APBRmetrics

Why are possessions estimated using crude formulae, instead of being properly and precisely counted?

For example, Basketball-Reference (http://www.basketball-reference.com/about/glossary.html) uses

0.5 * ((Tm FGA + 0.4 * Tm FTA - 1.07 * (Tm ORB / (Tm ORB + Opp DRB)) * (Tm FGA - Tm FG) + Tm TOV) + (Opp FGA + 0.4 * Opp FTA - 1.07 * (Opp ORB / (Opp ORB + Tm DRB)) * (Opp FGA - Opp FG) + Opp TOV)).

I believe the NBA still uses (http://www.nba.com/thunder/news/stats101.html):

Pos = .96 * (FGA + (.44*FTA) - OR + TO)

I understand that for earlier years, we don't have more detailed play-by-play information and so precisely counting possessions for earlier years may not be feasible. Instead we use the above formulae to estimate the number of possessions.

But surely it is a simple matter to count possessions for games taking place today, now that we have all the play-by-play data we need? Why isn't this being done?

I could say the same of many other stats — e.g. True Shooting Attempts is defined as FGA + 0.44 * FTA. Instead of using the crude 0.44 multiplier, surely we can nowadays actually determine how many True Shooting Attempts a player actually made in the game?

Good points all.
Player per-100-possession stats are also summed based on minutes rather than possessions. Go figure.
On a given team, all players are assumed to go the same pace -- proportionate to minutes -- even when we know their on/off team pace.
Team and opponent rebounds/100 are also treated the same in calculating Reb%, even though we know there are on/off team rebound rates for players.

And shouldn't we count a possession starting with 6 seconds in the quarter, as .25 possessions?

True Shooting Attempts is defined as FGA + 0.44 * FTA. Instead of using the crude 0.44 multiplier...

Yes, and in a box score a player goes 1-2 from the FT line, no FGA, and his TS% = .568
Is that a "true" shooting% ?
We failed to come up with a better concise term some years ago.

I always use the first formula. In some cases, I used even more sophisticated formulas. I don't like "too much estimation" or "bias". I always try to minimize that factor in any scenerio.

My guess is that estimates are still used commonly because it's a bit easier and, generally speaking, good enough for most purposes. I'd rather use exact numbers, but if I can estimate something that's 99% accurate, than that 1% of error won't matter for most projects.

nbawowy counts them (or tries to, at least).