Personally, I risk no more than 2% on my open trades. This isn't a Sniper rule per se, but something people generally agree upon as sound money management, regardless of the system used. If, rather than keeping the % risk constant, you decide to keep the $/pip constant, then in a highly volatile market, you account might be wiped out very quickly. The larger the stop loss is in this case, the larger the risk, since risk = stoploss x $/pip.
One thing you lose by setting a constant risk % is a definite relationship between pips gained and $ gained, as said in Salvador's post, so that at the end of each month, you might see yourself making pips, but not money (which, by the way, is a clear demonstration that pips are not units of currency).
However, in the long run, I would guess, without proof, that a gain in pips probably corresponds to a profit in dollars. I say this because Sniper seems to work for stoplosses of any size (within a reasonable range), which means that for any $/pip value, the system seems to be profitable.
I think it's interesting to look for potential market conditions in which a system doesn't work. If a system works only when the market is highly volatile, then the losses encountered in low-volatility periods (high $/pip) may outweigh the profits from high-volatility periods. However, that's just a hypothetical situation. Just like if GBPUSD stopped trending, then Sniper would no longer be profitable. Whether these scenarios seem plausible or not is for each of us to judge based on our own intuitions.
But it does amaze me that amidst all the randomness of the market, there seems to be sufficient regularity in it that makes it possible to make a profit using a system of rules like Sniper.
H