Fernando wrote:f) So it could be -and I am almost sure of it- that Excalibur AND Nelson produced intentionally weak or at least not too much strong engines to cope with that new market, so different to the old one, where amateurs were always looking fro strength.
g) Why, then GM? Because there is also a slice of the market constituted by middle level players that like to play a relatively strong machine, but NOT THAT strong.
I agree that marketing has to influence the types and strength of machines but I had already carefully considered all of that when coming to my own conclusion that Nelson contributed little to none of the actual GM engine.
The actual hardware is virtually the same amongst all of the machines. Over time there was further miniaturisation, however the actual capability of the hardware varied little. And you could buy expensive premium Excaliburs with this hardware (DTTC) as well as extremely inexpensive $30 devices (various plastic chess devices with the proper tournament levels) - the only difference between top end and bottom end models being clock speed (half in the cheap plastic models). So really, the marketing comes down more to program flexibility rather than actual hardware. And it is not that difficult to make a strong program play weakly whilst still maintaining it's maximum playing strength. At a really simple level you program interrupts to minimise thinking times or you restrict ply searches and extensions. Again I would simply say why would any sane company chuck away such an excellent program as GM, since the easiest way to have a universally appealing product (as opposed to a mass market one only or a niche one) is to simply pack the program in something small and relatively portable (to reflect modern trends) and add some novice levels to it. That way it caters from raw beginners through to club players.
And in any case, if you really want to create a truly decent and entertaining novice level to casual player machine, it is a far better approach to use a fairly mature and relatively sophisticated program and handicap it with thinking time restrictions and very basic hardware. Such an approach produces a far more useful, entertaining, educational and desirable opponent for a novice or casual hobby player than taking an inherently weak and relatively neanderthal program and obtaining the ELO strength by way of brute force on the hardware side. Because by doing that, you still have a program that plays unconvincing chess, even if it is in outright terms significantly stronger than a more sophisticated program deliberately handicapped to play at the casual / novice level that it is aimed at.
Yet time and time again, we see the same Excalibur program with it's extremely dubious positional and strategic play - only to suddenly have this GM that did everything so very well, only to chuck it all away again and go back to 1978 on steroids all over again. It makes no logical sense at all. Not financially, not from a marketing perspective and certainly not from the perspective of maintaining a company image.
The fact of the matter is that the hardware itself was never a financial consideration in the era we are discussing (late 1990s to late 2000s) since as I say, you'd find the same innards in a $300 machine and a $50 machine, the only significant differences being the clock speed, quality of the housing, voice / lack of voice, etc. So it only came down to software. And it is actually easier to make a strong existing program play weak chess than to keep developing a mediocre one version after version (as Nelson did - you can hit certain key combinations on most of his machines to see which version that machine happened to have).
It will never make sense to me that Excalibur instead decided to throw everything learned from GM away especially when they continued - for some years afterwards, to attempt to market strong machines. Long after GM came onto the market, Excalibur were still marketing and advertising machines purported to play at the 2000 ELO level. So they were still apparently serious about strength or otherwise completely deluded. So the more likely explanation is that whatever legal and copyright agreements were in place when GM was created had expired and were only in effect for the GM product. Thereafter, Excalibur had to go back to Nelson's "base" program.
Perhaps if in the post-GM days Excalibur only marketed their programs as mere toys rather than as alleged strong expert machines, I would agree, but all the evidence points to them still attempting to pull out all the stops so to speak (even to the point of having a product named after a World Champion) , though obviously limited technically in their capability to do so. Afterall, I have little doubt that Nelson at his absolute best could produce no more than 1600 genuine ELO from a machine on H8 hardware with 32K ROM.
Chess is like painting the Mona Lisa whilst walking through a minefield.