1. YOU were LUCY. You could boot up to the 68000 in your machine instead of the 68030. A3000 owners had no such fortune.
2. Now, you're underscoring exactly what my point was to begin with: bad code is bad code, no matter what the platform.
3. No op-codes "changed" with each new processor (although many did take fewer cycles to complete). Each 680x0 processor had complete user-mode backwards compatibility with prior chips. They also had additional user-mode opcodes, which didn't break prior software (as they were unused). Heck, on the 68020+, they even allowed word/long word accesses on odd-addressed lines, which would generate an address exception error on a 68000. The changes which broke software occurred in supervisor mode, with new stack frames, new supervisor registers, etc. All of those things were stuff handled by the OS. When a game took the machine over it had to explicitly follow Motorola's recommendations for dealing with interrupts. But yet, few did, and many programs broke. The only reason user-mode program would break on an upgraded processor would be either (1) self-modifying code unaware of the cache or (2) software timing loops.
4. If you had enough money to buy a machine & a development kit & get your game published, you definitely had enough money to get minimal (Certified) developer support from Commodore. No if's and's or but's about that one. Minimal developer support included all documentation needed to write system friendly programs. In fact, the only document published by Commodore which doesn't focus on the OS is the Hardware Reference Manual, which DOES detail exactly HOW you should be banging the hardware, and games which followed their directions to the letter also worked on AGA systems without patches.
5. Sorry to hear about the CD development stuff... yeah I know, I priced out the equipment at the time myself and came to the same conclusion. (For the curious, single-speed CD-R drives & crude burning software costed around $15000 at the time. Blank media was somewhat less expensive, at around $50 a pop)
6. When codetapper fixes the games, does he not add a *lot* of software from the WHDLoad suite? Isn't that bloat? Don't you require a SHITLOAD more memory to run the WHDLoad versions than the original?
7. Lets just say that there's an awful lot more going on in Quake than there ever was in any Amiga game ever written. Transcendental math functions, transformation of matrixes, etc. You'd be more accurate to compare a modern 3D game to some of the Amiga rendering programs. And believe me, they need that much horsepower & RAM.
I'm having a little bit of difficulty in understanding your post, sorry. I do, however, agree that QA is unfortunately an aftersight in modern PC games. Corporations force the programmers to push the product out the door in time for next months quarterly results. Let's hope that the XBox can change all that for us. MS is going to use a console model for release of games.
Let me further state that yes I have been stung by this personally. I bought Unreal when it came out without anyone in any review telling us that it ran like UTTER SHIT on any non-3dfx card. I had to wait TWO years for the final patch which made it playable on my system. Needless to say I did NOT buy Unreal Tournament & will never buy an Epic game ever again.
No, unfortunately most of the games DID make it past the cracktros. Some didn't, yes, but many did, and subsequently failed after loading, sometimes spectacularly. I had about 2 dozen originals, too. None of them worked on the 3000 except Black Crypt.
I'm going to close my reply to you by sayin, games now are more complex than they've ever been, and therefore, harder than ever to "get right". It doesn't take a rocket scientist to figure out that a 200,000 line program has more potential for errors than a 10,000 line program.