I submitted the event US Chess Wednesday night. Mostly it went very smoothly. Chronologically:
I liked that validation results were presented on the server without the lag of email.
I didn’t like that validation complained about two out of state players not being from the state as the majority of participants.It’s of no use to me and cluttered the validation.
I didn’t like that MUIR cared so much when someone’s name from the import was “Mike” when the ID has the name “Michael”. I had a lot of these. It should just use the Member’s real name.
I found it confusing where to including assistant TDs. They were in SwissSys but, as reported elsewhere, not in the report. I found the entry template for TDs a little klunky
I LOVED how fast the tournament was rated. It was just a few minutes.
All that said, I have a question about results from that tournament. A new player signed up and played 2 games in one section plus an extra game. MUIR rated each of them as if he was unrated.
From the section
From the extra game
His MUIR page doesn’t have either result (I recall it’s because that’s now aligned to supplements?) and it’s not clear which was rated first. So my questions are 1. why? and 2. what should his provisional rating be when I set up the next club event for next Wednesday. If it helps, the extra game was played first.
At this point they are trying to not give away age of a player so they have blanked out some of the info and they are working on adding the number of games not showing up.
1300 is the rating of an adult who only won two games against very low rated players so that was rated first. 1300/2
1849 is the rating when you add in the 3rd win including the win over the 1465 so 1849/3
Since the player does not have 4 games he is technically still unrated with a live rating or once it gets into the next supplement.
This also matches up to the section numbers 3 & 4 ordering.
I was once asked by a kid does the extra game get rated before or after the regular tournament games and I told him I ordered that section last so it gets rated after the regular games to which he replies I should have ordered it first since the extra section was over (last game finished) before the regular section’s last game was finished. For established ratings I am thinking it seldom matters more than a single rating point or two?? – but in an edge case it could mean missing a norm or class.
The ordering for newly rated events is being worked on, when we do the first rerate they will be rated in section number order, assuming the starting and ending dates are the same for both sections.
I’m hoping for improvements in the ‘does this person belong’ coding, I think what the previous validation program was doing is complaining if the state code was different AND the age or rating was well above or below the average of the field. But it took me a couple of years to tune in that code to where I was happy with it, not too many false alerts and few people slipping through the cracks.
What the name thing is noticing is that the name you entered (probably Mike) was different from what was in the player record (Michael). I’ve suggested some refinements to that, I don’t know if they’ll be used. First name differences (Bob, Robert, Bobby, Bud) are a big challenge, and so are ones where the first and last names are reversed because of cultural differences. I wonder if anybody has a table of first name alternatives, I can think of at least a half dozen for Michael.
Doesn’t the pairing program pick up the name from the supplement file, which should use the one in the member record. There is one issue with that, the name field in the supplement files is not long enough to accommodate some really long names. Players from India tend to have really long first and last names, for example. Programs that use the all_players file (tab delimited format) may help that, because it has the whole name since it isn’t a fixed length field. I’m kind of hoping that will become the first choice for the pairing programs in the future.
Picking up the 5 TDs from the import files is also on the list of corrections in the pipeline.
MSA did not have most-recent event ratings on it, either, only published ratings, so that’s no change.
I don’t think I’ve run a test on rating the extra game before or after the main section since we switched to floating point ratings, back in the integer days any differences were usually no more than a point or two, which could have been due to rounding ratings to integers. We once looked at trying to lump all of the sections in an event together if they had the same starting and ending dates, which would put the extra game in with the other games, but that got messy and affected bonus calculations. And I’ve seen some two-day events where the extra games section was listed as a one day event, which could affect the order in which they are rated. (If it is a saturday-sunday main event and the extra games section is listed as saturday-only, then the extra games section would be rated first.)
I don’t think I’ve run a test on rating the extra game before or after the main section since we switched to floating point ratings, back in the integer days any differences were usually no more than a point or two, which could have been due to rounding ratings to integers. We once looked at trying to lump all of the sections in an event together if they had the same starting and ending dates, which would put the extra game in with the other games, but that got messy and affected bonus calculations. And I’ve seen some two-day events where the extra games section was listed as a one day event, which could affect the order in which they are rated. (If it is a saturday-sunday main event and the extra games section is listed as saturday-only, then the extra games section would be rated first.)
Here’s kind of an interesting scenario. Suppose you have a three week game-a-week event and there is an extra games section for each day. So the 4 sections are
Our program director insists on using live ratings as of round 1 (or player’s first game) so usually not from a supplement. I routinely import a club list with updated ratings and that list has friendly names. Also, the ID was used as a source of truth under the old ratings system which then looked up the real name and avoided the validation issue. The new system could do that.
About the out of state check, I could see it helping clubs near state borders when new players join an event. I can’t see major organizers ever proofreading the validation output for the state. Besides, more and more registrations are online (and assumed right?).
TD/A never looked at the name fields in the upload files, because I could never figure out if they helped spot any potential problems. (And I’m the fool who added them to the 2C specs.)
The new system does look at them, and my guess is that will eventually be downgraded as not very helpful. I spent decades learning what worked best, and I threw out some ideas that I thought were great but in practice were a mess along the way, and it will take time for the new system and its overseers to figure out what works best under it as well.
To me, the roughest spots are for those TDs who enter their events online, and we’ve already got several ideas in the pipeline for speeding that process up.
Someone asked me in email why we’re spending all this time and money rewriting the system, and I gave that person a 3 word answer: I am 76.
Well, that’s kind of a mixed bag. Newer players don’t remember their IDs, I’ve typed mine so many times it probably would show up on a brain scan. And the thing that would help people the most with getting the right player ID, showing things like birthdate and address, creates too many privacy issues.
You built a strong system with tons of institutional knowledge baked into it. A rewrite was definitely needed considering your desire to step away.
The problems are with Leago. Why did they think they could do it this quickly and cheaply? And there seem to be numerous and blatant Chestertons fence violations from the team. And so many obvious bugs. It makes me worry about the quality of the partner we’ve all gotten into bed with.
It has taken too long relative to what they said it would take.
That estimate was way off. Which means US Chess probably mis budgeted.
I could have never judged how long it should take in absolute terms because US Chess didn’t really share very much of the scope until a few weeks before going live
I’ve said from the beginning that this was sorely needed.
Not so much a desire as a recognition that everybody’s time runs out at some point, usually before they expect it. Officially, I retired in 2016. And my wife would say: Yeah, right.
Having read all the responses to the RFP numerous times, I think Leago was by far the most qualified bidder. And they weren’t the lowest bidder, as I recall. (The highest one quoted a price that was several times larger than our annual revenue.)
In some ways, the feature list for the initial rollout may have been both too expansive and too restrictive, but having spent MANY hours explaining things to the Leago staff, they understood the basic tournament model, but the nuances that you get from working with the system for a long time weren’t there yet. (I was a TD and organizer for 20 years before I took a shot at writing the previous system, a task several people had attempted but failed badly at, many insiders in US Chess thought having an online submission process was impossible. But several people including Bill Goichberg were smart enough to realize I knew the business model and basically left me alone for a good six months while I worked on several different approaches to the task, during which time I came up with the rerate system, something the Ratings Committee initially thought was mathematically invalid.)
Some of the bidders struck me as not knowing our business model at all. And some had the classic attitude of “I have a hammer, so what you are is a nail.”
This is a common business HR concern and complaint by employees. For US we tried with some success to have both legal First & Last & a separate callname / preferred which sometimes looked nothing like their legal name maybe 75% of the time it was entered as obvious nickname for their first name - middle name - first name initial and middle name or just inititals – maybe with a title prefix. I know quite a few folks who the nickname is part of what is in the lastname field –
When I manually register folks in swiss-sys, I override the name quite often when I know the person uses the nickname (I think Michael to Mike is most common) – It is usually worth it just to reduce the width and most are just removing middle initials or prefixes/suffixes
I think the membership system (CIVI-CRM) has both formal and informal name fields, too. But I don’t know if we’re utilizing them.
The Mike/Michael thing is kind of funny. All through elementary and high school I went by Michael and my best friend went by Mike. He became an architect and is an expert on Frank Lloyd Wright restorations and within his profession is known as Michael, while in the chess community I’m know as Mike. (And my grandfather almost always spelled my name Micheal.)
