tl;dr: all you handle kids are scaring the straights away, and it’s a problem for us.

When Google+ launched this summer, it required users to register under their “real name”  — not necessarily their legal name, but the vaguely defined “name your friends, family or co-workers usually call you.” A lot of people thought this was a bad idea, including me. Real names harm a variety of people in different situations, while psuedonyms are an important tool of privacy in a medium where every public utterance is recorded and forever searchable. Although the issue of online identity is far broader than Google+, the launch of this new service was seen by many as a chance to reexamine this point, and the policy triggered a public backlash which came to be known as the nymwars.

Last week, Google+ rolled out a partial reversal of this policy, allowing arbitrary names, but only for new users, and only if they are an already “established identity.” Again, this seems an impossibly vague standard. Also, why? Why can’t I just call myself whatever I want?

Certain answers come out in  Google+ chief architect Yonatan Zunger’s recent thread on the topic. It seems that, rightly or wrongly, Google has certain strong ideas about what “kind of community” Google+ is supposed to be. Moreover, they claim that only a minority of people have strong feelings about the use of pseudonyms online, and that they have data showing that the use of “handles” drives other people away.

In this sense the use of real names is at heart a business issue, just as many folks suspected. But not the business issues that have been most talked about. One standard argument is that Google wants your real name for the benefit of advertisers, or for the benefit of state authorities. Of course Zunger and others could be deceiving us (or themselves), and I certainly believe that Google is engaged in a competitive deathmatch to be the dominant online identity provider. But the “advertisers” and “authorities” arguments for real names seem to me weaker than they first appear. After all, it’s not your name that drives personalized advertising algorithms, it’s the content you produce and where you are in the social network. On the commerce side, no one needs my real name to take my money, because payment systems are ultimately tied to credit cards, bank accounts, or phones. These sorts of social and financial links also make real names much less interesting/useful from a law enforcement point of view, especially given that Google will already turn over all your information to government authorities when asked. In either case, names are really only useful as a (very unreliable) key to match between multiple databases. Your behavior is much more telling. For an illustrative example, consider that it’s possible to accurately guess your age, gender, and political orientation from public Twitter data.

So let’s look closely at what Zunger said about the recent names policy change, in response to detailed questions. This is his stated reason for the original real names policy:

First of all, you might ask why we have a names policy at all. (i.e., why we don’t simply go with the JWZ proposal) One thing which we have discovered, while putting some miles on the system, is that it is indeed important to have a name-based service rather than a handle-based service. This isn’t a matter of functionality so much as of community: You get a different kind of community when people are known as Mary Smith than when they are known as captaincrunch42, and for a social product in particular we decided that the first kind of community is the one we want to build. In order to do that, we want to establish a general norm that the names you put in to the system should be names, not handles.

Zunger is talking here both about what “kind of community” Google+ is intended to be, and how he thinks that sort of community can be established — by making rules in an attempt to encourage certain norms. He distinguishes between “names” and “handles.” I’m not really sure I immediately know how to tell these apart. Further adding to the confusion, Zunger is also clear that Google+ is (now) not concerned with whether you are psuedonymous or not:

Our name check is therefore looking, not for things that don’t look like “your” name, but for things which don’t look like names, period. In fact, we do not give a damn whether the name posted is “your” name or not: we will not challenge you on this basis, nor is there any mechanism for other users to cause you to be challenged for this.

In regards to a question about “anonymity,” he says

it depends on what you mean by “anonymity.” If you mean that the name on your account isn’t associated with you in meatspace, I think that we support that right now.

Ok, so why bother restricting names at all? “You claim evidence that a no-handles policy is better for discourse,” wrote Sai. “I’ve seen zero proof of this, and indeed proof to the contrary.” Zunger responded that the policy is

not a no-handles policy, but a rare-handles policy. I don’t have data which I’m at liberty to share, but we got very strong feedback about this one, especially from less technical users, and also very disproportionately across genders: women liked handles a lot less than men. (This is somewhat reflected in the populations which have the highest density of handles: e.g., people who are old-time Internet users and whose handles date back to usernames)

Yet Zunger also admits that real names don’t constrain bad behavior the way he was hoping they would:

We thought this was going to be a huge deal: that people would behave very differently when they were and weren’t going by their real names. After watching the system for a while, we realized that this was not, in fact, the case. (And in particular, bastards are still bastards under their own names.) We’re focusing right now on identifying bad behaviors themselves, rather than on using names as a proxy for behavior.

What’s going on here? Zunger says he both “got very strong feedback” in favor of real names and “bastards are still bastards under their own names.” How can both these things be true? He explains a little farther down the thread:

Actually, it’s not that people think that nyms are abusive at all. It’s that people react differently to seeing that they’ve been circled by John Smith, versus seeing that they’ve been circled by CaptainCrunch49. Various categories of user tend to react very negatively to the latter, say something to the effect of “who are these strange people?!,” and log off and never come back.

…

The initial policy was different, and it was based on a number of reasons, such as the theory that permanent names encouraged good behavior (turned out not to be true) and the theory that name-based services have a different ambiance, and lead to different collective behavior, than handle-based services. (Seems to be true)

…

It’s definitely an issue of perception, not security. But handles are only used in a fairly limited subculture, and a lot of the past intersections of that subculture with the broader culture have been negative: people associate handles with trolls on forums. … Obviously, not everyone with a handle is a bad actor, but handle namespaces have acquired this rep in spades.

When pressed on whether real names are “more engaging and encourage interaction”, he says

This does, in fact, seem to be the case — people seem to interact really differently when they see names and when they see handles. This is one of the main reasons why we continue to think that this distinction is worth preserving.

So Zunger is claiming that the goal of excluding “handles” is based on user behavior differences that can be seen in the data — not “bad behavior” but other things, the only one of which he’s specified is leaving and never coming back. This is a core business issue. But it’s also a user experience question. Zunger refers twice to negative experiences of women on G+, and this is consistent with what I’ve heard from my female friends who complain of “creepy” people adding them to circles. Even when not creepy, “who are all these people adding me?” has been a common refrain with G+. This is a problem which is made worse if people choose psuedonyms which they don’t aren’t already commonly use elsewhere — how do I know who you are when you add me to your circles? Of course, there are psuedonym-preserving potential answers to this question, such as seeing that a known friend vouched for them.

Part of the reason there is such heated argument about the use of psuedonyms is because there’s so little data. The best large scale evidence I know of is Disqus’s figures, which lead the company to conclude that “pseudonyms are the most valuable contributors to communities” in terms of comment threads. Zunger isn’t releasing any data, but he drops many hints about the content of Google’s data set, which contains much richer information than comments. It appears that Google has done some social network analysis on psuedonym use:

There’s a lot of clustering asymmetry in this, however. Generally, if you know at least one person who has an unusual name, you’re likely to know a lot of such people; i.e., people with unusual names travel in tightly-connected clusters. That’s largely because these names tend to be tied to particular subcultures. The problem we’re really encountering here is of culture clashes: people from one culture absolutely freak out when they encounter people from a very alien culture. That’s actually a very deep problem which affects a lot more than names, and it’s one that I’m spending a lot of skull sweat on lately. (I can tell you more off-line) If we can find a good way to deal with that, then the handles problem goes away too, and we can just revert to the simple jwz solution.

And so we get to the current confused state of affairs: you don’t have to sign up under your “real” name (whatever that is), but you have to meet some vaguely defined standard for “established” names. The new criteria are spelled out in the post by VP Bradley Horowitz:

If we flag the name you intend to use, you can provide us with information to help confirm your established identity. This might include:

– References to an established identity offline in print media, news articles, etc
– Scanned official documentation, such as a driver’s license
– Proof of an established identity online with a meaningful following

As a matter of practice, Zunger explains that G+ uses machine classification to decide whether a name is allowed, augmented by humans in the uncertain cases:

The classifier is training to get the (huge number of) easy cases right, not the hard ones; those are always going to be passed off to actual humans. … The goal is that most things which are marked as “not a name” are genuinely cases of something being meant as either a nickname or an organization; whenever the appeals process is triggered, and even more so whenever something passes an appeal, that’s a sign that the first-stage check failed and we need to improve our rules. So then we can look at the pattern of appeals, see if there are classes of names which we are systematically getting wrong, and learn from this to improve the process and reduce the chance of someone being sent through it incorrectly.

But Zunger hasn’t yet really answered the question of what qualifies as a legitimate name at signup — what will pass the human review process that is used to train the machine classifiers? And while there is a new “nickname” feature available to people who have already created an account, you can’t only be known by your nickname, which shows up in addition to your “real” name.

So where does this leave us? On G+ you have to use some name you commonly use elsewhere. The policies are ambiguous but strongly favor the sorts of names most people use to introduce themselves in person, rather than “handles.” You can use a handle on G+, but only if you can convince a human that you have been widely calling yourself that elsewhere. In other words, we still can’t call ourselves whatever we want, because some number of people are going to fail the (unspecified) name check. You can’t create a brand new identity in order to explore, say, an openly gay existence online, or to see how people would treat you if they didn’t know you were a kid (which is something about the internet that was really important to me when I was 14.)  The weird thing here is, overall the policy doesn’t sound like it’s really about whether your name is “real,” but whether it sounds like “not a handle” in a way that doesn’t frighten other people away.

So what’s so bad about this? Identity is complicated in real life. We are all different people in different contexts, and expressing yourself online has the risk of smushing all those contexts together in a way that loses something, such as the ability to reveal yourself fully without fear. Google understands that such problems exist, or they wouldn’t have done the sociology research that clearly influenced the “circles” feature. But as Moot has argued, maybe it’s only under anonymous conditions that we are authentic. The current name policy is not going to work for many people in many cases. Zunger knows this:

I completely agree that “well, shut them out then” is not the right thing to do. But I’m currently stuck between shutting out a small number of people, or creating an environment in which a large number of people (especially women, and especially people who are already feeling uncertain in the online environment) feel a hostile environment and get shut out, too. I do not like being in this situation and am actively trying to work on real solutions which will allow us to bridge the gap, and make this a good environment for everyone.

Which is a lovely sentiment, but how is this to happen?  Zunger freely admits that the recent changes are supposed to work for most people, most of the time. In this sense the problem is more that G+ is aimed at the mainstream than that it specifically excludes you. It sucks to be different, and G+ is not trying to solve this.

And there’s one of the crucial questions in the whole nymwars debate: are we arguing about rights, or are we arguing about what serves “most” people? I think you could make a good argument that the ability to choose one’s name online is a right, just as it is a right offline. Where the waters get muddy is that Google+ and other huge networks are private property, and can set more or less any rules they want. Yet we all use them; we depend on them for public interaction. Rebecca MacKinnon explores this problem at book length in Consent of the Networked.

Meanwhile, Zunger seems intent on preserving the names/handles distinction:

I’m making a tradeoff in this service by restricting the space of names to things which are, by some criterion, “name-shaped.” On the one hand, the exclusion of handles has a nontrivial cultural effect, because handle-based cultures such as Internet fora, YouTube, some parts of fandom, etc., have established cultural norms which are (on the very large-scale average) ultimately somewhat similar to one another and very different from those in many name-based cultures, such as G+, FB, or meatspace. Since we have made an explicit decision to make G+ a name-based culture, and since the large bulk of our users come exclusively from such cultures (i.e., have little or no familiarity with handle-based cultures), there are significant culture clash risks associated with culture mixing and we’ve chosen to resolve those by basically excluding handles. (With rare exceptions for very established handles, which is an exception people are used to because they see those cases as intrinsically exceptional; as an extreme example, Lady Gaga) On the other hand, this excludes identities which come from handle-based cultures.

…

When the excluded identity is in the second category, then this is frankly working as intended: I’m trading off one virtue of social health (building up a unified culture on G+) against another virtue of social health (allowing as many identities as possible to be represented on the service).

…

The resolution that we’re aiming for amounts to attempting to structure the name restrictions as narrowly as possible in order to attain the social health virtue of building up a name-based culture.

Zunger hasn’t yet explained why a “name-based” culture is a “social health virtue.” He already said that “name-based” cultures don’t control “bad behavior” and “bastards.” But we do have one really important clue: he claims that only a small number of “subcultures” use handles, and that handles drive away people who aren’t used to such cultures. If Zunger is telling the whole story, G+’s critics are right in that this is a business issue, but it isn’t so much that advertisers want real names. Instead, this names policy seems solidly about being acceptable to as many potential users as possible. Which is not such a terrible goal, but it’s by definition anti-subcultural, and that does kill some of the genuine and important ways that people enjoy interacting with each other .