I'll betray my age, I'm afraid, in order to contradict point 2) above. I was a research/academic type back in 1985 who picked up a Lisa with the proceeds of a ziplock bag software royalty payment, and got what turned out to be a unique perspective on early object-oriented software evolution. The Lisa guys got to see what the Smalltalkers had done at PARC, and with the help of Eurolanguageguru Niklaus Wirth, set out to reproduce this world in Object Pascal on the 68000. The result was an environment that looked so much like Office 2000 does today that you'd be floored. The enviroment was an inspired but slapdash heap of first-draft object-code that embodies key notions from Smalltalk 80 V1 like BitBlt, and windows, and a host of other goodies, with orthodox coding notions from the Pascal world. It further let me see how these ideas, once set, could be shorn of inheritance, and much of the polymorhism, and cast as a simple Pascal API in the Mac Toolbox. Ideas appear first as methods. As they are reused, cut and paste variants can appear, either in new classes, or subclasses. As commonalities are discerned, these are promoted to superclasses. As what changes is really understood, the variablility can migrate to components and parameters. Once objects really mature, they become black box components. This, and the subsequent chance to see both Smalltalk V1 and V2, were what inspired our early observations about object evolution (such as those in
http://www.laputan.org/drc/drc.html and http://www.laputan.org/dfc/discussion.html). It was fascinating to see what was improved, and what was kept, as subsequent drafts of these systems appeared...

There is no other education in OO techniques like looking over the shoulders of the masters like that. Indeed, it is possible to track the patterns and idioms in Smalltalk-80 through the Lisa ToolKit into the Mac Toolbox, then into MacApp, and into other OO frameworks like Interviews, ET++, and on into Windows and MFC. It's a great tale, and we've sworn we'll get around to documenting it one of these days...

--BF, still basically a recovering Lisa victim...

When I saw my first Lisa I was done in for it. The really hard question is, when the 68000 helped build these machines (Lisa, Mac, Apollo DN100, Sun-1 and many more) and most everyone I know liked the Peddle-Melear model more than the Faggin Shima model, where I think adressing and memory layout where the biggest pain and Peddle had showed before that you could build his architecture (6502, 25$) a lot cheaper than even the Z80 (70$) why did everything reverse. Fast ascension to mass production is the difference I see. Super cheapo overall packages. Intel and MS wanted a lot less and talked more than Motorola and their Partners. The clone industry was supercreative in terms of products and price. But:
"Had IBM chosen the 68000 they would have killed the workstation market before its inception." They could even have made Xenix into a huge success. But Intel and MS (who then propably) sold more to the 6502 world) felt weaker than Motorola and DRI (a SW for Intel company) so they made the compromise deals with "the Evil Empire).
Collective engineering mindsets might come in too, pretty complicated.
So my next revelation came with Smalltalk/V on the PC which I saw before the MAC version. Programming the MAC in Pascal (which I loved as a language) was very difficult if you came from C64 Pascal.
Back to my money argument:
a Lisa was totally unaffordable for someone like me. I could afford a Mac from the SE30 on when I bought a Classic from a friend who upgraded. On the other hand an aquaintance from then who worked in tuwien.ac.at replaced an Apollo machine with a MACII with Smalltalk/V made a large bargain.
We should not forget that Sun, Apollo saw $20.000 and above minis as their competition and wanted to stay in high price markets, so did the commercial Unix SW crowd (Sun, Oracle etc.)
Where did Adele Goldberg want to go?

All of those systems made large contributions to what we take for granted now Only they did not make their makers rich. Intel, MS (implementation company, no idea company), IBM and Compaq made contributions to implementation and massproduction technologies
The invisible hand works in mysterious ways.

It is interesting that you mention that the original reason for Intel's and Microsoft's success may have been that they originally felt weaker than Motorola and DRI. I guess this is a recurring business pattern - companies rise to prominent positions because they start small in a difficult or even hostile environment.

For instance, take all the software that is released as open source because people know they don't have the marketing budgets to push it against the big companies' products. Now of course, at least so far, open source software hasn't made people rich, at least not comparably rich to the way boxed software did. Maybe "being successful" in the sense of making a big splash is becoming less and less connected to making big money.

guess you make a very good remark on the matter here. I'd like to add that you could also look at this with the background of a consumer capitalism vs. services exchange markets debate. It might be that the extraordinary profits made by "successful packaged SW" just stem from the mixture with a still long living but basically outdated form of economy. .net looks like a further development of tying services into a uniformely reproducable consumer product world.
Still it will take some time for many people to dream of gold shitting donkeys and buying into spiced up haystacks and force fodder.