Age sourced from fide db rounds differently on diff screens

[ijm8710]

Seen this with most players. Seems one screen rounds up. The app is showing it wrong as well as I believe 17 is correct per Wikipedia

[PawelPlutaUek20]

The problem

I don't think it's a rounding issue.

In javascript, the age is calculated from the start of the event (which would be December of the past year)

lila/ui/analyse/src/study/relay/relayPlayers.ts

Lines 147 to 149 in f24a385

	const year = (tour.dates?.[0] ? new Date(tour.dates[0]) : new Date()).getFullYear();
	const tc = tour.info.fideTc || 'standard';
	const age: number | undefined = p?.fide?.year && year - p.fide.year;

In scala, the age is computed from now:

lila/modules/fide/src/main/FidePlayer.scala

Lines 44 to 47 in f24a385

	def age: Option[Int] =
	val nowYear = nowInstant.date.getYear
	year.map: birthYear =>
	deceasedYear.fold(nowYear - birthYear)(deceasedYear => deceasedYear - birthYear)

I think that one issue is that Fide does not share the players birth date, they only share the birth year:

ID Number      Name                                                         Fed Sex Tit  WTit OTit           FOA SRtng SGm SK RRtng RGm Rk BRtng BGm BK B-day Flag
21874697       Seemann, Jakub                                               POL M   GM                           2525  12  10 2415  0   10 2359  0   10 2008      

Here's the event mentioned with it's starting date , just for the record:

https://lichess.org/broadcast/53rd-rilton-cup/bCQK73un

Solutions

A possible solution would be to always display the current age, not relative to the start of the tournament. This way it would show the same age in both places, but it would still be wrong. Another solution would be to display the birth date instead of the age.

[ijm8710]

Only things that are surprising are

1. not sure why the example of
   A player born in September is affected by a December event
2. ignoring what fide shows, internally lichess shows 17 on one screen and 18 on another (in the dual photo I shared). So it's weird that "we" calculate it two different ways. Are we claiming both are wrong?
3. if you feel we can get birthday from Somewhere why not just infer it relatively from there?

[PawelPlutaUek20]

1. Lets imagine an event from the year 2000. For some people would be probably more interesting to know that Magnus at the time of the event was age 10 and beat some GM, rather than knowing that he's currently 36.
2. Since fide does not share the day and the month of the birth date, it's basically impossible to properly compute the age, so yes, they are currently both wrong. One computes an approximation of the age of the player when the event started and the other and approximation of the current age of the player.
3. I think one of Lichess’s core principles is to keep the codebase simple. Inferring the birth date from multiple different sources is not ideal. Probably replacing the age with year of birth would be the most correct. But I guess age is more user-friendly.

[ijm8710]

I don’t disagree. I was just surprised the example showed him as 18 under either method when he isn’t 18 yet. I understand you’re saying these are approximations, but it sounded like they were tied to the timing of the event. The event ran in December–January and didn’t change his age—his birthday isn’t until September. He was 17 throughout the event and remains 17 now. He won’t turn 18 for another eight months. I get that the event ended in 2026 and he will turn 18 at some point that year, but that hadn’t happened yet.

I guess it solely uses the years and not months.