Forum for Direct Comparison
Why link to Roberts page on Quality of DC when none of it is useful? [#2227]
Why indeed? Of the 3 advantages listed on that page none of them are correct. The page finishes by saying that Robert would have won a tournament once if DC was used as the tiebreaker. What merit does this external page have? It is a rehashed version of a discussion page in which it was pointed out to Robert how wrong his ideas on Direct Comparison where. The difference is that on the external page none of the refutations are written. I fail to see why Senseis Library should link to information which is not only misleading but deliberately so.
I reinstated the link on the basis that in most cases censorship is a bad thing. Robert's Quality of DC reasoning maybe faulty thinking, or not. I've not followed this DC argument, so I haven't an opinion on who is correct. I'm just saying I think allowing linking to a page where someone attempts to outline their argument seems reasonable to me. Other's may see this differently.
If censorship is a concern, perhaps an alternative to deleting the link would be to add information (i.e. somebody that has a differing opinion about the quality of direct comparison could write about it and post their article as well).
The problem with that is that is has already been done on various talk pages on SL. Robert's ideas are not censored, they were widely debated, but they were not supported by anyone apart from Robert (at least in the discussion pages I saw). I really don't see the merit of linking to that page.
A member of the EGF rules commission wrote a page about a tie breaker, that somehow reflects in the advice on tie breakers given in the EGF Tournament System Rules (preference for Direct Comparison). Obviously that is highly relevant, regardless of your opinion about its utility. And I really don't get why you are so obsessed about this particular link.
The proposition that Robert saw no support in any of his claims regarding tie breakers is provably wrong. I remember to have supported him just recently in a discussion about the limited utility of SOS as tie breaker and I am not at all interested in this kind of stuff.
Kind regards Tapir
Agreeing with something Robert wrote about SOS hardly supports his ideas on the different subject of the quality of DC. That argument is as convincing as the page I wish to delink. Yes Robert wrote this page, yes Robert wrote that section of the EGF rules. It contains his personal preferences. Robert dislikes ratings. Robert prefers DC over SOS because he would once have won a tournament if DC would have been used. So what? As a point of principle, on a technical subject page linking to false arguments cannot have merit.
Oh, if you need another wording: Yes, I agree with Robert that Direct Comparison can be usable tie breaker. It often breaks ties, sometimes arbitrary, and has the further advantage of being applicable in Round robin tournaments where SOS and similar tie breakers are not. And it doesn't pretend to measure anything, which is a big advantage as well.
If you keep on picking on Robert on this page as you did before, we need to lock the page.
Kind regards, Tapir.
Shall we discuss ![[ext]](images/extlink.gif)
http://senseis.xmp.net/?DirectComparison%2FDiscussion ten more times to gather a more statistically reliable viewpoint? Or shall we simply say that sticking to your original viewpoint carries more importance.
Yes, you can discuss this as often as you want with Bass and Robert. Picking on Direct Comparison on the basis of its arbitrariness while preferring a tie breaker which is commonly referred to as lottery (SOS-lottery), talking of a statistical misunderstanding being the base of Direct Comparison while talking about meaningful measuring of "playing strength during this tournament" without even trying to establish detection limits for this measurement to just give two examples - is not necessarily more convincing than the arguments of Robert.
The point is tie breakers are not for measuring anything, but for breaking ties. As long as this central misunderstanding is at large of course everyone will claim that his preferred tie breaker is "measuring something", even if it is not at all.
Cheers Tapir
Yes and MMS does not measure anything, we don't know its limits. Lets just stop holding tournaments. We have all been idiots.
If you organize amateur tournaments not out of love for the game but mainly to measure the strength of players (when playing a lot of games and setting up rating or asking for professional evaluation does the same or better) there is of course a problem.
I read some of the discussion that happened earlier regarding the Direct Comparison method, and I think that this topic is very interesting. In my opinion, as people point out on the page, it's certainly possible for the direct comparison method to be prone to error for some of the reasons outlined on the page.
However, as is also suggested, the SOS method is prone to error as well. Personally, if I were to select a tie-breaking method on my own, I might opt for something that takes a player's rating into consideration. But even considering such an approach, it is not directly clear to me whether a tied player of lower rating, higher rating, or even one with a better ML performance rating "deserves to win".
When it comes down to it, none of the tie breaking methods seem to escape from error to me, and it seems fair to select SOS, DC, or even a random selection of the winner provided that the tie breaking methodology is clearly established and known by the players prior to the tournament. Even if the selection of a winner in the event of a tie is completely random, if the players know about it in advance, it seems OK to me.
This is because somebody that "loses out" in such a situation can't really complain since their tied opponent(s) performed just as well by the normal tournament ordering. The important thing to me is that the tie-breaking strategy is announced to the players before the tournament begins.
MMS contains errors, SOS contains a sum of those errors, SOSOS,SODOS...etc. Consistency is an important beast. If we are to consider statistical accuracy for one tiebreaker we cannot then ignore it for another, for example DC. I can quite understand why a player might prefer Direct Comparison, that does not mean I need to pretend to understand any false reasoning for that. :)
Provided I understand your point, I agree with you. I think that all of the systems have errors, and some may prefer DC, SOS, MMS, etc.
What I want to say is just that I think that it is important for the players to know beforehand which tie-breaking system is being used. If the players know beforehand about the tie-breaking system, then there is no problem IMHO. It's on the same line with knowing the ruleset that's being used, the time controls, and so on.
If the players all understand how things are going to work and there are no surprises, I don't see any problems with any of the systems.
This is important yes. I know of one European Tournament were the tiebreakers used were not those advertised, for those losing out on an international place this was a bitter pill to swallow.
I wrote down my opinion about the said paper in January, but didn't publish it beyond the EGF rules commission.
Because satire seems to get censored here, I seem to have no choice but to publish the non-satirical text.
As the page is locked, I cannot add the link there directly, so here it is:
http://iki.fi/bass/go/qoqodc.html
-Bass
Concerning your comments on my webpage:
On my page, I state: "Only a Player's Own Performance Affects His Direct Comparison Value" This statement, like all statements on my webpage refer to an application of Direct Comparison after the one tournament has already been completed.
You are right to observe that "Opponents' performance against third players affects the DC score" if two tournaments with different game results are compared and that "Pairings affect the DC score" while the tournament is still in progress.
Neither of your two statements contradicts mine though because you study tournaments that are still dynamic with respect to their game results while I study one tournament of that the game results of all rounds are already given.
Presumably I should have stated this context more clearly. Quite like you should also make this distinction between dynamic tournament still being in progress and static, already finished tournament clearer.
The next question then might be how to interpret all these findings for better assessment of the tiebreaker's quality.
Your statement "A game of go between two players will not always produce the same result as the previous one. Therefore, it makes no sense to take the result of one game and say that its result is also the result of the next game [in] a[n omitted] playoff." is a good one; much better than what you had tried to explain about this aspect previously.
Concerning significance, we all need a much better reference to formal definitions of significance. Currently I lack time to go that deep and therefore do not discuss your related comments yet.
Concerning frequencies of how often Direct Comparison does break ties in a particular tournament system, this should still be studied in much greater detail. So far only subjective guesses have been made.
You state: "It truly is unclear why some games of a player should be evaluated more often than others"
Ok, but this consideration can be taken against every tiebreaker. E.g. against SOS, where games with a high SOS contribution value more than games with a low one.
Since you made interpretation mistakes concerning my stated advantages of Direct Comparison, your conclusion "DC has only disadvantages." is a wrong implication.
That the EGD does not show the tournaments where I made place 1 by tiebreaker proves its incompleteness. (I also do not recall details, but I am sure to have tournaments by tiebreaker.)
My anecdote I have chosen because I remember it much better than lots of similar anecdotes I have heard from others.
You write: "If a tournament system measures something, then care must be taken to measure the same thing in the tie breaker."
If the same thing were measured, there would be no tiebreaker but the Swiss / McMahon Score would simply be repeated.
You claim that coin tossing would be more meaningful than DC. Where is your proof?
Can you please abstain from technical discussion in this thread and open a new one or use the discussion page?
Sure. But can you explain please which discussion page is for which purposes and for which not? E.g., why don't you want us to discuss technical details here but somewhere else?
Just so that we are clear on this, you seem to be saying that in your evaluation you have used the word phrase "Only a Player's Own Performance Affects His Direct Comparison Value" in such a sense that
1. If a change in a game's result would change the score, that is not "affecting". 2. If a change in pairings would change the score, that is not "affecting". 3. Your evaluation should only be applied in a context where by your definition it is impossible for anything to "affect" any scores whatsoever.
Please disprove at least point 3 by a counterexample.
-Bass
Hi
Regardless the tone of the above discussion, I agree with Peter HB, Tapir and Herman (I think), that we can link to RJ's page, since it is at least on topic, and maybe factual correct or plain wrong, and let the people do the thinking.
I'm not biased to either viewpoint, but do suggest we close this topic now.
Cheers,
Dieter
isd, you're criticising Robert Jasiek's page for not being useful, yet you add a link to http://users.durge.org/~gandalf/Dic.html ? And you make the sarcastic comment "Shall we discuss http://senseis.xmp.net/?DirectComparison%2FDiscussion ten more times ..." when it's you who chooses to stir up the discussion further?
If your goal is to improve the quality of SL, there are better ways of going about it. We already know that you disagree with Robert Jasiek. In repeating yourself, you don't make yourself look any more intelligent. If you have nothing new to add, then please leave the topic alone.
If Robert's page remains then I believe that my page provides a useful perspective on the subject. Satire is not a crime.
See above: "If you keep on picking on Robert on this page as you did before, we need to lock the page."
This page has undergone a long and tortuous series of edits and discussions. Once upon a time page was edited by Robert once to put in all his perceived advantages and disadvantages. Some people disagreed with these and edited them, a huge discussion page erupted. After some time the current page formed. I would like to believe the current page showed some consensus. If so, why include a page Robert constructed basically containing his original views which are no longer on the page because (perhaps) they did/do not stand up to reason. SOS-Lottery doesn't have anything to do with this, it is not an aspect of DC, nor is the precise formula one might use to calculate the standard error in MMS as a function of rounds and possibly rating. Love of Go is not important either. This is actually about the contents of the page. Yes I removed this external page link. I consider it to be misleading. PeterHB reinstated it, that's fine. So I created a discussion as to whether or not it should be there - better than an edit war. Perhaps I have been too frank in my views of the contents of this page? I don't see why you lock it though.
I am very well aware of the page history. However, you are continuing to misrepresent even the latest dispute. You did not just delete the link, PeterHB reintroduced it and then you started a discussion, but you are obviously obsessed with whatever Robert does, that is why you put this satire on your personal homepage. Even while there obviously is no lack of opportunity to express your opinion on Robert, on Direct Comparison, on Robert's preference of Direct Comparison (ample evidence for this is given by the lengthy discussions here).
And obviously you can't cope with anyone expressing disagreement with your implicit assumptions (Robert is always a loner and no one supports any of his ideas. Anything said against him is necessarily a sound argument.) That is why we read such insightful and highly on-topic comments as
"Shall we discuss ext http://senseis.xmp.net/?DirectComparison%2FDiscussion ten more times to gather a more statistically reliable viewpoint? Or shall we simply say that sticking to your original viewpoint carries more importance. "
"Yes and MMS does not measure anything, we don't know its limits. Lets just stop holding tournaments. We have all been idiots."
"Yes I see now that you are right, MMS cannot show us anything about the love of the game."
Cheers Tapir
PS And of course I have no objections to add a link to the page Bass wrote as that one aims on discussion rather than ridicule. Although the proposition "Not in a single one of his 144 tournaments found in the EGD has Mr. Jasiek gotten place 1 by any tie breaker at all." is provably wrong. ( http://www.europeangodatabase.eu/EGD/Player_Card.php?&key=10213203, sort after placement)
That is quite an irresponsible comment. You seem to only consider that I debate this topic only as a means to annoy Robert. You have not once discussed here any of the 3 advantages listed on the external page or the validity of a story about Robert winning a tournament if Direct Comparison was used instead of SOS. Instead you discuss SOS lottery and love of Go with me. What relevance do those have? I am not saying that Direct Comparison should never ever be used, I have stated so in various places here. Yet, if an argument is put forward that is patently wrong, why should we accept it? I don't think we should link to the page in its current form, it contains a tidier version of arguments already made on the discussion page, I don't think it adds anything of real merit to the current page.
If you want to disagree with my points please do, but please don't censor.
It should be obvious that my arguments above tried to show by example, that Bass' explanations are not necessarily always (though admittedly they are more than once) more sound than those of Robert and that your agreement with him (on the discussion page) doesn't indicate there is a consensus.
That is all, I did not intend to discuss all the issues touched by my examples here. And I do not see any necessity to contribute to the discussion of Robert's page, when the page here changed in the meantime and hardly resembles Robert's original anymore.
Cheers Tapir
PS I wish you may contemplate about whether any of your comments may have been "irresponsible" as well.
Sensei's Library represents one possible collection of information on DC. My page represents a different collection of information, which is independent of editing wars and flame baits. Every different collection of related information is worth being linked to. Quite like webpages by others are worth being linked to. This allows the reader to get a broader range of information so that he has an easier chance to inform himself more objectively.
To tapir,
You claim that my statement is provably wrong. Would you please present that proof by pointing out the exact tournament on the EGD, where Mr. Jasiek has gotten place 1 with a tied McMahon score.
Thank you.
-Bass
He won the 4th Baduk Botschafter Cup (german bracket) after a four-way-tie in the playoff, thus winning the right to represent Germany in the KPMC 2008. (The kind of win that doesn't end on the first place, like our European champions.)
Furthermore, I understood "Not in a single one of his 144 tournaments found in the EGD has Mr. Jasiek gotten place 1 by any tie breaker at all." is indicating that Mr. Jasiek never got a first place regardless of tie breakers. (That would surely wrong as he won several tournaments, but that may be a misunderstanding on my part.) But also the limited meaning that he never won a tournament after breaking ties is not exactly true. See above.
Cheers Tapir
I see. That doesn't really qualify as a "place 1" in my opinion, but ok, since it is a minor matter anyway, and as Robert says elsewhere in this thread, it's quite plausibly just a symptom of the EGD lacking data.
I have won something in between 10 and 40 tournaments, likely about 20. I do not know the exact number because I do not archive a list of wins. A number would be pretty meaningless because the tournaments are of very different importance. Also the numbers of participants say little about a tournament's value.
Tapir,
I think I would allow just the text:
"A satirical opinion on the
to be included on the page.
I don't like satirical baiting, but I feel being anti-censorship should be even handed. isd has as much right to have a link to his way of stating his argument as R.Jasiek. The wiki is for everyone. I think the links should be there and let readers decide their own points of view. I think readers can distinguish that it is an external link, and as such is not SL material.
If the linked to text was on SL, I think I would have attempted to change it, as baiting doesn't fit with the request for civility of discourse on SL.
I don't mind that you've locked the page for a while. Its a reasonable attempt to get things to calm down.
Perhaps some of the others ( HermanHiddema & Unkx80 perhaps ?) have wiser heads and will see a peaceful 3rd way out of this tomorrow. ( I feel a little out of my depth interfering in this disputed tie-breaker subject that I don't properly understand ).
Just to be clear on this point: I have no objection to isd or anyone else linking to a satire which is clearly marked as such. I was merely pointing out the contrast between isd's opinions stated in this forum and his action in editing the page. If isd's satire deserves to be linked to, then Robert Jasiek's own page is at least equally deserving of mention.
It is clear that qualities of tiebreakers are a hot discussion topic and will remain an interesting research topic for years. Therefore 1) currently agreement cannot be reached and 2) as much information as possible should be collected. Providing external links on the topic can only help, regardless of whether they state facts, beliefs or satire. Partial censoring would hurt progress. Subjective comments on particular links (like the current "widely disputed" as a comment on my link) also hurt progress. Progress is made when all opinions and contributions are openly available. isd, I tolerate your webpages - can you tolerate mine, too?
One common misconception is the idea that majority opinion would equal truth. Rather truth should be revealed by factual evidence and correct reasoning.
A ruleset that has been adopted by an association or federation has surpassed the state of expressing only the authors' preferences; it expresses the body's preferences.
Part of the problem is that these pages have sought to suggest Advantages and Disadvantages of various tiebreakers. This has allowed irrational thinking to enter the pages as an incentive to find ways of promoting favoured tiebreakers appeared. If Aspects has been included instead perhaps a lot of trouble would have been saved? I have updated my satirical page to reflect this.
![[Welcome to Sensei's Library!]](images/stone-hello.png "Sensei's Library")