There's nothing wrong at all about that; however, just because something is fun and somebody happened to have spent multi-thousands of dollars on a fun piece of equipment doesn't somehow make it more accurate, and that others should feels as if they cannot have a high-fidelity headphone system because they cannot afford $10k for an boutique tube amplifier. One can go a very, very, very long way for under $1000 start to finish, and have a very high fidelity system.
Tube-rolling for accurate sound is kind of like having a set of screw drivers and a bunch of nails. You can try whacking the nail with the different screw drivers in the set to see which works best at driving the nail---and you may even likely find a nice hefty screw driver that can get the job done---but the right tool for the job is still a hammer. Two things about this, 1) trying to find the best screw driver for pounding nails is stupid, 2) a nice set of screw drivers are perfect when you have a bunch of screws. A screw driver is simply not the best tool for the job when you want to drive nails, just as tubes are not the best active components to use when you want to have a highly linear, low distortion amplifier.
I love tube amps---I just use them where appropriate. You can certainly build a very linear, very accurate tube amp---it just costs you more to do so. If you want High Fidelity sound reproduction, you don't need a tube amp (in fact, you probably should avoid tube amps). If you want High fidelity from a retro amp design, tubes are the way to go---they're very retro indeed---it's just going to be a lot more expensive to get the same fidelity performance levels that one could get from a solid state design.
Now, if you need a high power, kilovolt-level power amplifier for a high impedance load, then tubes are probably a very attractive active component to start your design with. I just can't think of any dynamic headphones or speakers that would fall into that classification.
To actually address the OP's question/topic, I don't think tube rolling is pointless. I think it could be very educational if approached in the correct manner. 1) do a little background reading on amplifier design. 2) do a little background, nay, history reading on the operation principles of different types of valves. 3) tube roll and compare the change in response/distortion between different tubes---compare their data sheets--- see if you can make sense of the changes based on the circuit design and the different tube specifications. 4) learn.
Nobody these days invents anything new by brute force trying N! combinations of some system parameters before something magically happens to just work. That's sort of how evolution works and it takes Millions and Millions of years to get anywhere with that approach. Don't tube roll blindly. Formulate a hypothesis based on established principles and systematically test in a controlled way.
Edited by ab initio - 3/25/14 at 9:59pm