In my experience talking to authors, this definitely has a pretty large effect on what people write on the site. Though I think the effect is often not really dominated by worrying about the perception of other LessWrongers, but about the perception of other random people on the internet who might take what you say out of context. Or potentially future organizations you might be collaborating with that might comb through everything you said on the internet in search of red flags and a highly risk-averse hiring or collaboration policy.
At least within the rationality community, I view commenting almost always as a positive signal about someone, and think this is pretty common. My best guess is that the reputation management of this type doesn’t actually have super much of an effect on other people in the community, though it does have an effect on the broader world, and it’s also hard to be confident here.
LW developers create an official and convenient anonymizing option, where posters can choose not to have their profile name attached to a comment/post but can reveal it later if they choose.
I’ve been thinking about doing that for a while. It’s a good amount of pain from an implementation perspective, if you want to avoid allowing indirect methods of deanonymization (like voting on an anonymous comment and checking which users total karma changed, which you can either do programmatically for hundreds of users, or manually for a few users if you have a suspicion about what user it is).
Thanks for your thoughts on this. Would a soft anonymization feature work, just hiding their name and warning them that there are still ways a determined person could find them out? Or perhaps disabling the link between post/comment karma and total user karma?
I think the fact that you can create pseudonymous accounts already functions as a superior anonymizing option without cluttering up the UI, complicating the backend or permitting the attack habryka outlines.
Secondary accounts provide better security because they remove a broader attack vector. If you trust Less Wrong to anonymize you then you are vulnerable to Less Wrong getting hacked. If you manage two separate accounts then a compromise of the Less Wrong servers will not deanonymize you.
Yep, though UI affordances are surprisingly powerful and people seem to forget that this is an option pretty frequently. I do think the fact that it’s super easy to create a new account is a reason why we haven’t implemented this.
Sometimes people are also worried that they shouldn’t use a second account because that’s a bit like sockpupetting. There might be some way of making it clear to people that we don’t mind, as long as you don’t upvote your own content.
In my experience talking to authors, this definitely has a pretty large effect on what people write on the site. Though I think the effect is often not really dominated by worrying about the perception of other LessWrongers, but about the perception of other random people on the internet who might take what you say out of context. Or potentially future organizations you might be collaborating with that might comb through everything you said on the internet in search of red flags and a highly risk-averse hiring or collaboration policy.
At least within the rationality community, I view commenting almost always as a positive signal about someone, and think this is pretty common. My best guess is that the reputation management of this type doesn’t actually have super much of an effect on other people in the community, though it does have an effect on the broader world, and it’s also hard to be confident here.
I’ve been thinking about doing that for a while. It’s a good amount of pain from an implementation perspective, if you want to avoid allowing indirect methods of deanonymization (like voting on an anonymous comment and checking which users total karma changed, which you can either do programmatically for hundreds of users, or manually for a few users if you have a suspicion about what user it is).
Thanks for your thoughts on this. Would a soft anonymization feature work, just hiding their name and warning them that there are still ways a determined person could find them out? Or perhaps disabling the link between post/comment karma and total user karma?
I think the fact that you can create pseudonymous accounts already functions as a superior anonymizing option without cluttering up the UI, complicating the backend or permitting the attack habryka outlines.
Secondary accounts provide better security because they remove a broader attack vector. If you trust Less Wrong to anonymize you then you are vulnerable to Less Wrong getting hacked. If you manage two separate accounts then a compromise of the Less Wrong servers will not deanonymize you.
Yep, though UI affordances are surprisingly powerful and people seem to forget that this is an option pretty frequently. I do think the fact that it’s super easy to create a new account is a reason why we haven’t implemented this.
Sometimes people are also worried that they shouldn’t use a second account because that’s a bit like sockpupetting. There might be some way of making it clear to people that we don’t mind, as long as you don’t upvote your own content.