FM75

Now adding an automated learning component is a very fine idea!

According to everything that BBO personnel have posted, the rating system is based upon data. They don't have people making judgments on the "starriness" of a pair. That makes it completely "objective" to my way of thinking. The algorithm takes a number of measurements and calculates a number between 0 and 5. They could do it subjectively, by crowd-sourcing, I suppose. They could show a friend list, country, and whatever objective measures they have and ask someone to rate between 0 and 5. That would be subjective.

At this point, I think it would be fair to delete my initial post. It seems like the discussion is just about politics, capitalism, and socialism. I don't see any comments about how bureaucracy can destroy an otherwise fine idea and program. Likewise, apparently, nobody disagrees that the program is broken - perhaps because the highly competent bureaucrats and Congress have done such a fine job designing and funding it. The Title/subtitle as a reminder: Government Intervention in Medicine What could go wrong?

It is funny that you did not rethink the assertion that "any continuous distribution can be corrected to a normal distribution ...", after your powder example. First of all, by definition a Gaussian (normal) distribution is defined as a function over the domain of all real numbers. In any event, the universe is full of non-Gaussian distributions - The energy density as a function of frequency of a black body and the energy levels of electrons in an atom spring immediately to mind. In fact, any measurement that is bounded for any reason, for example temperature, can't fit a Gaussian distribution. The age of all of the members of any particular species - yes, non-Gaussian. That said, I can't imagine a reason for imposing a gaussian distribution model on a player skill ranking at anything - except that perhaps the person making the measurement plans for the results to be Gaussian and is willing to skew the measurements in some way to "make them fit" his postulated distribution function.

It has been explained several times, in this thread. No, BBO did not publish their "algorithm" which they have described as "under development". That said, what is the point in describing what medicine you should take based upon your individual genetic makeup, and the likelihood of whether it will work and for how long? You would not understand the science behind it, even if you were given the volumes of details about it, or even the abstract describing the technique. Would that make it of "apparently no practical use"? Well perhaps to a sceptic like you, it would not be of any practical use. You could criticize it as much as you like. But a doctor treating you, with knowledge of the methodology - absent the details - and the understanding of how well it worked, would be crazy not to take it into consideration. He might even be found culpable of malpractice if he ignored it. Note: I am not suggesting that BBO's algorithm has the same level of predictive ability, but analyzing the predictive value of their model should/could be their next step.

Background My mother is very ill. I took off work today, to leave to visit her in a senior rehab center, a 90 minute drive for me. The medical and family issues involved are not relevant. But her condition, and the information available to her family caused my brother, a board certified endocrinological surgeon, to cancel his appointments for the rest of the week, and begin a 5 hour drive yesterday to visit her as well. His concerns were that the care and the information we had gotten from our brother and sister-in-law (RN and newly licensed LPN) did not really make sense - details irrelevant to this post. Government Intervention and Background I have no medical training other than basic first aid and some self instruction on wilderness medicine. My brother has not only practiced medicine, but he taught medicine as a professor in a university, and also has had a management role in surgery at a VA hospital. He is reasonably computer technologically savvy - for an MD - maybe even for anyone not in computer software industry. Does this scare you? Given his VA background, it was natural that at some point today, we had time to discuss something other than the medical issues facing our mother. He no longer works in the VA, but I asked him about what his take was with the current controversy (for lack of a better description). He allowed that it was common practice - eventually - that the policy of a time limit between request for treatment and actual treatment, led naturally to a distortion of reality. Patients with a complaint, for which the backlog exceeded policy, would have "delayed entry" into the system. This mechanism, via "political natural selection", became widely known, and in his opinion was likely "universally practiced". He also mentioned that there were delays, in some cases, due to patients not being able to meet on the scheduled (perhaps delayed) dates. So far, this may seem like a management problem? To be fair, this practice became well known to the top level management of the VA. It gets worse The VA had a computerized system built in the early 90's to track patient histories. It was DOS based (most bridge players used computers in that era). It eventually was DOS running under Windows. This version did not really get any "smarter". But it gets still worse (IMO and his). The VA decided that they needed extensive data. They mandated that all of the doctors in the system get "training" for analyzing possible mental issues - psychological and social. They also required counseling on smoking cessation, drugs, and even what firearms the veterans had in their homes (I am not making this up!). This training was not specialization specific. For example, the same "course requirements" existed for surgeons, family practitioners, etc. So there was not a 30 minute course for surgeons on smoking cessation, and a longer course for GPs, etc. Darwin and government got the expected result. Soon the "test answers" became well known. It ain't just the VA Once the lawyer/regulator gods got the religion, it started leaking into Medicare. HIPPA - originally all the data was aggregated - in compliance, of course, with regulations prohibiting personally identifiable information. The next step is auditing, by the government,. the information provided. Can you audit without personal information? Now the whole idea is moving into medicare. Does a surgeon who operates on organs in the endocrine system have the time to counsel on smoking cessation - to check that box to get reimbursed? Is the ownership and possession of firearms, by individuals presumably highly trained on how to use them by our government, germane to surgery on their thyroids? Opting out Unsurprisingly, you can expect doctors to "opt out" of medicare. At the VA, a doctor could only treat 12 patients in an 8 hour day! Even if it were not financial, would you want to limit the use of your expensive knowledge and expertise to treating only 12 patients per day (40 minutes per patient) because of paperwork?

As mentioned, Dealer has some pretty nice features built in. These make for an easier project that is still interesting. You could have plenty of fun "integrating" your application with an existing system. A good software developer is "constructively lazy". He/she does not reinvent the wheel, but "stands on the shoulders of giants". Apologies for a truly "mixed metaphor" there.

The simplest answer to this question is "yes". I have done this. There are many ways to automate it. I would recommend using python, because it is easy, and as fast as anything else. The response time of BBO will be the "bottleneck". If you know Java, or a shell scripting language and curl, perl, etc., start with what you know instead. Once you "script" it to download the files (one hand at a time), then you will need to "massage" the files into a format that meets your needs and "bridge toolset". If I were not employed, I would probably build such a system - using a web-based interface - and charge a subscription for using it. There are open-source tools that you could integrate into a full blown product. Eventually, I will complete this project for my private use.

It is "scientific" IMO, to consider what is the worst 15 partner could have. If he had that, what are the odds that opener will find the opening lead? 16 is slightly above the average 15-17 opening hand. With 16, what is probable, etc.? Could a partnership agreement leave you short of a cold slam? Hard to assess this, but it certainly depends on the depth and quality of the agreement. Finally, of course, what is the form of scoring? How strong are the opponents? If duplicate, how strong is the field?

I alluded to that in my request. Thanks for directions, but I was hoping that you could change your posts in such a way as to not get as many other results.

Ben, I like your problems. Is it possible that you could either edit the titles in such a way that a BBO search finds just these. It can only search for words of length >3, can search by author, and by which topic holds the post. At first glance, this looked like draw trump and ruff the last club, but 10% of the time trump might split 4-0, which would require 3-3 split to pick up the last club - about 36% until you discover the 4-0 trumps at which point it gets less likely. Can't throw a losing club under another winner. What else could I do? Ruff diamond high, return to dummy, ruff diamond high. Ah yes, a dummy reversal. With a surplus of top trump, I don't even need to worry about how diamonds break. I guess that puts me somewhere in the intermediate spectrum.

Statistics of hand distributions meeting your constraints. For example, how likely were the hands to occur? What was the average, std deviation of each individual suit, points held, both individually and as partnership.

Hand rotation needed?

1♣ - 1♦(neg) - 1♥ - opener has 4 card ♥ (and perhaps ♠) (in our Cambridge ♥ variation, the rebid is either natural or a subsequent NT bid will show a balanced hand and hcp range - and does not promise a 4 card major - likely looking for NT contract or a part score which might be a major) 1♣ - 1♥ - positive and 5+♥ 1♣ - 1♠ (positive, balanced no 5c M) - 1NT (right-siding NT and asking for controls) 1♣ - 1N (positive and 5+ ♠ - various asking bids available including control ask or suit quality ask (agreeing spades as trump) 1♣ - 2m (positive, if has 4 card M, the hand is (24)(52), or has 6 card minor, or (13) or worse in majors. Not Moss-Bathurst, nor Meckwell, but missing 44M is pretty infrequent,

Mike Lawrence, in one of his books, spent several pages on the weaknesses of the "short club" opening. He pointed out many defects - typical of partnership agreement holes in the bid. He did not mention the problem faced by the partnership in the play when they end up defending. Perhaps if he revised his book, this would be added. http://www.bridgebase.com/forums/public/style_emoticons/default/rolleyes.gif (Or agree to never defend after a 1♣ opening. http://www.bridgebase.com/forums/public/style_emoticons/default/wink.gif ) Play it if you want. But it would make good sense to find his discussion and make sure that your partnership has satisfactory answers - or don't play it. After shoring up your agreements on the bidding, clearly you need to discuss your defenses, which certainly should allow for your partner to make decisions based on the auction and his holding if he is the opening leader, as well as decisions based on what he sees in dummy after the opening lead.

I felt a bit embarrassed to even ask this question. Now I am glad I did. We use it very broadly. In constructive auction after 3rd suit if at the two level, in competitive auctions, regardless of who opened, after X of a weak two, and in one mildly unusual auction (precision) 1♥ (1♠) 2N = lebensohl, with the only additional restriction that it is not lebensohl if it is some other artificial bid in our system such as unusual no trump or a Romex ask to make short suit game try, or if it is an asking bid or a response to one. While there might be some odd cases where partner could elect not to bid 3♣, we do consider it obligatory. 2N = relay to 3♣ seems like a minimum. "Lebensohl" means many things to many people and nothing to a few. I felt that if asked, we might also say lebensohl, but was worried that somebody might later object "that was not lebensohl" which could be true from their understanding. Clearly explanations of any follow-on alertable bids, after 3♣, seem required if asked, but I wanted to make sure that at any point we had our "ducks in a row" with respect to any required disclosure.

If a partnership has agreements on various uses of lebensohl, then clearly the 2NT and 3♣ bids are artificial and need to be alerted. (In particular, I am interested in ACBL, but other answers are invited.) Assuming that 2NT is really an artificial temporizing bid, "requiring" partner to respond 3♣ is "relay to 3 clubs" sufficient as an explanation? If lebensohl is used in many different competitive situations, or even possibly only in a constructive situation, does anything more have to be provided at the time it is bid? If there are alternatives that might have been available - e.g. a 2 level bid, should they be described, or is that something that should be explained only if asked? Likewise, is the explanation for 3♣, simply "bidding 3♣ as directed'?

I would rate quantum electrodynamics (includes relativistic effects) as significantly harder than quantum mechanics. Each is impossibly hard without an understanding of calculus and partial differential equations, and the first is truly difficult with limited knowledge of particle physics. If you punt on rating problems, analogously, we might ask first graders to solve Ph. D. level problems in any subject. Rating is (I hope, now) obviously important. Does it need to be one dimensional? Likely not. Perhaps Ben's problems need some sort of rating system that includes pre-requisites. Even then, there may be an occasional problem that can be solved with either of two prerequisites - and resulting in two different solutions, to be truly interesting. I have seen this in a published software program by our revered host. http://www.bridgebase.com/forums/public/style_emoticons/default/biggrin.gif Can we give Ben his own section for his problems?

As a project, you could write a program to "manage" scripting the dealer program on BBO. It is capable of reporting various forms of statistics. Two weeks won't be much time for a project like this unless you are pretty darn good at the language you are using. That said, just because one exists should not stop you from writing your own. You could look at the links mentioned to see what BBO/Dealer support. You need to do several things - one generate random deals, two establish a mechanism for specifying constraints, and three establish whether a deal meets the constraints, and finally do something with what you got - report statistics or whatever. You might find that any one or two of those components might take a week or more. Deciding how long software development takes experience. The best advice - predicting is hard, especially about the future - and software engineers routinely underestimate how long it takes to develop something. Sounds like fun. Report back if you do something like this. :)

Engineers [mathematicians] aren't boring people. They just get excited discussing boring topics. http://www.bridgebase.com/forums/public/style_emoticons/default/biggrin.gif

Also Ctrl-A in windows selects the entire chat. Windows users should know this, but maybe many do not. Mac users just use Command key instead of Ctrl Key for the same functionality mentioned.

Speaking of ratings systems for compatibility... I have noticed women's dance shoe advertisements in the left side of the screen lately. Surely you know that I am male. (I recently bought some birthday gifts online. Of course, I know why they show up there.) http://www.bridgebase.com/forums/public/style_emoticons/default/biggrin.gif Sure, those advertising matches are not perfect for you, since I don't click on them. But they really don't bother me! I am willing to bet that the day everybody thinks the compatibility rating system is good will be the day after everyone voluntarily stops using the windows client. It is probably also a safe bet that you will always have more complaints about it, than compliments. I like the idea, and I have not seen any ratings that I found unusual. If I hit one, I won't lose sleep over it either. I will click on a woman's dance shoe advertisement. http://www.bridgebase.com/forums/public/style_emoticons/default/cool.gif

Looks like a great idea. I can imagine that Singular Value Decomposition might work quite well. After all you have some objective measures in the mix as well as known "friends" and enemies. You could also likely use "facebook" style networks with the friends component. I hope it works, and expect that eventually you will have it working quite nicely.

:) Which is why I suggested that an effective study would need to be done against a range of competitive bidding systems. Of course, your results likely would vary against each, which would show you when you get your best results, and when you get your worst - or conversely, how best to compete from the set of systems measured, against the preemptive style.

Or simply "share" the hands, or even make them "public" which are the mechanisms most expected on the internet today. Dropbox, Evernote, etc. being examples of this technique.