Just for fun, let's see if we can make a reasonable estimate of this.
In what ways can a U.S. poker player end up owing more money in tax due to his or her poker activity in a year than he or she made at poker? The two biggest ones are probably those that will cause even losing and break-even players to owe tax due to their poker play:
#1: Loss of Standard Deduction outweighs wins
Every losing player who would otherwise take the Standard Deduction and loses it for having to file, say, $6,000 in Gambling Winnings and $6,700 in Gambling Losses, will end up paying more in tax for losing at poker than he or she would by playing no poker. This will also apply to players who win only a small amount and don't file as professionals.
#2: Players in "bad tax states"
All losing players in states which disallow Gambling Losses as a deduction will owe tax on their net loss. Winning amateur players in this state can owe more tax than they made if their gross is much higher than their net.
Let's ignore the winning players who trigger these (hard to estimate) and just focus on the losing players.
My personal guesses of the relevant parameters:
# of U.S. poker players: A
PPA survey claims 55M U.S. poker players. I don't know how broadly this survey went, but it's safe to assume not all of these individuals play poker more than a casual game a few times per year. Let's assume that somewhere between 50%-80% of these players play poker at least every month or so (even a small player can trigger the above tax issues).
My estimate: 27,500,000 to 44,000,000
% of such players that lose money on the year: I remember reading somewhere that somewhere between 80%-85% of PokerStars players are down money on PokerStars lifetime. To be conservative and to account for different types of games, the range could be wider.
My estimate: 65%-90%
% of people who take standard deduction if their poker weren't taxed: I have no idea, and I couldn't find anything on the internet. My understanding is that many people do just take the standard deduction, but maybe that's just among young people.
My estimate: 25%-90%
% of the above who lose their standard deduction due to proper poker tax accounting: Hard to estimate for the casual player, as it depends on frequency and stakes of play, so I keep it broad.
My estimate: 10%-50%
% players living in a bad poker tax state: These 10 states comprise about 30% of the total U.S. population, but they probably have fewer poker players per capita than states like Nevada. Also, some of these states merely limit itemized/gambling deductions, which won't affect the low-volume losers and break-even players.
My estimate: 10%-35%
With these estimates, restricting our consideration only to losing players who fall into #1 or #2, we are looking at 3,141,875 to 27,907,000 Americans who currently owe more tax on their poker play than they made at poker in a given year. Adding in the consideration of winning players owing more than they made via #1 or #2 adds another 200,000 to 4,000,000 based on some loose assumptions. I welcome more informed estimates of some of these parameters. I tried to be conservative with the low end, and even if I was off by a bit, I think we're looking at at least a million.
Naturally, of these million+ Americans, I suspect that somewhere between 11 and 13 of them are actually dutifully paying their taxes correctly right now.
With a potential future of online poker sites that report all players' tax information to the IRS, it might be useful for some organization to make a more formal estimate of this to demonstrate the impending disaster of these million+ people suddenly all being forced to pay their taxes properly. I see this as an important argument towards an internet poker bill which includes a change in the tax code for internet, if not all, poker winnings (if this is politically possible). Note that the only change necessary is allowing poker players to net their entire year's play instead of reporting gross winning sessions and deducting losing sessions, which is easily demonstrable as a reasonable and practical approach for a game like poker anyway.
Cliffs: I estimate that at least a million (possibly much more) Americans currently owe more tax because of their poker play than they make at poker.