USCF ID now starting with 130

Running Chess Tournaments
1

Checking my tournament history, one posted 12/2/04. See and did play with a new player with a USCF ID number starting with 130, in fact there are two new players with these first 3 numbers. If talking with a person that joined the federation for the first time, as a director able to tell the player their number will start with 13?

Wounder how many years did the numbers were issued for 10, 11, 12: when they were issued for the first time and when they were no longer issued? It is a fair question, as a player or as a director, it can give a idea how long ago they first started to play chess.

2

There may be others who know more of the history than I do, but here’s what I know, with a few guesses thrown in.

First, here’s a quick count of the members we have in our records by the first 2 digits of the ID:

10 - 15745
11 - 6011
12 - 519096
13 - 2260

IDs that begin with a ‘2’ come from the pre-numbered JTP forms that were first issued in the late 1980’s. There is no correlation between those numbers and when the IDs were first used.

When the USCF first assigned ID numbers (around 1978) they did so using a list that was ordered geographically, more or less moving east-to-west. The ‘100’ numbers, for example, all appear to be for people who were living in the east. There were no IDs issued that began with ‘109’.

Since I was living in Illinois at the time, my ID number (10339324) is in with a bunch of other Illinois players from that era.

I think that after they numbered all the existing members, they started assigning numbers to new members that started with ‘11’.

I suspect that at some point, probably when they changed computers, they jumped from ‘11’ to ‘12’. There are no ID’s that begin with 116, 117, 118 or 119.

I don’t have complete information on how many ID numbers were assigned each year. Tom Doan and I have taken stabs at trying to find the first ID for each year, since IDs have been assigned sequentially over the years, with the major gaps coming when they moved from one hardware platform to another.

I know that the ID’s starting with 1251 were assigned back in the mid to late 1980’s, probably around 1987. By June of 1992 they were assigning IDs that began with 1257. (That’s when the USCF started issuing rating supplements on diskette files, so we have pretty good track of the numbers issued from then on.)

IDs that begin with 126 were first assigned in about September of 1993.

IDs that begin with 127 were first assigned in about March of 1997.

IDs that begin with 128 were first assigned in about June of 2000.

IDs that begin with 129 were first assigned in about June of 2003.

The first ID that begins with 13 was assigned in November.

Shortly after we switched to the new membership software earlier this year, we started using a variant on a MOD 7 check digit for the 8th digit, which means that IDs are no longer issued consecutively. (If you ignore the last digit, they still are issued consecutively.)

3

Does this mean that 8 times the 1st digit, plus 7 times the 2nd digit, plus … plus 2 times the 7th digit, plus 1 times the 8th digit, is supposed to add up to a multiple of 7? (That’s what it means for the mod-11 checkdigits used by book ISBNs.)

If so, then there’s a weakness. Namely, the 2nd digit won’t affect the checkdigit at all, i.e. two ID numbers that are identical except for the 2nd digit will have the same checkdigit.

Bill Smythe

4

Nolan:

Thanks Nolan. Have been looking at my number for years, (12313120) and find not that many people with that digits (123). Very sure it was issued in 1980 or 1981, my memories of childhood and the federation are not that clear on dates. It could have been being away from the federation as a active member from 1987 - 1997. When looking at the 3 digits, do not see many 120, 121, 122 or mone 123. Can see a lot of members with digits of 124, 125, 126, 127, 128 and now 129. It could be like the first two digits of 10 and 11, not issued that much, for the reason not seeing the three digits of 123.

5

Bill, ISBN uses a MOD-11 check digit, which is why there is an ‘X’ in the check digit of some ISBN’s. That wasn’t an option for the USCF.

MOD-11 is considered to be a bit better than MOD-10 for error catching, but we’re using a variant on a MOD-7, which does a better job of catching single digit errors and two digit transposition errors than either MOD-10 or MOD-11.

MOD-7 also has the virtue of being somewhat tuneable to specific data patterns, which has to do with what kind of data errors you get.

The precise scheme was chosen after an analysis of a list of known ID keying errors from crosstables submitted over the last two years.

Doug, because of the USCF’s data purge policies over the years, we are missing many of the ID’s that were asssigned in the distant past, roughly from 1978 to 1987.

However, we have about 98% of the ID’s issued from about 1251xxxx on, which I think takes us back to mid 1987.

Here’s a list showing the total number of ID’s we have by the 1st 3 digits of the ID and the number of those that are current USCF members:

mem total active

100 2852 1790
101 2915 1899
102 2558 1679
103 2987 2048
104 2937 1858
105 912 457
106 142 42
107 124 38
108 140 46
109 178 57
110 890 337
111 919 268
112 1066 228
113 1205 251
114 1533 344
115 398 80
120 1814 390
121 1828 370
122 2369 462
123 4313 701
124 74409 4179
125 92622 4070
126 98879 3847
127 98282 6209
128 99030 17002
129 45550 27672
130 2383 2118
(27 rows)

6

Actually, I was just asking a technical question.

With MOD-11 applied to a 10-digit number, what happens is that 10 times the 1st digit, plus 9 times the 2nd digit, plus … plus 2 times the 9th digit, plus 1 times the 10th digit, is supposed to add up to a multiple of 11.

In other words, to calculate the 10th (check) digit, you would take 10 times the 1st, plus … plus 2 times the 9th, and then:

  1. Divide by 11 and consider only the remainder (a number from 0-10 inclusive).

  2. If the result is 0, the checkdigit is 0. Otherwise, subtract the result from 11, giving a number from 1-10 inclusive. This number is the checkdigit (use the letter X for 10).

BUT, if a similar scheme is adopted with MOD-7 instead of MOD-11, the scheme will not work well if the number of digits (including the checkdigit) exceeds six. This is because the 7th digit from the right (e.g. the 2nd digit in an 8-digit USCF ID number) won’t figure into the checksum at all. 7 times the 2nd digit, when divided by 7, will always give a remainder of 0, no matter what the 2nd digit was. In other words, two USCF IDs with different 2nd digits, but with all other digits the same, will have the same checkdigit, so the checkdigit will not have performed its function of detecting a data entry error (if in the 2nd digit).

Bill Smythe

7

I don’t want to debate mathematics, because I don’t need to understand the theory, just the practical uses of it. (That’s why I was in the engineering college at Northwestern while you were in the math department.)

I can say that the types of ID errors that the USCF gets, which are largely two consecutive digit transpositions in the last 4 digits of the ID, lend themselves to a customized MOD-7 check digit scheme. As long as the multipliers of those digits are relatively prime to each other and not a multiple of 7, it will catch nearly all such transpositions, certainly more than what one would expect a MOD-10 or MOD-11 check digit scheme to catch. (Only the internal computations are MOD-7 the final one is MOD-10.)

I suspect you can look up the details in Knuth’s “The Art of Computer Programming”, probably volume 2.

8

OK.

I was hoping somebody could explain to me the exact scheme used in USCF’s version of MOD-7. Apparently it’s not exactly as I speculated in my last post, especially since you use the word “customized”. I assume it’s something like N1 times the first digit, plus N2 times the 2nd digit, plus … plus N7 times the 7th digit, where all of the Nn are in the range 1-6, and no two consecutive Nn’s are alike, or something like that. When you say the final computation is MOD-10, does that mean you divide the result of the previous step by 10 rather than 7? That would make sense, because otherwise you would have only 7 digits available instead of 10 for the final digit, which in turn would mean that the checkdigit could not always distinguish between two different values in the same digit position somewhere else in the string.

Anyway, if the scheme always catches single-digit errors, and always catches transpositions of two consecutive digits, that would seem to be the best that can be expected, and exactly what is needed.

Bill Smythe