Is there a drone bubble? It's not clear whether this is more like the .com bubble, when a lot of useful stuff was built but a couple of years too early, or more like the housing bubble, when a lot of stuff was built in the wrong places to the wrong standards at the wrong prices and will probably never be worth much. It's the nature of a bubble, of course, that it's precisely at the top of the bubble that the commitment to it is greatest.
One of the things the RQ-170 incident tells us about is some of the operational limitations of the drones. Typically, they are piloted in the cruise from locations that may be a long way off, using satellite communication links, but when they land, they do so under local control via line-of-sight radio link from their base. This allows us to set some bounds on how much of a problem link latency really is, which will take us circling back to John Robb's South Korean gamers.
Gamers are famous for being obsessed with ping-times - the measurement of round-trip latency on the Internet - because it's really, really annoying to see the other guy on your screen, go to zap'em, and get zapped yourself because it took longer for your zap to cross the Internet than theirs. Typically you can expect 40 or so milliseconds nationally, 60-80 inter-continentally...or several hundred if a satellite or an old-school cellular operator with a hierarchical network architecture is involved. A sat hop is always clearly identifiable in traceroute output because latency goes to several hundred ms, and there's a great RIPE NCC paper on using the variations in latency over a year to identify the satellite's geosynchronous (rather than geostationary) orbit as the slant-range changes.
On the other hand, roundtrip latency across an airfield circuit a couple of miles wide will be negligible. So we can conclude that tolerable latency for manoeuvring, as opposed to cruising, is very little. Now, check out this post on David Cenciotti's blog from January 2010. Some of the Israeli air force's F-15s have received a new communications radio suite specifically for controlling UAVs.
You might now be able to guess why even drone pilots are going through basic flight training. Also, this post of Cenciotti's describes the causes of six recent hull losses, all of which are classic airmanship accidents - the sort of thing pilot training is designed to teach you to avoid.
That said, why did all those drones get built? The original, 1980s UAV concepts were usually about the fact that there was no pilot and therefore the craft could be treated as expendable, usually in order to gain intelligence on the (presumably) Soviet enemy's air defences by acting as a ferret aircraft, forcing them to switch on the radars so the drone could identify them. But that's not what they've been doing all these years.
The main reason for using them has been that they are lightweight and have long endurance. This is obviously important from an intelligence gathering perspective, whether you're thinking of over-watching road convoys or of assassinating suspected terrorists (and there are strong arguments against that, as Joshua Foust points out). In fact, long endurance and good sensors are so important that there are even so-called manned drones - diesel-engined, piloted light aircraft stuffed with sensors, with the special feature that they fly with intelligence specialists aboard and provide a much faster turn-around of information for the army.
Their limitations - restricted manoeuvre, limited speed and payload, and high dependence on communications infrastructure - haven't really been important because they have been operating in places and against enemies who don't have an air force or ground-based air defences and don't have an electronic warfare capability either. Where the enemy have had man-portable SAMs available, as sometimes in Iraq, they have chosen to save them for transport aircraft and the chance of killing Americans, which makes sense if anti-aircraft weapons are scarce (and surely, the fact of their scarcity has to be one of the major unreported news stories of the decade).
But then, the war in Iraq is meant to be over even if the drones are still landing in Kurdistan, and the US may be on its way to a "pre-1990" military posture in the Gulf. This week's strategic fashion is "Air-Sea Battle" and the Pacific, and nobody expects anything but the most hostile possible environment in the air and in the electromagnetic spectrum. And the RQ-170 incident is surely a straw in the wind. Also, the Bush wars were fought in an environment of huge airfields in the desert, and the ASB planners expect that the capacity of US bases in Japan and Guam and the decks of aircraft carriers will be their key logistical constraint. (The Russians aren't betting everything on them either.)
I think, therefore, it's fair to suggest that a lot of big drones are going to end up in the AMARC stockpile. After the Americans' last major counter-insurgency, of course, that's what happened. The low-tech ones are likely to keep proliferating, though, whether as part of the Royal Engineers' route clearance system or annoying the hell out of Japanese whalers or even playing with lego.