It's sloppy chess, but it might work at a higher level than many people want to admit. For an extreme case, take Sam Sloan's Damiano Defense with 3.Nxe5 fxe5?? where he knowingly accepts a lost position in standard time controls to have the opportunity to demoralize his opponents. I think the Halloween Gambit (Four Knights with 4.Nxe5) is almost -/+ (or at least =/+) with accurate play in a couple of the most critical lines, but when I studied it carefully, it soon became clear to me that it is not an easy opening to refute "cold" OTB. It's funny. Sometimes I have a fascination with "improper" openings. If this were Go, they would be called
hamete in Japanese which roughly translates to "trick plays." Often they are most likely incorrect, but a strong player should be able to figure out why beyond the fact that they are not
joseki (standard). I will almost never play an opening that I believe to be incorrect, even if I find it to be interesting. It is matter beyond practical chances for me. As a consequence, there are many openings that I have studied that I never intend to play, even if it is unlikely that I would ever face them with the opposite color. I imagine a lot of people would think that I am wasting my time. I see it as a form of training, but perhaps it could be argued that it's not the most efficient use of time.
In regards to TN's claim that White is just as likely to make a miscalculation as Black, I would tend to agree in general terms if the positions were "cold" to both players, but that is often not the case and I think there may be a method to Gambit's madness.