1 out of 1 members found this post helpful.
Did you find this post helpful?
Yes |
No
Originally Posted by
ericfarmer
This may also be privileged information, in which case disregard the question, but I'm curious what is Don's involvement in your work with Tarzan? Before the previous thread was closed, Don commented that I was alone in wanting (or at least being willing to) post similar simulation results in this forum... followed a day later by Tarzan posting additional results of your simulation. Is Don similarly "steering" your work, do you and/or Tarzan submit results to Don for review prior to posting here, etc.? Or is this just you and Tarzan working together?
Tarzan responded about Don's interest and involvement in an earlier post. From my point of view Don's role has been as an interested advisor. I believe that he has an interest in the Tarzan system with respect to where it would land within the spectrum of the SCOREs of counting systems from simple, to more complex, to perfect play. He has been instrumental in vetting early results, spotting things that didn't look quite right and advising us as the project has proceeded.
I had approached Tarzan about simulating his system independently. As I see it, Don and I have both been working for Tarzan. I see the results of the project as belonging to him -- to keep or to publish as he sees fit. Having said that, I was a bit surprised that he posted results as quickly as he did, given that simulations are still underway on different games and conditions. I had thought that the results would all be published at once as part of a more comprehensive article. I believe that this is what Don was hoping for as well, both with respect to the Tarzan work and your perfect play project.
Originally Posted by
ericfarmer
On a more technical note, this is very interesting stuff. Are you able to describe how your simulation implements Tarzan's strategy? *Not* the details of the strategy itself, but is there an "interface" for specifying a strategy of which Tarzan's is just one example? (Or is Tarzan's strategy more "hard-wired" into your sim code?) I ask because this capability could be useful for evaluating other similarly more complex multi-parameter counts.
I can say that within my software, there is an API for implementing counting systems. There is general code for the raw counting of cards. There is general code for performing common mechanics like deck estimations and true count conversions. There is semi-object-oriented representation of a counting system (the code is written in C, not C++) which contains methods which must be implemented for each system. There are default methods and "objects" which are sufficient to implement traditional systems for which only the tags vary. The blackjack engine simply calls the methods for the counting systems that have been registered with it.
For more complex systems, like T count, the implementation is custom written C code for each of the required methods. All the heavy lifting is done there. I have implemented several custom systems using this framework and have not yet come across something that I could not implement.
Bookmarks