Theory of Information
We must always bear
in mind that our bridge bidding is a highly restricted language (just
35 + 3 elements and a restrictive grammar too) and therefore perfection
is simply not possible!
At the same time then; the greater should be our concern to get the most out of its limited possibilities! Would you not agree?
When one considers that the possible number of deals is:
53,644,737,765,488,792,839,237,440,000. On the other hand
the total number of possible bidding sequences is also very large, but a large percentage are unusable.
There is a mathematical theory of the information applicable to any
system with a continuous field or (as is the case with our bidding
language) a system of discrete values ...the legal bids available in
any context.... Do not doubt the power of this theory, whether
applied to discrete systems like in bridge, human languages, or the
electron
states of a Hydrogen atom or in non-discrete physical systems such as
fluiddynamics, general relativity, electrodynamics etc.
This theory is known as 'Fisher Information'. If one stops to realise
that this theory can be used to derive many of the most important
mathematical representations of the physical world, such as,
fluiddynamics (Stokes), General relativity (Einstein) and even Quantum
Mechanics etc - it is indeed a mathematical and a little known unification
of physics!
It can most definetely be applied to bridge bidding - a discrete system - too!
What it tells us is that the maximum information that can be passed at
any bid depends on the proper use of all available bids and crucially
the more economical a bid is to carry a broader scope of information. And put more quantitatively the frequency of usage must
decline, quite rapidly according to theory, with with each possible
available bid and a recognition that the information in each possible
bid will decrease, that is 'become more specific', as more bidding space
is consumed. This will not come as a surprise, I'm sure, but what maybe
a big surprise is that there is a more rapid decay in the amount of
information with each successive bid should convey than you may expect.
I know of no bidding system that comes even close to this. This is not
surprising as it would inevitably cause intense artificiality and given
that most of our bidding systems have naturalness at their heart the
outcome is clear: a battle between naturalness (easier to acquire) and coded (higher precision).
So, the theory would expect that economical bids like PASS, double, redouble or
the next available n(1-7) of ( NT should be very general and the more of precious bidding space
that is consumed the less information, that is more specific, will be inferred.
This is a clear endorsement of 'doubler' bids, bids that make require
further bids to unravel their meaning further - including further 'doubler' bids!
It would also suggest that in opening bids such as 'PASS' should have the
highest frequency (it most certainly does! c40%)
- it also suggests the concept of a forcing opening bid of PASS is
sound! Frequencies for 2 and higher level openings are all
lower in frequency.
Let's take a closer look at ACOL:
1 should be the most
frequent bid - this is far from true in ACOL.
Indeed frequency seems to increase from 1 through to a lordly 1NT! whose frequency has increased with the decades!
Consider precision's 1 opening which comes in at just under 9%. But while precision showed strength & 1 has a slightly greater frequencey than Major openings.
In Beta 1 1> > > > NT The frequency of opening steadily decreases.
Beta denies presence of a 5 card in and 's
frequency is higher than Major openings as it covers a higher CPU range
while Major openings are limited by Beta's strongish 2Ma
openings.
Betas 1 openings are all about the 7.5 to 8% with
D's>H's>S's> 1NT at only 6% So we observe a tailing off of
probabilities as required AND crucially more in line with Fisher than
ACOL!
In Beta 2C's 6%, 2D's (3-way) 5% 2Ma each 2% 2N(21-22) which may contain a 5 card Major: 0.5%
Well in 'beta we do address this failure in ACOL: increasing the
frequency of 1 at the expense of 4 card openings in 1& 1 and particularly
a far more constrained 1NT. This is a powerful endorsement of beta ACOL over ACOL.
In conclusion if the principles of information are ever taken to the
limit in a bidding
system even beta ACOL would be shown up but we would have totally
artificial system - harder to master and likely more prone to disaster
even fro EXPERTS!
Well I'm not
about to go about constructing such a futuristic system right now,
although admittedly I have given it some thought! Personally, I still believe in a balance between 'naturalness' &
'artificiality', the former more human and easier on the memory while
the latter requiring considerable mental effort and dare I say
prone to somewhat more prone to disaster when misunderstandings occur!
One option i do commend wholeheartedly is that the next bid is
conventional - which is frequently the case in Beta incidentally. Beta has artificial sequences like many systems.
Finally how do the most common systems of today measure up to this theory? Conclusion none is much better than any other! But Beta has made a considerable effort!
What of beta ACOL as compared to ACOL, well I claim a definite improvement! Consider these:
The above is not an exhaustive list! and
Its also arguable that the EBU's straightjackets have and still will
resist innovation in an over-protective attitude towards ACOL.
I strongly advocate that beginers in bridge learn a 5 card major
bidding system
SAYC or even PRECISION which will be a much quicker, easier and more
effective entry to
competitive bridge. ACOL on the other hand demands far more experience
to succeed putting beginners in a very weak position at the table!
Worldwide bridge as seen on BRIDGEBASE for example are largely 5
card major systems and even strong NT. A much more comfortable envirnment
for beginners than the local clubs in the UK pushing and protecting
ACOL. The lockdown may well have brought this to the attention of many
a bridge player in the UK in the last 18 months!