choice of color in rated games

I don't see why this is an issue. Just don't play people if you don't want to. Or create a seek with randomized color.

I know that lichess ratings don't correspond exactly with FIDE/ USCF ratings, but it would be good to know approximate percentile. I know detailed statistical information is coming soon on lichess. But when it does come, it will still be inaccurate because people choose colors/ opponents/ etc. Although I do not have empirical evidence to support this, I believe that if everyone (or most) around you do not have true rating (extreme e.g.: by playing 90% white games, etc.), it is hard for you to obtain true a rating also, because Elo is a relative measure.

Cafestream, I think I am correct in saying that there is no such thing as a 'true' rating. The number is only a relative strength indicator against those who you play . Carlsen's ELO is only a relevant number to those who have played FIDE and to those of his own generation. You may say that 90 percent white playing guy does not have his true rating but he does. It is true to him playing 90 percent white!
I always choose random colours :)

I always choose random as well. UNLESS

Sometimes I want to try out a new line from either a black or white perspective.

Players should be given the choice to play what side they want. It's like if you were at a chess park, you can sit on the side you want to play and if an opponent doesn't want to play the side that's empty, then they simply won't play.

I understand where cafe is coming from, and I do KIND OF wish that all rated games were randomized, but I lean more toward freedom of choice, and I highly doubt that the system will change even if several people were to agree with cafe anyway.

In principle, I agree with Dionysus #14. Freedom of choice is quite important! Also, I agree with the point that ratings will not change considerably if we change to a theoretically fair system.

After reading the comments here I believe that the overwhelming sentiment is that people want the choice, so I no longer would want to argue for random colors.

Just a little bit of nit-picking @ #13. - here is how I define true rating.. it is that number where your rating will tend to, no matter whether you start from 10 or whether you start from 5000. I know our strength does change from game to game but it will have a mean and a small range (e.g. 1550+/- 40 SD); in some games you will make huge blunders and play at 1000 level or in some games you will play brilliantly and play at 2500 level, but those will be rare. This mean wil be different if you play only white games, or only black games (e.g. with white games it could be 1580 +/- 30 and with black games it could be 1510 +/- 60. All these numbers are simply made up to illustrate the point.

This argument is pointless because even if the games were randomized the players who only want to play white will just abort the game, which is far more annoying (it already happens all the time as it is, it would just get worse). We're playing online, the ratings are already inflated by cheaters having their engines fail, drunk opponents, disconnects, etc. And in essence, players can already inflate their ratings simply by never playing stronger opponents. It's no different than only ever playing white. The pools would have been a good way to solve both issues, but they weren't popular, especially in longer clocks. So basically, it is what it is, there's no reason to fix what isn't noticeably broken.

@static_shadow : that problem is easily solvable. (1) Any game that starts (even before move 1) cannot be aborted manually. (2) Game aborts automatically if 1st move takes more than 30 seconds (except in chess960 where more time may be needed ).

But it any case, I now retract the idea of randomization, after observing that most people prefer the status quo.

@static_shadow, my answer was only regarding aborting games. I agree selection of opponents is a big problem, and I have no solution to that, except a pool like system which did not gain traction

Also, I have to respond directly to #15 because I missed it.

Cafestream, your numbers were made up to prove a point but they actually prove another point. That's all these numbers are...they're made up by statistical comparisons between games played over time against strength of pool. I can play the same one player 100 times and have a 'true rating' based on your statements, provided we alternate starting sides. It's all just hogwash and makes it seem you haven't the slightest clue how ELO or Glicko work. If you want a 'true' rating, your best option is to play at club level over the board against a pool of mixed opponents with adjustments for strength of schedule. That's the only way you get a 'true' rating, but even that can vary because a 1500 rated club player with 100 games in Calcutta (where chess is very popular and studied in schools) vs. a 1500 rated club player with 100 games in Indianapolis (where chess is more seen as a hobby) vs. a 1500 rated club player with 100 games in Hong Kong (where serious chess is rarer than the US, but on the rise) are all going to have much different skill levels based on the caliber of their opponents in their club pools and tournaments. Give the Indianapolis 1500 player white all day against the Calcutta player and I would imagine the Indianapolis player will end up at 1350 by the end of the day. There's no 'true' or 'accurate' way to measure skill over the board, it just isn't possible. The rating systems we have simply attempt to approximate that, and some systems are better than others and some communities are going to be a more accurate place to assess that skill level than others.

And to #17, if you want a system of 30 second game aborts, randomized forcing, and so forth, only play tournaments. 30 second aborts would ruin longer clocks where people are potentially more sociable at the start, and it definitely wouldn't fly in correspondence.

This isn't about people 'preferring' the status quo, it's about your request being unsound.