Basically, online it's trivial to masquerade as someone else (see IdentityTheft) because you control how you represent yourself. Since the only information people have of your identity comes from what you project online, they cannot verify you are who you say you are.
Indeed, not only can you masquerade as someone else, you can even be something entirely new. This is the idea behind online "handles" or "nicknames" (nicks). It's common to find people with nonsense names like "Commander Cool" or "Frozen Burrito" online.
Entire online careers and identities can be attributed to mere handles, the equivalent of anonymous post office boxes in the real world. If you've read EndersGame, you'll recognize the potential power: Peter and Valentine, both children, masquerade as Locke and Demosthenes and become world powers.
That being said, online identities have many problems. First, they lend themselves immediately to AnonymousIdentity. Second, it's hard to pin responsibility onto people who can shed their skin faster than a snake (cf RightToVanish). Responsibility is the first requirement to any conscious collaborative effort. (It isn't necessary to IncidentalCollaborations).
However, there really isn't much of a way to verifying that people are who they say they are without going back into the real world. This is feasible for very small collaborative efforts--say inside your office or on the OrgPatterns? site--but it can't scale to larger, distributed efforts. Besides, I think it's more interesting to build collaborations between strangers. -- SunirShah
Currently we don't have much of a security infrastructure in place. It's pretty easy to imagine online identities becoming virtually unforgeable through use of DigitalSignatures. Then it will no longer be "trivial to masquerade as someone else".
More infrastructure could make "going back to the real world" easier (or unnecessary, depending on how you look at it). For example, an online system could require users to buy insurance against losing libel cases. The insurance would be associated with a digital signature and could be paid for anonymously (eg through Swiss bank accounts). Then if (eg) DaveHarris libels you online, you can sue, win, be awarded and paid damages, with the money coming from the DaveHarris account - without you ever finding out who DaveHarris is.
The "right to be sued" is an important part of making communities scale. Say I offer to write some code for you. If you don't know who I am, you don't know if I can deliver. The right to be sued enables me to make a convincing promise anyway. You can know you'll get the code or you'll get compensation, and on that basis the collaboration can proceed with confidence despite us being strangers. -- DaveHarris
An important feature of OnlineIdentity is that it doesn't map one-to-one to individual people. One person may use more than one OnlineIdentity, and a single OnlineIdentity can be shared by a group of people.
In practice, it is extremely difficult to completely divorce two identities, even if one desires to do so. No WalledIdentity is perfect.
See also: IteratedPrisonersDilemma, WhatIsMultiplicity, UseRealNamesDiscussion.
As soon as it comes to real world transactions in the sense of WikiNomics, where money is involved, this terminates the masquerade. Take e.g. the transfer of domain-names in SocialDomaining communities from one registrar account to another. Although the members of http://NamePros.com communicate in their forums via their OnlineIdentity, this masquerade normally transforms very fast into name, address data via Google:Whois and confirmed Bank user account data. See also http://Aboutus.org/RealNames. -- FridemarPache