What do you think?
If you mean real spam, like those trying to sell you pills or scamming / phishing information out of you, bouncing the email won't have any effect at all. The messages are usually sent from botnets (ie: networks of hundreds of infected computers controlled by the spammer), the reply addresses are faked, etc...
If, on the other hand, you mean regular websites that have acquired a database of harvested email adresses and are spamming you with their newsletter, then either they care for the quality of the base and they'll usually provide you with an unsubscribe link, or they don't and there's no point in bouncing back the mail. It will just be ignored. Best thing to do against them still is legal action, since they usually cannot / will not hide their real identity, like professional spammers do.
Conclusion: automatic spam classification is still the way to go, and I must say that gmail is good enough at it that I'm rarely bothered by spam in my inbox.
- Algorithmic filtering is an arms race, I still get SPAM in gmail almost every day (albeit just one or two).
- Legal action is out of reach for most people (probably everybody who is outside of the US/EU).
- Just flaggin 'em as spam has lost the "HA - there you go: I just FLAGGED YOU AS SPAM!!!" feeling over the years, now it is more like "Duh hum another spammy email. Flagged it again".
Gmail would get blocked as a spammer, because...
sometimes the sender is spoofed
...so they would be sending spam to the spoofed sender, who doesn't want to read the spam either.