I have reasons to believe and to predict that
average PR's of humans against bots will be
comparatively lower in games lost and higher
in games won.

I dared all to share their average winning and
losing PR's but unfortunately (and expectedly)
nobody did so (at least not yet).

I suppose it must be true that when children
close their eyes and don't look, the monsters
under their beds are gone... ;)

So now, I'm taking the subject one step further
and predicting that average cube PR's of humans
against bots will be higher than checker PR's,
perhaps both in games lost and games won, but
especially so in games lost than in games won,
because the so-called "cube skill theory" is a
big cow-pie and people can get away with higher
cube errors than checker errors.

Once again, I dare all of you children to be
brave and look under your beds... ;)

MK

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/16/2024 12:06 PM, MK wrote:

especially so in games lost than in games won,

I meant the opposite. Sorry.

While here, I may as well go another step further
and bring in the cube errors in relation to cube
values but I will keep it at this for now because
I don't have a clear enough idea on this yet. It
seems that cube errors will be bigger at higher
cube values, which at least in my case seem to be
more often in games that I win than I lose. Feel
free to volunteer your observations and opinions.

MK

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/16/2024 2:06 PM, MK wrote:

I have reasons to believe and to predict that
average PR's of humans against bots will be
comparatively lower in games lost and higher
in games won.

Related observations have been made by John O'Hagan and
Douglas Zare.

https://www.bgonline.org/forums/webbbs_config.pl?read=163763

https://www.gammonvillage.com/backgammon/magazine/article_display.cfm?resourceid=6355

In particular, they suggest that when you're losing, you often
have easier decisions to make and so that will tend to lower
your error rate. They suggest that this effect occurs not only
when humans play against bots but when humans play against
humans.

---
Tim Chow

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/18/2024 6:12 AM, Timothy Chow wrote:

Related observations have been made by
John O'Hagan and Douglas Zare.

Only three paragraphs of Zare's article is
free to read but even with that little, it
is interesting that he is the only one who
mentions luck along with skill in the same
sentence.

One of my arguments on this subject is that
"luck+skill=1" is a cow-pie. Obviously Zare
won't say this. So, I wonder what more did
he say about luck in the rest of his article.

To repeat my argument: winner has always more
luck, thus more luck = less skill, thus less
skill = more errors, higher PR...

In particular, they suggest that when
you're losing, you often have easier
decisions to make and so that will tend
to lower your error rate.

I skimmed through the posts in the bgonline
thread, looking for keywords and skipping
most of the dwelling on the details of their
common arguments as you summarized above.

Unlike me, all you guys are both worshiping
believers who will never admit that cubeful
equities are inaccurate. They need to find
explanations that won't shake their faiths.

Bob Coca questions "Is there data indicating
this?" But how can any data indicate their
explanations based on decisions being easy
or difficult?

You comment that XG won't count sufficiently
easy decisions as decisions at all. Bot how
does XG know if a decision is sufficiently
easy or not? For that matter, how can humans
know?

I think you had participated in a few past
discussions about determining the difficulty
of positions, to find out that it's not easy
if not nearly impossible to do.

They suggest that this effect occurs not
only when humans play against bots but
when humans play against humans.

In humans against humans, players are much
likely to deviate from the bot play, and PR
is measured by the bot. In humans against
bots, at least the bot is totally consistent
and human is less likely to deviate from the
strategy to achieve the best PR.

Anyway, what would Occam's Razor say here..?

MK

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)