MatPlus.Net

 Website founded by
Milan Velimirović
in 2006

23:09 UTC
ISC 2020
 
  Forum*
 
 
 

Username:

Password:

Remember me

 
Forgot your
password?
Click here!
SIGN IN
to create your account if you don't already have one.
CHESS
SOLVING

Tournaments
Rating lists
1-Oct-2020

B P C F





 
 
MatPlus.Net Forum Internet and Computing IDs for Big Brother
 
You can only view this page!
(1) Posted by Iļja Ketris [Friday, Nov 12, 2010 15:18]

IDs for Big Brother


As some of you are aware, WFCC is developing an online system for submitting entries to FIDE Albums.

The system has a potential to become a tool to facilitate many other tasks, such as automatic calculation of ratings, organizing online tourneys and more.
It will also inevitably become Yet Yet Another Chess Problem Database, and we welcome any input from all the existing databases, and propose to join efforts.

One of the question we would like to ask the community deal with the introduction of a "person ID", a short code that would uniquely identify a composer, solver, judge, etc., much like FIDE ID.
We may follow the beaten path and just assign numbers to people (423098, 561187 etc.), or, we may come up with some other approach. At the moment, it seems unlikely that anyone would need to memorize this code for any reason, but we don't know how things may develop in the future.

My proposal is to model the ID after "NIC handle": two initials and small number, such as "AB12", "PR9" and such, so it would be relatively easy for John Doe to remember his handle as JD20.
I would like to know opinions of people, who I am going to encode beforehand ;)
 
(Read Only)pid=6354
(2) Posted by Siegfried Hornecker [Friday, Nov 12, 2010 17:29]; edited by Siegfried Hornecker [10-11-12]

With the NIC handle you create a system that is impossible to maintain for future generations if there are more than 100 people with the same initiands over history. This might not be an important issue now, or in 100 years or so, but it WILL be an issue at one day. So you should rather use a random number for each ID, with at least 8 places (better 16) consisting of numbers, letters and symbols.

The ID should not be publically visible and never be
The ID should be encrypted
The transmission should be secure - yes, this means the use of HTTPS protocol rather than HTTP

Remember: Data leaks can cost lives.


Or even if you are not paranoid, you should disconsider the NIC handle just for the reason that it causes confusion with the hhdb sources handle (which also is two letters and two numbers).
 
(Read Only)pid=6356
(3) Posted by Miodrag Mladenović [Friday, Nov 12, 2010 22:03]

I do think that this is a great idea. However I would recommend to use at least four letters and two digits. For example John Doe would be something like JODO20. Of course since UNICODE is supported by computers nowadays I do not see any reasons why this would not be some Cyrillic letter too. Like for example Milan Velimirovic (Милан Велимировић - МИВЕ01). Yes, I know some letters do have same shape but different code (latin and Cyrillic E etc.). However if everything is done electronically that can be easily controlled and we can always assign different numbers for confusing combination of 4 letters (same look different codes). Since there are no too many problemists this should handle IDs for some longer period of time. Also, I hope that online collection does not mean entering forms. It would be better if problems are sent through the XML standard format published on the WFCC site. I think that it would be nice to have IDs for problems too. I think they should be some 10-digit number starting with 1000000001 etc.
 
 
(Read Only)pid=6358
(4) Posted by Ian Shanahan [Saturday, Nov 13, 2010 13:29]

What about using birthdays, in the form yyyy-mm-dd? E.g. mine would be 1962-06-13. Extra letters could be used in the rare event of coincident birthdays. I like Misha's idea of a problem id number: but it gets complicated when anticipations happen (as they inevitably do).
 
   
(Read Only)pid=6362
(5) Posted by Hauke Reddmann [Saturday, Nov 13, 2010 14:21]

The Hashguys theorem: IDs are always too short.
I already witnessed number collisions between ~10000 images
coded with 13-digit numbers. (Not mentioning that BBS will be
at 9999999999999 in about 100 years already.)

I would suggest vanity coding (like on a cellular keyboard)
if I wouldn't fear the wrath of Siegfried. :-)

Hauke
 
   
(Read Only)pid=6363
(6) Posted by Iļja Ketris [Sunday, Nov 14, 2010 01:10]

Thank you everyone for your input so far, and let me comment.

I never said that the proposed format is "two letters and TWO digits", it's "two letters and as many digits as needed".
Nothing at all mandates having handles of fixed length, so JD2 and JD8976 are both valid.

I would also reject right away any non-ASCII characters in the handles, as they must be equally well suited for input by anyone, and we just don't expect anyone (or even someone!) to be able to input Cyrillic, Georgian, Armenian, Kanji, Devanagari, Arabic, Hebrew, Thaana etc. etc. at the same time just to look for a person, so it will stay alphanumeric, meaning 26 latin letters and 10 digits, which also means that we will be sorry for those whose names start with Ö, Ā, Þ, and other non-ASCII latinics.

The idea of having more than two letters is interesting indeed, we just have to figure out what to do with those having "Mc", "O'", "van der", "het" and such.
 
 
(Read Only)pid=6366
(7) Posted by [Sunday, Nov 14, 2010 12:50]

Ilha Ketris wrote:

>One of the question we would like to ask the community deal with the introduction of a "person ID",
>a short code that would uniquely identify a composer, solver, judge, etc., much like FIDE ID.

The difficult part won't be the code -- it will be how it is maintained, and how robust it is.
Who creates the code for example -- to ensure it is unique, there has to be a formal process.
Can mistyped codes be identified? What will be the consequences for mistyped codes? Are they easy
or difficult to fix?
Etc.

Decide on criteria for the code now. What kind of requirements must it live up to? Once those are
accepted, then design the code.

>My proposal is to model the ID after "NIC handle": two initials and small number, such as "AB12",
>"PR9" and such, so it would be relatively easy for John Doe to remember his handle as JD20.

But John Dawson, who has JD200, and who is not a great typist, will probably enter JD20 when he means
JD200 ... what kind of problems would that lead to? And when John Doe (or Joan Doe, for that matter)
changes name ... what happends then? (That is, is it a requirement that the code remains constant ... or
isn't it?)


For general robustness, avoid connecting the code to initials. (On the other hand, that may make it
more difficult to remember). Also avoid including information that may be sensitive. (For example,
a code based on date of birth would could mean that identity theft becomes easier -- in Sweden, for
example, the date of birth is two-thirds of the 'personal number' (a kind of social security number),
and the remaining digits are not too difficult to guess.)

For further robustness, avoid codes of unclear length. Rather make the codes self-extensible (so that,
perhaps, JD1 means there are exactly four/five digits following, JD2 exactly six/seven, etc.)

For still further robustness, include a check digit that catches simple mistypes (one wrong digit/letter,
one digit/letter added/deleted, and two digit/letters transposed.)
 
 
(Read Only)pid=6367
(8) Posted by Siegfried Hornecker [Sunday, Nov 14, 2010 13:27]; edited by Siegfried Hornecker [10-11-14]

And don't forget: There could theoretically be more than one composer with the same name. We already had one Hans Müller in Austria, and this is a very common name. So there must be other uniquely identifiable information.

Since you want people to contribute anyway, why not let them choose a nickname? Then you can assign a number to it and still have it identifiable. This works on Wikipedia and MatPlus. Just make sure that it is not publically visible, just for security reasons. Also, it must be possible to contribute on behalf of other people or via offline ways.

The upcoming century will bring means of war that could destroy internet access for countries for several years. Or if it isn't war, it is fascism, dictatorship, whatever criminal politics you want to name. Already in many nations internet is censored.

Too many nations to list here, including many so-called "western" ones, already have censorship laws on internet, thus should be regarded unfree, and it can't be assured that people from those nations will be able to access internet permanently. [1] [2] [3]

For at least all those nations other means of contributing to FIDE album must stay open.

[1] http://en.wikipedia.org/wiki/File:Internet_blackholes.svg
[2] http://de.wikipedia.org/wiki/Zensur_im_Internet#Zensur_durch_Regierungen (accessed version: ID 80873900)
[3] http://en.wikipedia.org/wiki/Internet_censorship (accessed version: ID 396043923)
 
 
(Read Only)pid=6368

No more posts


MatPlus.Net Forum Internet and Computing IDs for Big Brother