I love the F-35. Is it a great plane? Who can tell. But as a piece of software, as a marketable product and as a project, it’s complex, over budget, and disastrously late. Rehabbing late software projects seems to be where I eat these days, so I’m hooked.

The radar has to be rebooted mid-flight. Integrations with other military hardware are partial at best. But the laundry list of bugs is less critical to my eye than the fact that nobody knows what this product is for.

Defense acquisition is still in a postwar mode defined by MAD and summarized by JFK’s inaugural: ‘only when our arms are sufficient beyond doubt can we be certain they will never be employed’. ‘Beauty too rich for use’ makes for a pretty strange product category.

As developers of (sometimes) novel products, we experience some of the same headwinds as military contractors: trying to sell what’s next when the old model is still cutting edge, walking the line between features you can pre-sell vs real innovation, the difficulty of unseating established players, and various slow start phenomena (nothing is perfectly useful on day one).

Missions and minivans

I drove the family station wagon in high school which was a pink Volvo V-70 and somehow the other kids thought it was cool. It had a stickshift & a turbo but also was useful for odd jobs; carting around surf & sail gear, taking home some lost dogs we found on the road.

The F-35 is ‘multi-role’ in that it was designed to do multiple things well but it’s not a station wagon – it costs more, isn’t a reliable daily driver and doesn’t hold a lot of cargo. The existing aircraft it competes with are better and relatively cheaper at the job.

Categories that the F-35 is trying and failing to own:

Close air support: The reigning champ here is the A-10 warthog, which has a gun the size of a volkswagen and has the largest bullets of anything that flies, good armor for low-altitude survivability, and a relatively long loiter time.

High & fast jets like the F-35 are bad in this mission because they have limited loiter time over the battlefield. And high-altitude = less ability to distinguish friendlies on the ground.

VTOL: There’s a vertical takeoff and landing version of the F-35 because most international customers have ‘helicarriers’ rather than aircraft carriers with a full runway.

But VTOL brings vast complexity to the airframe. It’s an extra variant with unique maintenance challenges and many extra moving parts. It also melts whatever it lands on, so you can’t park it just anywhere.

This is a wishlist feature that shouldn’t have survived a sober cost-benefit calculation.

Dog-fighting: The dogma is that visual range air-to-air combat has been over since korea. If that isn’t the case then the F-35 is underprovisioned with a barely-functional gun, a weapons bay ‘the size of a purse’ (yes google it), and questionable manueverability.

Stealth: Ancient long-wave Soviet radars like the SA-3 are being used effectively for area denial in Syria. This is the same radar that shot down an F-117 over Belgrade. My read is that radar stealth technology is another ‘cold war calculus’ trick that is less relevant in small wars.

My point: nobody did a sane assessment of how the product would be used and what’s on the market now. Sane shopping is hard.

Enterprise sales & kitchen sink

John Boyd’s ‘fighter mafia’ in the 60s rejected the prevailing dogma that heavy & powerful fighter-bombers were the way to go. His philosophy resulted in the F-16 and F-18, which are not perfect planes but have been widely adopted because they’re cheap and maintainable. These ‘scrappy’ products, less competitive if you trust the customer’s wishlist, turned into a massive and sustainable business.

They’ve been used successfully in a wide range of conditions and both are still being manufactured (many model versions later). Some F-35 export customers are considering buying modern F-18s instead because they known how to use them and the F-35 is so darn late.

But defense acquistion is a lot like enterprise sales; sometimes the way to make a sale is to collect all requirements, put a price tag on it knowing you’ll have to extend the contract, and then overpromise & under-deliver.

If I have a point here it’s that enterprise customers need to be saner about what they’re buying; in my experience the way to do this is to include the user of the product in the decision, rather than only empowering senior managers to ‘buy their priorities’ (vs what is practical).

Air forces with lower budgets are starting to spend practically. The Phillipines bought KAI T-50s from South Korea – this is a ‘jet trainer’ rather than an actual fighter (low range, low armament, low speed etc) but it allows a small country to maintain an air force on a budget.

If you’re on a real budget you can have a Combat Caravan for the price of a Cessna Caravan, a hellfire AGM and 50 rolls of duct tape. See also the cheap, jig-less South African AHRLAC.

Asymmetric threats & TCO

A reaper drone costs $16 million before you put weapons in it. ISIS in Iraq fielded modified commercial drones and DIY contraptions made from plywood and styrofoam. Iraqi troops found an ISIS drone lab in Ramadi where a strapped R&D department was trying to bolt on a gyro system so they could program navigation.

If small asymmetric wars are the future, the F-35’s main enemy won’t be advanced radar-guided anti-aircraft missiles or the latest Russian or Chinese jet. It will be a cloud of cheap self-propelled flying machines made from scraps. The US has the largest defense budget but not for the sort of investments that are working – cyber & information, asymmetric stuff. If you look at high-ROI projects like drones made out of garbage, we’re probably last.

A 747 is objectively better than a goose but Tom Hanks still had to make a movie about the one time it wasn’t.

The Russians unveiled their ‘avangard’ hypersonic glider this week which they claim can maneuver at mach 20. I haven’t read anything about the measured speed of their test or the materials science here, but this feels like a win for Russia. They accepted their limited budget and made a wise compromise between operations and R&D.

We can be bankrupted, or just tired out. When Ford took over production of B-24 bombers in World War II, they turned out 1 per hour in Ypsilanti Michigan and delivered more than the army could use. As a plane, the F-35 is objectively better than the B-24 but as an investment it may be objectively worse.

Product designers & buyers can’t afford fishy accounting around ROI and TCO (though there are legitimate uncertainties in those numbers). We have to be as excited about operability, maintainability and affordability as we are about new capabilities.

Slow-start, readiness & operability

Every product has to deal with slow start problems, i.e. it may not deliver its value prop on day one. Social networks are a special case of this (nobody wants to be the only person on a social network) with personal value precedes network value as a proposed solution.

Like social networks, aerospace hardware is typically no good until it has been run through its paces. This isn’t just about fixing bugs in the product. Jet fighters have a social network-like system of maintainers and operators who know how to keep it flying, detect warning signs, and operate in degraded mode.

These networks are sustained by practice and, like an ecosystem, can be damaged by sudden shocks. Sweden experienced this when Russia ran their Good Friday 2013 air raid on Gotland. The Swedes had recently downsized their air force and couldn’t scramble Gripens in time to intercept.

New products are full of surprises. In the F-35’s case, deploying the machine gun pulls the plane off course (my Volvo had torque steer too). These things happen but also, this is what happens when you don’t fire the gun until 10 years into the project. The A-10’s gun produces as much force as one of its engines and is placed precisely for balance.

New products can also surprise us in how they interact with old products. An F-18 shootdown of an ancient Su-22 over syria last summer brought up this account of testing AIM-9X air-to-air missiles against ancient soviet flares. The AIM-9X, designed to evade countermeasures, had only been tested on friendly American flares. The decades-old Soviet flares fooled it every time.

The cold war defense acquisition mentality says that letting the budget dictate the capabilities is the same as admitting defeat. My experience tells me the opposite: piling on features while ignoring constraints (budget constraints, testing time, personnel, expertise) is a way to get features with unusable flaws. And furthermore creates a culture of ignoring reality that leads to worse failures down the line.