Plus, can you imagine the compound interest? :)
Let's take binge-drinking, for example. Let's suppose that each binge-drinking episode has a .0001% chance of causing a fatality. That is trivially low.
However, in order to consider the trade-offs involved, you have to think about the expected value of your life in either scenario. So you must compare E(life | no drink) with E(life | drink).
These are practically the same, but where they differ is that in the drink scenario, your expected value is broken down into two scenarios -- one if you die as a result of the binge drink, and one if you don't. So you can break E(life | drink) into
99.999(E(life + d)) + .001d, where d is the added enjoyment you get from that one night of binge drinking.
The comparison, then, is ultimately whether E(life) >=< 99.999(E(life + d) + .001d
You can tell from this that what really matters is the ratio of d to E(life), the enjoyment you get from drinking that one night, compared to your overall enjoyment of your ENTIRE subsequent life. As we extend your life expectancy, as long as we assume that life is a good thing, we make the value of your subsequent life higher and higher. This, in turn, should make any rational actor be less willing to take on risks that would cut it short.
You're putting too much faith in the judgment of centuries-old "teenagers". My guess is that they would squeeze 100-200 years of drinking and partying into 5-10 years :)
Also, how would your memory hold up? Would you even be able to really remember anything that happened over a century or two ago?
There no pressure to get your stuff together if you know you can mess around for a couple of centuries and still have time to get your life in order.