Nate Silver, statistics guru of the New York Times, addressed the HOF Steroids question the other day in a Q&A, and I thought his answer was insightful:
Would you vote for Barry Bonds or Roger Clemens to get into the Hall of Fame?
Yes, I think, in large part because the split-the-baby solutions to steroids use are hard to apply in practice. I might use steroids use as a tiebreaker for otherwise very close cases (and I think McGwire, Sosa and Palmeiro all fall into that category). But I don't think people should pretend that we can put each player's stats through some kind of algorithm and come up with "steroid-neutral" statistics. We just don't know all that much about who did and didn't use steroids, and when.