EM versus SS - The smoke-filled debate.
Flash, when it was first produced in 1979, used the Williams System 4 MPU board and eventually graduated to the System 6 MPU towards the end of production. These boards, from the System 3 to the System 6 formed the early part of Williams migration away from Electro-Mechanical (EM) machines to Solid State (SS) machines. The first solid state machine was a game called Flicker from Bally in 1974. For Williams, the transition began in 1976 with a game called Aztec, for which a whopping ten machines were produced. Aztec was also produced as an EM (and over 10,000 machines produced). For the next two years, Williams produced both EM and SS machines. However, the company switched completely as and from the first of January in 1978. With the benefit of hind-sight, it's easy to wonder about why this happened over a year, and why they didn't just stop making EMs on Friday, and start with SS's on Monday morning, seeing as the technology is just so much better.
I'm picturing a smoky conference room in Chicago (probably because I've just started watching Mad Men, long after everyone else). The young kids (probably retired, by now) are presenting their plan for a new and improved pinball machine. Yes, they're using acetates on an overhead projector.
They are pointing out that the benefits are significant. The ability to use timers for certain ball activities (like ball-save), and other fancy software routines. Complex scoring rules, of course. I mean, look at where we are, today - some games require a PhD in maths to know how to play the game. Probably the biggest change though, is in the cabinet. Gone are the rows of relays. Gone is the dreaded score motor. Gone too are the score drums, match counter, and credits drum. In that context, it seems like a no-brainer.
But then, the old guys get their chance to speak. That's all well and good, they say, but relays are simple. They don't "crash". Route operators know how to fix them; just carry a spare set of leaf switches, some sandpaper, and a bending tool. You can hear the CPU kids yelling in unison - "the score motor!?!?" Indeed, the score motor is quite the beast. It's like a 1960's Artificial Intelligence. Just when you think you've figured out how it can manage to score thirty points by pulsing the step-up coil for the "ten" drum three times, it defies logic and counts to forty. Or it doesn't count at all. Or it spins forever, making that distinctive EM pinball sound: da-da-da-da-duh! Somewhere, a switch contact is open and it should be closed. Or it's closed and it should be open. Not only that, there are hundreds of them. How is that easier to fix than software? A few of the older guys scoff. Loudly.
At this point, the financial people chime in. OK, OK, cool your jets, they argue. What are the cost benefits, here? With a flourish, the software guys whip a fancy graph onto the overhead projector. The numbers aren't immediately convincing, when considering the price of CPU chips and memory. Memory costs were $32,000 per megabyte in 1977. Today it's about a quarter of a cent per megabyte ($0.0027). An 8k byte static RAM chip had a list price of $250 and a 6800 processor chip cost a cool $175 before you added ancillary chips like the PIAs. It wouldn't be unreasonable to think the board might cost close to $1,000. Per pinball! That would buy a lot of relays.
The presentation might have included the enormous cost of wiring all those relay switches to the various solenoids - the cost of the wiring looms and the manual cost of hand-soldering every connection. Not to mention the rework costs if a single contact is soldered to the wrong switch. In those terms, the computer looks quite feasible.
At this point, the EM guys are on their feet. "What about the software costs?! You've included the costs of soldering relays, and yeah, that's pretty high, but who's going to write that software?"
Indeed. If only they knew. Today, the software industry is worth over $400 billion. Software engineers in the bay area (California) regularly earn six figure salaries. I'm willing to bet that the labour costs involved in producing wiring harnesses and soldering said harnesses to the various components probably hasn't gone up by that much in the last forty years.
No matter. I doubt few today would argue in favour of a return to relays. Certainly no manufacturer has considered it, and a few have made "retro" style machines, complete with score wheels. It also helps that processor speeds have doubled as per Moore's Law and memory costs have plummeted. Today, your average Raspberry Pi costs around $50 which is significantly less than the old System3
through System6 boards. Computationally speaking, however, they are about 1.4 million times faster. Plenty of opportunity to add complex game rules, to say nothing of callouts, graphics, video clips and music.
There are afficionados of old electro-mechanical machines, and I count myself amongst their numbers. Nothing can compare to the modern pinball in terms of game rules, bonus scoring, and of course multi-ball, but that score motor sound is intoxicating. It's not quite the same thing, listening to call-outs on Stern's Deadpool in comparison to the cacophony of bell rings, score drum step-up coils, relays chattering, and the immersive experience of an EM game.
Overall, I'm glad the EM guys lost the debate, and I'm glad every machine produced since the late seventies has used an embedded computer to operate the solenoids and keep score, but that doesn't mean I don't enjoy having a battle of wits with a score motor from time to time.