1/16/2018 Update - Parent Comments, and Nearterm Horizon

post by Raemon · 2018-01-16T23:59:01.367Z · LW · GW · 6 comments


  Easy vs Hard Bugs
  Nearterm Horizon
  Moderation tools

One issue I've ran into on the frontpage is seeing something in "recent comments", which is missing the context of it's parent comment. I got annoyed with having to click through the post just to figure out the context.

This was fairly easy to fix, so I went and added a "see parent" option for recent comments.

Easy vs Hard Bugs

I know a few people have flagged pretty substantial bugs (in particular with the editor, such as the weird way it forcibly parses markdown), as well as major features the site is still missing for parity with Old LessWrong (such as meetups). We haven't gotten around to those, so it may be frustrating to see "small quality of life" fixes like the Recent Comment thing rather than addressing those major issues.

This is mostly a factor of "easy things vs hard things." Generally each week we set some major goals for ourselves that involve a lot of thinking/planning, and we spend a few hours each day working on those. Then, in the evening, or on weekends, we work on easier things that just happen to annoy us a lot, that we're motivated to work on in our off hours.

In particular, changes to the editor require a major upfront investment of time. We do want to fix it but we'll need to set aside a couple days for it.

Nearterm Horizon

It also seemed helpful to clarify what's upcoming, with the overall plan is to reach feature parity with Old LessWrong. (Note: this is not a promise to get these things done in a certain timeframe, more of a "this is what we currently intend.")

Moderation tools

LW2.0 has made a lot of progress in consolidating discussion. But we're stilling running into an issue, where many people still prefer to have their conversations on private blogs or FB where they have more control over the commenting experience. On a private blog, there's a clearer sense of ownership and responsibility for deciding the tone of a discussion.

We are currently planning to shift LW a bit in this direction. Users with a smallish amount of karma (around 50-100) will be able to moderate comments on their personal blogposts. Users with higher karma (probably 1000) will be trusted to moderate comments on their own front page posts.

(Front page post comments will still have to conform to some basic site-wide rules, but authors will be given more leeway to delete comments that don't seem helpful)


Comments sorted by top scores.

comment by Chris_Leong · 2018-01-17T09:44:22.116Z · LW(p) · GW(p)

Firstly, I think feature parity with Less Wrong makes a lot of sense as a focus. Having two different sites is certainly less than ideal.

"Users with higher karma (probably 1000) will be trusted to moderate comments on their own front page posts." - I'm worried that this could go pretty badly. We should expect that anyone should be able to get over 1000 karma if they hang around the site long enough. Will they have the ability to actually delete comments or just to collapse them? I want to know if I'm only seeing part of a conversation. Will this ability be removed if people use it to essentially create an echo chamber?

Replies from: habryka4, SquirrelInHell, ingres
comment by habryka (habryka4) · 2018-01-17T20:33:05.672Z · LW(p) · GW(p)

The comments would be deleted (though the original author should still be able to recover them), and if someone starts deleting a ton of stuff we will reach out to them and chat with them. When I looked over the list of people who currently have high-karma, I would have been quite surprised if any of them would have abused their power much, or would have not been open to chatting about it.

comment by SquirrelInHell · 2018-01-17T10:56:16.395Z · LW(p) · GW(p)
We should expect that anyone should be able to get over 1000 karma if they hang around the site long enough.

I second this worry. Historically, karma on LW has been a very good indicator of hours of life burned on the site, and a somewhat worse indicator of other things.

Replies from: Raemon
comment by Raemon · 2018-01-18T02:16:39.907Z · LW(p) · GW(p)

We have an upcoming blogpost that goes into both the problems we're concerned about solving, and our thoughts on how to resolve them, and I'll probably hold off till then to dive into this too much. (I'll probably respond once more with quick clarifications if need be and if there's a lot more to discuss, will do so when I can dedicate a good chunk of time to it)

But it seems like there are actually 2 (3?) different issues here and I'm not sure which of them is more significant:

I. Should we become more like a network of personal blogs than a forum? This is certainly a marked change. The main reason we're considering the idea is because many people do seem to prefer discussing things in personal-blog-like spaces - sometimes because they own the space, other times because someone they trust owns the space and they have more of a sense that it's in the control of someone they trust (and people vary in what sort of people they trust and how they want discussions curated)

II. Is Karma a reasonable tool to determine trust? For any features that we might want to limit to trusted users.

Both questions are important, I wanted to make sure I didn't respond to one if the crux of the disagreement was more about the other.

Re: the Karma question (disclaimer: this is just some high level examples, not concrete plans)

If not karma, how would you determine who gets access to trusted permissions? The two main solutions I can see here is "some kind of systemized approach" and "fiat, careful decisions by the site admins." Both of them seem to have risks and issues. Systems can be gamed. Discretion of admins or existing trusted users can become insular.

(The third option of "don't ever create tools that can only be given to 'trusted' people is impractical since, at the very least, someone needs to deal with spam and trolls", and as the site grows you will need to grant that power to more people to deal with higher volume)

It's perhaps worth noting that I see karma as "the system we come up with to allocate trust, which will change over time as we think more about it", as opposed to "the system as currently implemented now."

There could be different types of karma to reflect trust in different domains (i.e. "I trust you to write things that are worth reading" vs "I trust you to care about making sure discussions are productive").

I do suspect both systemized and discretionary trust will end up playing a role - perhaps with certain permissions granted automatically by something-like-karma, and allowing trusted users to manually add newer users if they trust the new users' outside-site accomplishments and decision making.

If people have ideas for third options that seem radically different from either of those approaches, I'm interested.

comment by namespace (ingres) · 2018-01-17T20:23:12.699Z · LW(p) · GW(p)

Thirded. This entire "Karma = Trust" nonsense is a really glaring design flaw that seems to be getting propagated forward into future plans and features.

Karma measures time and investment into the site, but not trust. Conflating the two is very dangerous.

Replies from: habryka4
comment by habryka (habryka4) · 2018-01-17T20:28:38.674Z · LW(p) · GW(p)

My expectation is that it will work well-enough, though not perfectly, and that we will probably improve the karma systems in a bunch of different ways over time. StackExchange similarly gives you moderation and editing power as you reach a certain amount of karma, which you mostly get by just being active and hanging around, and their system is working reasonably well.