses should indeed figure into our figuring, but while Wheeler claims that Reapers are crash-prone, he provides no specific evidence. Rather, he just asserts that if the USAF is aiming to buy 396, but only plans to hold 256 in the operational inventory by 2017, then it must be forecasting the loss of 140. This is more nonsense. Today, at the behest of former Secretary of Defense Robert Gates, nearly all the available Reapers are flying over Afghanistan. That is not normally the way the USAF or any other air arm operates; a considerable margin of aircraft will eventually be reserved for training, depot maintenance, testing, and war reserves. Some margin would be forecast as attrition losses, but Wheeler offers no evidence that this must be the entirety of the marginal 140.

Wheeler would like a tail-number inventory, and that might be nice. A lower-cost approach would be a little secondary research. Just this February, Defense Tech reported that the USAF had lost seven Reapers to all causes since the program started flying in 2007 [5]. As Wheeler himself is quoted in that article, I would have imagined that he would have seen that figure. If we call February "through the end of 2010", thus slightly understating the hours, we have seven losses in 125,821 cumulative flight hours (another figure from the 2011 SAR). Note how that's all losses, combat and otherwise. That's one loss per 17,975 hours; as a single Reaper is supposed to fly 1,825 hours per year, the average aircraft lifetime on intensive operations can be taken as 9.85 years. That would be short for a jet fighter that mostly sits in the hangar, but with a turboprop that flies constantly, the USAF may be getting its money's worth.

That is, while his mysterious multiplication by four was bad, almost as bad is Wheeler’s treatment of availability. He cites a mysterious document of no clear authorship in claiming that the USAF’s 69 operational Reapers flew 97,727 hours in 2011. This works out to 29.5 hours per aircraft per week, a figure that he decries as just two flights weekly, since each aircraft must be flying at the maximum fourteen-hour mission. Actually, 29.5 hours is 17 percent of a week [6]. In missing this point, Wheeler has implausibly asserted that the Reapers aren't flying enough.

Admittedly, 17 percent a tad below the objective of 21 percent, and it's not quite enough for a four-ship to provide 24/7 coverage. With proceed time, it could be more like 12/7 coverage. But honestly, I don't know of any other military aircraft that spends 17 percent of its life airborne. Maybe tankers do in really intensive operations. Certainly F-16s and A-10s don't. By our estimates above, the fighters spend 3 to 4 percent of their time flying. Indeed, they can't do much better. The F-16 airframe is only rated for 6,000 hours, and even if rebuilt, really should not fly past 10,000 hours. At the usage rate of the MQ-9 (1,825 hours annually), any F-16 would be a total write-off in 5.5 years. That's just over half the time in which we'd expect to lose the MQ-9 to an accident. So which airplane has the unacceptable loss rate?

Oh, and incidentally, flying F-16s or A-10s that much would require a lot more than 145 people per four aircraft unit. For one thing, how many pilots could actually fly that much? That is, after all, the whole point of remote operation. Crew rest is just a whole other concept in the cubicle. But we'll ignore that for the moment.

So, in wartime usage, how do we put all these numbers together, to compare MQ-9 CAP costs to F-16C or A-10C CAP costs? Remember that the recurring, annual costs are labor and variable flying costs. For the MQ-9s, the USAF needs 105 people; for the year, that's about $10.5 million. For 7,300 flying hours, the USAF would pay $17.2 million in variable costs. For the F-16s, the service would need 145 people; for the year, that's about $14.5 million. But for 7,300 flying hours, the Air Force would pay $37.3 million for the F-16Cs, and fully $44.2 million for the A-10Cs.

So much for the recurring costs, which are massively in the MQ-9's favor. But what if we add in the nonrecurring costs? Buying another four MQ-9s, with today's capabilities and at today's prices, would cost about $89.9 million. (That's subject to negotiation with General Atomics, of course, but it's a decent approximation.) Amortized over an expected 9.8-year life, that's about $9.2 million in annual depreciation costs. That is, if you want to keep flying Reapers after the campaign, you'll need to pay something approximating the depreciation as a recapitalization cost, to buy new aircraft.

One might presume that there's no equivalent cost for the fighters, as they are already in inventory, but this is not quite so. For the manned jets to perform like the drones, the USAF would be flying the wings off them. With an expected life of 5.5 years, the F-16 unit would require recapitalization nearly twice as fast as the MQ-9s, and the annual amortized charge would be almost $8.6 million per aircraft, and fully $34.1 million in depreciation for the whole set. There would be a slight difference for the time value of money, but this isn't worth describing. Oh, and note that this only considers the wear-and-tear on the fighters, and not the occasional losses due to accidents. Those have taken 36 F-16s and 8 A-10s out of the inventory since 2001 [7].

As noted above, because the A-10C is out-of-production, I won't define recapitalization of a worn-out fleet as an option. But the comparison is still stark. For a 7,300-hour per year four-ship CAP, the estimated costs for MQ-9s are $10.5 million in manpower, $17.2 million in variable flying expenses, and $ 9.2 million in depreciation, for a total of $36.9 million. The estimated costs for F-16Cs are $14.5 million in manpower, $37.3 million in variable flying expenses, and $34.1 million in depreciation, for a total of $85.9 million. At more than twice, that's a marked contrast.

Now, someone might note that depreciation is a non-cash charge, and thus assert that it's not a real cost. That's true, if one just wants to fly the F-16Cs into the ground, and not bother to replace them. Heck, with a thousand around, why not? Even then, the annual cost of operating and replacing four MQ-9s on a CAP ($36.9 million) is less than operating and dumping four old F-16Cs ($51.8 million). Again, there's just no comparison, and actually no way to treat the fighter fleet.

Now take a deep breath, and remember that very thought of comparing these aircraft is silly, because no one is crazy enough to try routinely to put 1,875 hours annually on a fleet of F-16s.

And yet, as a last ditch effort to save the argument, we might even get the claim that this comparison would only apply in these high-tempo, wartime conditions. In peacetime, it would be different. Really? Arguably, the F-16 aircrews would still need to get in their 200+ hours to maintain proficiency. How much flying is required for Reaper aircrews to maintain the same? Possibly zero. That’s right. The aircraft can pretty much be stuck in a box with the label “Open in Case of War”, as the aircrews do not need actual aircraft to train. Their operational experience is undertaken completely in an office space; training on a simulation of the flight hardly differs from training with an airborne aircraft. But even if actual flying is desired, somewhere the FAA will let them fly, the per-hour cost of the MQ-9 is so much lower than that of the jets that it's still clearly the better choice as an on-hand, just-in-case, overwatch aircraft.

In short, there’s no comparison of the MQ-9 to the A-10C or the F-16C: for the mission for which it was designed, persistent surveillance and on-call strike, the MQ-9 is clearly the better buy. Revolutionary? It feels like it, but even if not, it comes with a solid business case. Are my numbers spot on? Possibly not, but they're likely materially accurate, at least when the competition is "now multiply by four." In short, including these aircraft in the force structure is good idea simply to save unjustified wear-and-tear on the fighters, which might actually, someday, again be needed for the big war.

Here at the end, you might be thinking "wow, that guy clearly spent some time on that." And you'd be right. But this sort of thing is important in our business because nonsensical numbers often get bandied about in half-baked analyses when someone with an axe to grind needs an argument, and can't make the argument he wants. Wheeler's real intent in shown in the second sentence of his Common Defense Quarterly article, in which he holds forth on "what little has been written about them as physical systems, rather on the morality of their use..." The fact is that Wheeler doesn't like drones just because they're drones. He probably knows, though, that plenty of folks are rather taken with them because they're robotic—what better to use to terrorize the terrorist than an all-seeing eye with a Hellfire? So, instead, Wheeler has concocted a hack job to go after them.

If this keeps up, I think that I'll need to set up my own one-man think-tank too. Maybe I'll call it the Project On Gadfly Oversight. The logical URL has been taken, but there's not much there anyway.

Jim +1-512-299-1269 www.jameshasik.com

—————

NOTES

1. See the MQ-9 article at Defense Industry Daily on 5 December 2011.

2. As I prefer audited figures, I pulled the cost estimate from the USAF's 2010 Selected Acquisition Report (SAR) on the MQ-9: $3,051 per hour in 2008, and inflate this at the general CPI rate to 2012, for $3,310. That's not perfect, but it's close. (As the A-10 and F-16 are too old to have SARs, I'll have to take Wheeler's numbers there as workable.)

3. William Petroski, "Air Force seeks to cut 492 Iowa Air Guard jobs," Des Moines Register, 6 March 2012.

4. For this rough but appealingly round figure, I rely on David S.C. Chu's briefing on military manpower costs at the 2008 Western Economic Association Conference in Honolulu. Chu was totaling up all current military personnel costs, including salaries, bonuses, housing, and healthcare. He was omitting, incidentally, those amazingly unfunded military retirement costs.

5. Christian Lowe, "Drone Losses Debate," Defense Tech, 9 February 2012.

6. Of course, we really should use audited numbers, and not those which can’t be attributed. So, to check Wheeler's figures, I used the Pentagon's 2011 Selected Acquisition Report for the MQ-9. By that document, the Reaper fleet had accumulated 125,821 cumulative flight hours through 31 December 2010. By 4 February 2011, the USAF had taken possession of 65 aircraft, but only 43 were operationally active, and in 16 CAPs. Initial operating capability was achieved in 2007, and General Atomics' production is understood to have been rather steady. Thus, with lag time from acceptance to operations, it is conservative to assume that the USAF started with zero MQ-9s at the beginning of 2007, and brought another 16 on line in each of years 2007, 2008, 2009, and 2010. This straight-line forecast provides an average operating inventory of 21 aircraft over 48 months. Dividing through, those 125,821 hours work out to 124.8 flying hours per airplane per month, or again, 17 percent of each month.

7. Otto Kreisher, "The Aircraft Losses Mount," Air Force Magazine, February 2010.

Comments

[Joe Katzman]

Hi James. Very good analysis. The one thing I'd mention is that the general is probably right about F-16 buy cost.

Iraq's cost for downgraded F-16C/D Block 52s is $46 million per only if you don't mind leaving things like radars out of your F-16. Those items are contracted separately, and add up fast.

[James Hasik]

Ha! I missed that. Thanks for the insight, Joe. I suppose that one *could* resurrect that idea from the early 1990s for an A-16, without a radar, as that might have a point in this context. But really, the next time I run the numbers, I'll use the general's figure.

[Curt Steigers]

Excellent analysis but I think you missed a few points that further make the MQ-9 economical for its mission set.

1. Although you said the analysis part is a wash, it really isn't. The MQ-9 can piggyback the sensor information on the installed satcom datalink. The F-16 would have to add the datalink either to the existing F-16 airframe or in a pod. In any event, it adds some additional cost to the F-16/A-10 to meet the capability provided by the MQ-9.

2. And while we are talking about datalinks, the flight personnel in the F-16 need to be in country or at least in nearby countries while the MQ-9 flight personnel are at homestation. As a result, the MQ-9 flight personnel cost a whole lot less to include flight to and from theater, billeting and support in theater, personnel equipment, medical in theater, etc. How much additional cost does it take to support one person on the other side of the world?

3. You mentioned continuing training costs but failed to mention initial training costs. An F-16 pilot has a much longer and more expensive training pipeline to get to his first squadron. While MQ-9s are currently largely flown by personnel trained on other manned aircraft, ultimately they can be flown by personnel with much lower cost flight training, probably on the order of millions of dollars per person less. Whats the difference in training costs between basic flight training in a T-6 and MQ-9 pipeline versus basic flight training in a T-6 followed by T-38s and F-16 transition time? Plus, the opportunity cost of the longer pipeline for the F-16 pilot needs to be figured in. If the average F-16 pilot took a year to get through to his/her squadron and the average MQ-9 guy gets there in 6 months, I potentially get 6 more months of squadron time out of each MQ-9 guy in a career. Of course, you may be able to extract a longer commitment from the F-16 guy which justifies the longer pipeline.

4. Flight crew utilization. Any number of common aches and pains can down a manned aircrew, a UAV pilot not so much. And things like water survival are not so critical. So each pilot can fly both more hours and more reliably meet his flight schedule. So you need a lower standard of redundancy to ensure you meet requirements, ultimately reducing MQ-9 pilots to do the same mission. Not to mention that since the MQ-9 guys are at homestation, in an emergency you can easily draw on additional personnel to meet requirements.

5. Since an A-10 or F-16 uses significantly more fuel, and cost of fuel can be significantly more expensive in theater (and to be fair sometimes it can be cheaper), a generic comparison of fuel costs may not be accurate. For instance, if the cost of fuel is twice the average in the forward location, while the percentage difference of fuel costs per hour flown will be the same, the absolute difference between the two will be twice as great.

6. One minor point in favor of the F-16 is that it doesn't require very expensive satcom links to operate, but since the sensors are all processed outside the theater, you probably couldn't reduce the absolute number of satellites.

[Kazys Shukis]

Jim .... Nice article! What attracted me to it was claim that Wheeler's work was BS. It's refreahing to me to find someone who isn't afraid to stand up to people like Wheeler.

[James Hasik]

Thanks for both Curt and Kasys for the endorsements. As Curt has some recommendations, I should acknowledge the advice in detail:

1. Agree. I really should think more about the datalink costs, but you may be quite right about that.

2. Agree. I missed the difference TDY costs of airmailing the whole of the mini-squadron's strength around the world. I'll adjust next time.

3. Agree. Training a drone pilot almost certainly costs a lot less than training an F-16 pilot. The USAF may have the numbers, but if I can't find them (next time), I will think about how to estimate these.

4. Agree. I didn't miss this; I did actually think about it, and alluded to it in my point about crew rest. All the same, I didn't address it quantitatively, and I probably should have tried to do so. One would need rather a few more pilots to somehow keep the same number of F-16s or A-10s airborne.

5. Agree. I totally missed that. I just took the USAF's fuel costs as fuel costs, and we've all read a lot about how fuel costs crazy amounts in Afghanistan.

6. Mostly agree. I admit to having left out the datalink costs, and frankly, satellite bandwidth should be relatively easy to price. There's something of a market for it. I do take your point about the fighter possibly needing to export its imagery, but we could price out a line-of-sight link with a big team of analysts at Bagram. When I have more time, or someone wants a thorough rethinking of this, I will definitely do that.