Final, definitive Post about Nelson....

spacious_mind · Post by **spacious_mind** » Sun May 31, 2015 12:20 pm

I give up, facts, logic and appeasement obviously does not work. Maybe its a language thing.....

I am moving on, enjoy yourselves folks.

mclane · Post by **mclane** » Sun May 31, 2015 1:19 pm

i also concentrate on facts.

fact is we have no other programmer.

horvath has completely different style, behaviour and evaluation.
danielsen too. the programmers can mainly be identified by the different playing styles the programs have. over the years one can identify
schroeder programs, spracklen programs or others by their own playing style and evaluations etc.

nelson comes close from playing style, only the old nelsons are not strong enough. this can be explained by the usage of the different hardware (more mips) and features like big book, permanent brain etc.

this is fact, isn't it ?

the gap is from 1394 sensory voice to 1845 of the igor.
451 ELO difference in 17 years.

if we consider 140 elo difference due to speed and 50 due to permanent brain, 50 due to bigger book with transpositions, we have 240 elo.

now could a programmer make 211 ELO in 17 years by adding something to the search and evaluation function ? IMO yes.
especially if you consider it was 1980. you can easily see that the 1980 nelson program is very rough and unfinished. its a child of its date.
and in 1980 many search features were not used and invented.
nullmove, pre processing, iterative search etc this was all not common.
it all came later in these 17 years.

If you have a different opinion you could come with a name of another programmer (e.g. horvath, danielsen or xyz), you should IMO show parallels in games or positions between programmers different machines.

mclane · Post by **mclane** » Sun May 31, 2015 1:45 pm

the relation nelson spracklen is even worse. I wrote

The difference in elo between Nelson and spracklen maschines in The early days was not that big. 1394 with Nelson on z80a with 4 MHz (NO permanent brain). 1550 spracklen with 6502 2 MHz (permanent brain).

There is 1 year of development between the program versions. Nelson was 1980 and spracklen was 1981.

If we conclude that z80a with 4 MHz equals 6502 with 2 MHz we can estimate the difference between them was maybe 100 elo, if we give 50 elo for permanent brain feature.
Also spracklens book was 3500 plus and Nelson only 1000.

but i was mistaken. z80a 4 mhz equals 6502 1 mhz, not 2 mhz.
that means spracklen program had doubling in hardware speed !

so 1550-70 for the doubling of the hardware,
then it is only a difference of 86 between the champion from 1981 and nelson from 1980.

now subtract 50 ELO because of permanent brain and you have 36 elo.
but the spracklen book is 3.5 times bigger then nelsons.
so this could easily make 36 elo.

so where is the big difference in capability between spracklens and nelson ?!

the difference is vapourized.

its marketing.

of course spracklen made progress too.
and nelson ? he had his program. he put the listing in his desk.
and never ever used it again ???

thats very unbelievable.

spacious_mind · Post by **spacious_mind** » Sun May 31, 2015 2:06 pm

mclane wrote:i also concentrate on facts.

fact is we have no other programmer.

horvath has completely different style, behaviour and evaluation.
danielsen too. the programmers can mainly be identified by the different playing styles the programs have. over the years one can identify
schroeder programs, spracklen programs or others by their own playing style and evaluations etc.

nelson comes close from playing style, only the old nelsons are not strong enough. this can be explained by the usage of the different hardware (more mips) and features like big book, permanent brain etc.

this is fact, isn't it ?

the gap is from 1394 sensory voice to 1845 of the igor.
451 ELO difference in 17 years.

if we consider 140 elo difference due to speed and 50 due to permanent brain, 50 due to bigger book with transpositions, we have 240 elo.

now could a programmer make 211 ELO in 17 years by adding something to the search and evaluation function ? IMO yes.
especially if you consider it was 1980. you can easily see that the 1980 nelson program is very rough and unfinished. its a child of its date.
and in 1980 many search features were not used and invented.
nullmove, pre processing, iterative search etc this was all not common.
it all came later in these 17 years.

If you have a different opinion you could come with a name of another programmer (e.g. horvath, danielsen or xyz), you should IMO show parallels in games or positions between programmers different machines.

Thorsten,

Ok let us wipe our minds clean of everything. What is easier.

My assumption

Take Horvath + 1 addition (Spracklen Attack Tables) = Mirage = 1 modified Horvath.

Krypton owned Horvath program. Excalibur (Samole/Nelson) had what they could legitimately give to Krypton "Spracklen Attack Tables".

I am not the one who added Spracklen to the conversation. By all accounts Nelson did himself so I am assuming no theft is involved.

Your assumption:

A) From Horvath (Levels)
B) From Horvath (Evaluation)
C) From Spracklen (Attack Tables)
D) From Krypton (Teach Mode, these exist in almost every single Krypton computer)
E) User Manual (Krypton's)
E) Everything else Nelson? (what is everything else?)

Now explain especially since you indicate stealing in your previous post. What everything else means? What else would Nelson need to have taken from someone else to get his program up to speed?

1) Ponder? (Already exists with Horvath)
2) Opening book? (from who? From Horvath? From Spracklen?)
3) Engine? (who? his? Spracklens?)

You are adding to your own stealing theory in the direction that you are going.

I am not accusing anyone of stealing. Hence my statement of "I give up".

I have told you repeatedly that I have taken Nelson well beyond your 4x speed difference and it doesn't matter, the speed difference does not make the program any stronger because it has too many dumb moves, horizon effect dumbness, zero endgame knowledge. Who would teach Nelson endgame knowledge or is that another "oh by the way I also included Spracklens Endgame tables"?

At some point you have to take the easy path, manufacturer's don't have time for people fiddling about, computers have to be built quickly and sold to give you salaries that you can take home to your family and make them happy. This job was not about fiddling around.

Best regards

mclane · Post by **mclane** » Sun May 31, 2015 2:40 pm

where do you see that i talk about stealing ?

i am saying that nelsons last machine came into the shops 1980 and that we talk about igor/grandmaster etc. from 1997.
thats 17 years of computerchess history and development. and in these 17 years computerchess algorithms made big progress.

i do not believe that horvath has anything to do with the igor/GM/ivan
program.

the horvath programs can be identified easily by the HELLO in the display and the passive playing style.

the arrangment of the levels was in the DUTY concept of the producer of the dedicated chess computer. when the programmer was hired he got this duty book and subscribed that he would program the machine following these duties.

thats why some dedicated chess computers all made in the same factory or all made under the same producer have similarities in the levels.

spacious_mind · Post by **spacious_mind** » Sun May 31, 2015 3:25 pm

mclane"]where do you see that i talk about stealing ?

There is no need for stealing code from spracklen or using horvaths engine or whatever conspiracy you want to construct.

Maybe I misread you, but I don't think so. The fact that you are stating this is to get this to be plus in your favor where in fact the opposite would be true.

i am saying that nelsons last machine came into the shops 1980 and that we talk about igor/grandmaster etc. from 1997.
thats 17 years of computerchess history and development. and in these 17 years computerchess algorithms made big progress.

We are talking about Mirage, 1994. We are talking about identical hardware and playing strength. Legend is equally as strong as Mirage, what coincidence is that?

i do not believe that horvath has anything to do with the igor/GM/ivan program.

Don't forget to add Mirage. I am ok with your different opinion.

the horvath programs can be identified easily by the HELLO in the display and the passive playing style.

I agree with you on passive style, that has troubled me as well for a very long time leaving me scratching my head of "why a computer that has everything from other Horvath computers, plays so differently but still ends up every time being the same strength as Horvath?"

But no more. I have to thank Cameron with his "Spracklen Attack Table" disclosure. It filled in that missing link. "SAT" does not add to strength, instead it prioritizes "Aggressive" moves to be played first. It overrides what "Horvath" would normally play. Hence good aggressive moves and bad aggressive moves equals at the end of the day "Zero strength improvements/miniscule strength improvements". They cancel each other out, but "Passiveness is overridden". I can see this I can visualize this through my computer tests that show the same thing.

I have taken programs of every manufacturer that has settings for this and they all change how they play but at the end of the day the "Normal" setting and the "Aggressive" setting end up back to being very close to same playing strength. But you can follow many changes in moves that were chosen throughout each of these test. I have posted a lot of these tests here in the Forum.

"Hello" is no argument. Mirage/Igor/Ivan/GM say "Chess". That is just changing one 5 letter word to another. To me that is another indicator for "Horvath" confirmation and that it comes from Krypton, because the "idea" comes from Krypton computers.

the arrangement of the levels was in the DUTY concept of the producer of the dedicated chess computer. when the programmer was hired he got this duty book and subscribed that he would program the machine following these duties.

But it is from Krypton right?

Best regards

mclane · Post by **mclane** » Sun May 31, 2015 4:21 pm

i do not have a mirage. I can't say much about it though. maybe it is same program as ivan/igor/gm. i don't know.

horvath program is very different from igor/ivan/gm.

it plays similar to chess friend/pandix from horvath .

when i said there is no reason to steal i referred to theories of others in this thread. i do not believe there is a need to steal anything.

no matter if 1994 or 1997. 14 years are much for computerchess development.

spacious_mind · Post by **spacious_mind** » Sun May 31, 2015 4:32 pm

mclane wrote:i do not have a mirage. I can't say much about it though. maybe it is same program as ivan/igor/gm. i don't know.

horvath program is very different from igor/ivan/gm.

it plays similar to chess friend/pandix from horvath .

when i said there is no reason to steal i referred to theories of others in this thread. i do not believe there is a need to steal anything.

no matter if 1994 or 1997. 14 years are much for computerchess development.

Thorsten,
Yes we agree!!
I agree to everything you just posted. Horvath under normal circumstances has a different play behavior!

But in a non normal circumstance where someone took out the Horvath styles and replaced them with a prioritized Spracklen "SAT" style, you would now get this as a priority for moves that are played. As a result of this you also get the occasional really stupid moves like 8. Qf3?? which under normal circumstances a good program which Horvath is, it would not play. After all you only need to go 2 deep or to see it is a bad move.

8. Qf3?? is a Spracklen move, Spracklen's play it and Igor/Ivan play it.

Best regards

klute · Post by **klute** » Sun May 31, 2015 5:03 pm

mclane wrote: There is no need for stealing code from spracklen or using horvaths engine or whatever conspiracy you want to construct.

Hi mclane

I'm not sure if that comment is directed at me, but if it is, I've made at least two or three posts in the last few days trying to make it crystal clear that if the Excalibur GM program employs Spracklen Attack Tables, there would be absolutely nothing professionally, commercially or legally untoward whatsoever about that.

I think even some otherwise very knowledgeable people here simply do not understand how chess computers and programs are developed.

Regards

Cameron

klute · Post by **klute** » Sun May 31, 2015 5:16 pm

spacious_mind wrote: Quality by the way is a failing that moved around with Nelson. He could never really ever get that under control at Fidelity or at Excalibur. He lacked something in this area.

Hi Nick

I broadly agree with your poor quality comment. Even if Excalibur was still a going concern, I doubt I would be prepared to risk stocking Excalibur products.

I feel that with Excalibur in particular, Nelson was frequently pressured to push products out the door before they were ready.

Regards

Cameron

klute · Post by **klute** » Sun May 31, 2015 5:48 pm

spacious_mind wrote:Maybe its a language thing.....

Hi Nick (and everyone of all races, creeds, colours and languages)

In mainly English language forums like this with a significant population of non-native English speakers, any misunderstandings or misinterpretations quite probably are "a language thing" much of the time.

I sincerely admire the multilingual abilities of the non-native English speakers here. I have what I suppose you could call a limited working proficiency in French and I can translate German slowly, but that's about it. If I was writing in a native French environment I could probably get the thrust of what I wanted to say across, but there would be language nuances I'd miss that only a native French speaker could understand.

Regards

Cameron

klute · Post by **klute** » Sun May 31, 2015 5:54 pm

spacious_mind wrote:Qf3?? is a Spracklen move, Spracklen's play it and Igor/Ivan play it.

Hi Nick

One of my favourite (if that's the right word) Spracklen identifiers is the tendency to play knights to A3, H3, A6 and H6 in the opening.

Regards

Cameron

spacious_mind · Post by **spacious_mind** » Sun May 31, 2015 6:21 pm

klute wrote:
spacious_mind wrote:Maybe its a language thing.....
Hi Nick (and everyone of all races, creeds, colours and languages)

In mainly English language forums like this with a significant population of non-native English speakers, any misunderstandings or misinterpretations quite probably are "a language thing" much of the time.

I sincerely admire the multilingual abilities of the non-native English speakers here. I have what I suppose you could call a limited working proficiency in French and I can translate German slowly, but that's about it. If I was writing in a native French environment I could probably get the thrust of what I wanted to say across, but there would be language nuances I'd miss that only a native French speaker could understand.

Regards

Cameron

Hi Cameron,

Thorsten knows I speak fluent German albeit my grammar and spelling is getting worse and worse as time passes (but that applies to my English grammar and spelling too

). Thorsten has complete control of English there is no miscommunication

I was being facetious because Thorsten likes to only read every second word

Despite our arguments I also have the highest regards for Thorsten.

Apologies, to everyone, nothing untoward intended to fellow people from earth

Best regards

klute · Post by **klute** » Sun May 31, 2015 6:22 pm

spacious_mind wrote: I have told you repeatedly that I have taken Nelson well beyond your 4x speed difference and it doesn't matter, the speed difference does not make the program any stronger because it has too many dumb moves, horizon effect dumbness, zero endgame knowledge.

Hi Nick

I agree. I did similar tests 30 years ago to the ones you've recently done. In my case, I played a match between an Advanced Voice on the 11 min / move level, and an Excellence on one of the lowest levels (I forget which one, but no more than 30 sec / move) with pondering disabled.

The Advanced Voice didn't come out on top!

I do have a soft spot though for the top-of-the-line Nelson machines just prior to the Spracklen arrival. They're still majestic chess computers.

Regards

Cameron

spacious_mind · Post by **spacious_mind** » Sun May 31, 2015 6:23 pm

klute wrote:
spacious_mind wrote:Qf3?? is a Spracklen move, Spracklen's play it and Igor/Ivan play it.
Hi Nick

One of my favourite (if that's the right word) Spracklen identifiers is the tendency to play knights to A3, H3, A6 and H6 in the opening.

Regards

Cameron

Argh! my tests start past the opening book, but will look for similarities beyond book.