On the Senior TD exam I only got partial credit for the pairing question even though I checked my answer with SwissSys using the USCF default settings and it paired it exactly the way I did. Does this mean SwissSys doesn’t always give the most optimal pairings? They refuse to tell you what the “correct” answers are on the test so how am I supposed to learn what the “best” pairings are if SwissSys supposedly can’t even get it right?
If swiss sys and you both say the same then maybe you are misreading some fine point in the pairing scenario and thus you set up swiss sys with the same incorrect data/settings?
Of course it is frustrating not to be given the correct answer with reasoning.
I did get the pairing wrong on my senior test – I just missed a better pair in haste – swiss sys caught it.
Possibly, but I’ve checked it several times and don’t think I’m missing anything.
They do the same thing with WinTD, btw. I have never gotten 100% on a pairing question.
The Senior TD test is a problematic test:
- Testing Requirement.
Difficult objective examination designed to evaluate the applicant’s knowledge and judgment in situations not clearly addressed by the rules.
So, despite the claim of “objectivity”, the Senior TD test covers situations that the rules admittedly don’t cover well against a multiple-choice pick one of n scenario.
…that’s a subjective test with very little in the way of feedback.
1 Like
Couldn’t agree more. There were multiple questions on the test that I got wrong or only received partial credit for but after reviewing those questions, I still feel the answer I gave is just as good, if not better, as any of the other options.
US Chess Rules do not really have an overall “optimality” criterion in difficult cases where you have to do fairly wholesale rearrangements of a score group to get a valid pairing. For instance, suppose that to correct (alternation) in a score group, you can do a 70 point switch combined with a 30 point switch, or you can do a 60 point switch and a 50 point switch. Which do you do? The rule book is silent on that—either is within the rules.
Where the pairing program might differ from the strict interpretation of the rule book is in handling an interchange. Given a choice between a 1 point interchange and a 79 point transposition, the transposition is preferred. Given a choice between an 81 point interchange and an 82 point transposition, the interchange is preferred (per the rules). Except almost no NTD would ever pick the interchange in that situation (players may not like transpositions, but they really, really don’t like being swapped from the top to the bottom half), and it’s possible that your pairing preferences lean towards the latter. (Note, BTW, that FIDE Dutch rules don’t permit an interchange unless there is no set of transpositions, no matter how wacky, that can achieve the same level of color correction).
But FIDE Dutch is completely deterministic, right? That is, while USCF might have several different “rulebook permissible” pairings to be argued about in a given scenario, FIDE Dutch will only have one “most permissible”?
I remember not being in total agreement with the points awarded on the pairing questions when I took the Senior exam (closed book) at the US Open in 1987. I guess things haven’t changed much since then.
I have pretty much given up on the idea of a Keller Plan style of online TD training/certification, with a large pool of questions, all with explanations of the grading of each possible answer, in large part because nobody wants to write the 500-1000 questions that it would take to set up an initial test bank and then add another 100+ questions every year to keep it fresh. (I don’t want to write them, either!)
This would require a radical change in TD certification philosophy, from ‘see what they know’ to ‘make sure they know what they need to know’.
In addition, I’m also not in agreement with some of the multiple choice questions that I only got partial credit for or got “wrong”. I feel some of the multiple choice questions are completely subjective and have two answers that are both equally legitimate and not getting full credit on some of those questions is infuriating. If it turns out I’m wrong about that and one of the answers is clearly not as good as the other, it would be nice to know why cause I’m at a loss on why I didn’t get full credit for some of the questions.
For the pairing question, maybe the trick is to see how SwissSys pairs it and then choose different legal pairings!
IMHO, using SwisSys or WinTD to help you select the right answer is not quite cheating, but not really testing your knowledge of the pairing rules, either.
If a player comes up to a TD and complains that the pairings were wrong, and the TD’s response is “Nope, SwisSys/WinTD says they’re right”, that’s not going to be a very convincing argument.
I came up with an answer to the pairing question by myself and then just checked it with SwissSys afterwards and it paired it the same way I did.
1 Like
But did you understand WHY SwisSys paired it that way?
US Chess’s flexibility in pairing rules is both one of its strong points and one of its weak points, because two TDs (or two different sets of pairing program parameters) could pair the event differently and both be within acceptable definitions of ‘correct’.
And if you think a discussion of pairing flexibility can get testy here, try it at a FIDE meeting! I had the opportunity to spend several hours with Christian Krause (author of FIDE’s Krause format) in 2006, and it was possibly the highlight of my visit to the Turin Olympiad.
Yes (remember, I came up with that same answer before checking it with SwissSys).
1 Like
I wouldn’t guarantee that SwisSys’s logic was the same as yours, as it is quite possible for two different interpretations of the rules to result in the same pairings.
I’ve dug through the SwisSys logic on some sticky situations a few times and had more than one discussion with Thad about them.
And yet only one - or possibly none - would get full credit, and no explanation of why.
As I think you noted, this system is a little like trying to get cattle to put on more weight by weighing them instead of feeding them.
1 Like
My wife is an IT associate at the University of Nebraska. One of her duties is working with professors on their online tests.
In a recent class, the professor asked the students to submit possible questions for the final exam (with answers, of course), and several of them made it to the exam.
Maybe we could institute a ‘bounty’ system to encourage our experienced TDs to help build up a question bank?
I’ve also heard of professors using ChatGPT to generate possible exam questions, (and students who asked ChatGPT to produce potential exam questions, too), but I suspect ChatGPT hasn’t been trained on the US Chess rulebook. 
When I was in grad school, I took a course in instructional design (at the time I was in a PhD program but chose not to complete it), and we had a couple of very lively classes on the differences between ‘testing’ and ‘learning’ approaches to course design.
Did you find other pairings which accomplished the same level of color correction and were also “correct” according to the rules? You’re given two months to send in your answers—you have time to double check your work.
What’s the point, if there’s no way to evaluate one set as better than another? As you noted earlier:
1 Like
If you can do maximal color corrections with a single 50 point swap or the same with a 40 point swap, either of those is permitted, but one is clearly better, isn’t it? Now, if you are quickly doing pairings by hand, depending upon whether you start at the top or start at the bottom, you could find one or find the other, and software might, using its logic, be more likely to find one or the other. In either case, you/it might not look any further since the pairings are “legal”.
If you are doing a Senior TD exam, you aren’t under time pressure—you have two months to turn in the results. You can look more carefully at the options before submitting your answer.
2 Likes