The WMWA-FM main studio in Glenview had a very primitive automation system at the time. A couple of open-reel tape decks and a mechanical carousel cart changer were driven by a controller programmed by some 10-position slide switches, and control tones recorded onto the tapes. It consisted of "Play the device indicated on this switch until it plays an end tone, at which point start playing the device indicated by the next switch. Lather, rinse, and repeat." Someone still had to manually turn on the transmitter and adjust its power. When our remote studios at North and South were built, a phone remote was added to the transmitter. Its primitive voice-and-touchtone interface read you the current state, and you dialed numbers telling it to turn on and off, or raise and lower the output power. We were supposed to adjust the power once an hour--and it really did need it, it would drift about five watts one way or the other from its target of 140.
At the time, the intarwebs was still the domain of large corporations and academia. I had a dialup account on a Unix box that dialed into DePaul a couple times a day to pass traffic via UUCP. You couldn't just wire the stuff into a PC and jack it into the wall. Ethernet cards were over $200, and consumer broadband was 7 years away. You could've done analog modems over dry pairs, but that was Expensive™. The voice menu solution really was the only practical approach. (Though, I'd argue you could've written a power-control program to run on an Apple II.)
Fast forward to the year I graduated college and started at Qualcomm. I built my first Linux box, and there were plenty of methods for doing analog I/O, and multiple channels of stereo sound. Affordable systems were fast enough to run the new MP3 encoder in a reasonable amount of time (2-4x real playback time,) and the decoder would run full 44.1 kHz 16-bit stereo playback without killing the machine. That was the year Keith would start DJing parties at HoF with a PC and MP3s instead of CDs. It was then Blindingly Obvious that you could run a radio station this way.
I have no monopoly on good ideas. Plenty of companies have stepped into that field. Predictably, all of them did it with Windows software. Audio recording and playback was clunky and worthless for a few years, 'till Microsoft got that shit straight. The software was (and continues to be) expensive as all fuck, and uses annoying licensing schemes involving dongles and resource limitations. If I'd still been involved in radio, I would've immediately started work on a free solution under Linux. (We were still years before OS X in those days, and the cooperative multitasking of System 7/8/9 at the time wasn't something I'd trust to such a real-time-critical application.) Digital equipment was making its way into the high school by then... Pinelli had bought a couple of "digital cart machines" that used Bernoulli 44/88 MB disks for storage. You could conceivably (and expensively) replace that old mechanical sequence system at WMWA with a PC driving a few of these Bernoulli-cart decks. You'd still have to manually change them, or build a carousel, but it was feasible.
Sure enough, you can't keep a good idea under wraps forever. I was flipping through Linux Journal's web site today looking for something else, and found an article about the Salem Radio Labs. Salem Communications operates about a hundred commercial radio stations. Their chief engineer in Seattle had the same idea I did, and has developed a full-blown commercial-scale and quality operations and automation system running on Linux. All the code is GPLed. Some of the systems have Windows clients, to accomodate users for whom that works better. I am beyond impressed. The interface is very nice, and designed along the same lines as existing commercial broadcast equipment--it's familiar to people who've spent their lives in that business. The audio editor supports marking fade and segue points in media files, to allow for properly cross-fading music. It patches into commercial telephone equipment for managing talk shows, complete with caller screening support. It talks to RDS encoders to put text ID and music track information out for radios that support that. It deals with archiving and logging. It operates audio switchgear--Salem's Seattle operation runs five stations out of seven studios, all of which are totally interchangable.
I'm sure the guys doing commercial Windows software hate this.