Mr. Church is absolutely right, and I have a telling story from the decline and fall of Lucent to back it up:
The last project I worked on was something essential to Lucent's survival as an independent entity. Many of you have heard of sophisticated telephone switches, like the Western Electric (AT&T) then Lucent 5ESS (http://en.wikipedia.org/wiki/5ESS), a really impressive monster that's truly 5 9's reliable (or it genuinely tried to be and by and large made the grade).
The new way of doing this is (or was as of 2001) to split the functions of such a switch into a Media Gateway Controller AKA Call Agent e.g. the Lucent Softswitch, which understands things like billing, and a Media Controller which does the actual work of connecting A to B and converting as needed (e.g. POTS to VoIP). There's even a dual ITU/IETF protocol for communications between the two, Megaco AKA H.248 AKA RFC 2805.
It should be obvious that being able to offer a complete solution was essential for Lucent's future, as conventional switch sales were at least projected to plummet, and my group was part of effort to create the other half the company needed, the Media Controller. Which was a challenging thing, including that true 5 9s reliability requirement (just say no to hard disks, I was among other things the database guy in testing, and the operational database lead and I had to vet and sign off on each other's designs (mine was a data warehouse approach for storing fine grained events generated by the test system, Oracle was real fun for that sort of thing and we had a 106,000 seat license (!) from them, although headcount was going down to 35,000), and I was personally responsible for testing his system).
ADDED: going off the Yourdon Death March chart included in OP comments, this was a "mission impossible" project, high happiness (fun, challenging project), at least a decent chance of success initially (e.g. we were working with the 5ESS people, who taught us a lot of neat stuff about real 5 9s.)
Anyway, at a certain point, completely insanely a company wide headcount cull was applied to this effort, and I'm pretty sure it was sufficient to kill the project (my testing area was gutted, going from "maybe, just maybe we can get this work done in time" to one engineer plus a little outside help, with one manager, a failed programmer so she wasn't of much if any help).
One of the things that really stood out was the layoff of a pretty famous person in the company, a jack of all trades who e.g. had the energy and skill to take the chosen (Linux based) RTOS from the vendor and get it up and running in our environments, kick the tires and minimally validate it, etc. Don't know if he was a PITA like I was in part, but it's hard to imagine he wasn't at least a bit, he was the sort of guy who was in a position to say "This won't work because I just tried it".
The managers flaunted his layoff like a badge of honor, an implicit "If this guy can get fired no one is safe", and blatantly lied about the grounds. They claimed he was "inflexible" about what he'd work on, when it was a very visible matter of written record that he'd very productively work on just about anything. I never even talked to this guy, I just learned of his background and merits from the written progress reports and mention of him in an occasional conference call meeting, he was perhaps the most visible "exceptional" employee in this make or break project.
And yet he was purged.
(As was I, although that was by abuse from an envious boss and may not have even been part of the headcount cull; while I wasn't hardly so visible, I was the only serious software engineer on the testing effort, and there was no way the company could build the new test system without someone of my caliber. And you'll never create a true 5 9s system without serious testing.)
And as many of you know, the bottom line is that Lucent was bought by Alcetel and is a shade of its former self, including Bell Labs, e.g. http://en.wikipedia.org/wiki/Bell_Labs#2000s:
"As of July 2008, however, only four scientists remained in physics basic research, according to a report by the scientific journal Nature.
On August 28, 2008, Alcatel-Lucent announced it was pulling out of basic science, material physics, and semiconductor research, and it will instead focus on more immediately marketable areas, including networking, high-speed electronics, wireless networks, nanotechnology and software."