I understand that TR/X is generally weaker than X/X, excepting draw. The point is that TR/X is high variance and leads to games where the game is effectively over by T4. TR/Baron is crappy odds, but if it hits it can mean getting a Plat in play with the 2nd shuffle (T5). Likewise, TR/Coppersmith allows for an early plat. A number of other cards - like navigator, can make for high variance TR openings (TR-Navigator - leave 2 cards on top of your deck and look at playing a T4 gold).
Pretty much, these things mean that if a player goes for them and misses collision, he loses, which makes the game less fun for the other player as he is no longer challenged.
If they go and hit, it makes it very hard for the other player to come back.
Either outcome makes a for a game lacking in challenge and often fun.
In multi-player, say 4er, going for something like this might be your best odds of winning. A 30% chance of getting your TR/Smithy and getting first crack at the good cards (like goons) is much better than mirroring and taking your 25% odds.
WoG: yes I get that, TR/X is worse on average, but it is high variance. For many cards, when it works ... it REALLY works. It decides the game very early on.
RJ: None of 3 cards can draw better than +4 cards with TR/X - leaving you with a net +2 cards; further most all of the +2 cards options are stuff that tends to be weak - wishing well, shanty town, moat etc. Even with an early gold and some cycling, you still have a weak deck. That is not true with the 4's where payouts can be much bigger (I see multiple ways for TR/X with 4's to hit 9 coin) and much more useful late game.
Currently the highest variance TR/X openings are TR/swindler (2 chances to turn a starting buy into something useless) and TR/Chancellor. Both have a shot a buying a T3 gold, but both are pretty weak. Variance is just not as high for TR/X with 3s as with 4s.