You use the phrase "outed", which to me implies public exposure, so my post is based mostly on that premise. If that's not the case, please just clarify and ignore the applicable parts.
In theory that sounds nice, and to a certain extent there are some similar systems in place already...but I just don't see it working that well overall.
We already have a system like this for sex offenders, and it's got some serious issues. In such a system it would either be up to each site to have a public "troll list" or they would have to have some central repository. In the case of the former, anytime you look at a sites list you won't be able to tell which ones are just someone who made an offensive comment or two on a single site and "repeat offenders" who are on many sites lists. You'd have to collate the lists somehow and even then you'd have troubles of differentiating the various "levels", mistaken identity, etc. In the case of a central repository that's a whole 'nother can of worms.
Then there's the question of what constitutes an "identity" for the purposes of being able to post. A name? An address? Those don't seem suitable to me for such purposes, they're too fudge-able and temporary. A State ID# or SSN seem too personal and permanent. If the former are compromised, you can limit the damage somewhat, the latter less so. There's the issue of proliferation. Even if you use addresses, every single time you put your info out there is the increased risk of it being compromised. It's not unreasonable to want to limit the amount of times you expose yourself to the risk of security breaches.
There's the issue of the practical burden it would place on the sites. This isn't so much of an issue since it would be voluntary, but it's something to consider.
I think the ethical issue of whether it's ok to out someone who obviously has enemies on the internet is a tricky one. Again it gets to the issue of what constitutes an "identity" for these purposes. If it's too specific it becomes dangerous to release, or even store...and if it's not specific enough it becomes dangerous due to confusion, or loses it's sting.
There's also the question of whether it will really deter anyone. As much as they like to try and evoke the image of a deliberate prankster, I don't think most trolls really think things out that much. It strikes me as impulsive behavior. So they get hit on one site after an impulsive trolling and just move to another one.
Assuming that it's specific, I think the possibility of some sort of violent retaliation needs to be considered. You could argue one reaps what one sows, but I don't think that's an attitude that belongs in the consideration of organizing a society. We should value protecting everyone over punishing wrongdoers. A balance needs to be struck between removing the cause of harm from society and not discarding anyone.
Overall I wouldn't be surprised if such a system comes into being. Some walled gardens where members are authenticated based on a one way hash of some personally identifiable information, or a cryptographic key-fob or something...and if they cause a ruckus their access will be revoked. However I think such a system would have enough problems that it wouldn't take off except in isolated pockets that highly value having an exclusive discussion among a verified group, and there I think it would be more about that than preventing trolling.
For trolls, I think measures similar to what is mostly used now will continue to be the main defense. The ability to moderate, vote, etc.