Trust level 3 means regulars- change the criteria?

Continuing the conversation about improvements to Community, or rather, evolving it from a home for the construction team to operations, data, and the science community, let’s talk about trust levels.

Trust level recap-

  1. Basic- visit once, spend 10 minus, read 30 posts in 5 topics.
  2. Member- visit 15 days, spend 60m, read 100 posts in 20 topics, reply to 2 topics.
  3. Regular- over the last 100 days, read 25% of all posts, in 25% of all topics, & reply to 10 topics.
  4. Leader- placed there by staff.

Member status/trust level 3 is not sticky; one can lose it by not being around for the last 100 days. Unless one is placed there by fiat, I assume.

The question I have is about our user statistics:
trust level/ count
0 / 285
1 / 207
2 / 48
3 / 300
4 / 136
What should be a power law is a dual power law, because of Rubin staffers being put into trust level 3. It looks like in the first power law we’d have ~10 regulars. These would be outside the staffers. I wonder if they exist? Is there a way to tell?

The thing is likely the meaning of trust level 3/regular. The high wall seems to be over the last 100 days read 25% of all posts, in 25% of all topics. A concrete proposal would be to lower the percentage numbers, the idea of growing community regulars.



The forum has many groups. When someone is onboarded into a Science Collaboration by the SC’s chair, or a Rubin staff member is onboarded into a team, the onboarding procedure is to add the person to the corresponding group. Membership in those groups grants an elevated trust level (because we “know” the person). That trust level is generally 3, as you found. I’d love for this to be automatically linked to the Rubin LDAP, but we haven’t tackled that yet. So at the moment, we’re relying on team managers and SC chairs to do this work, but I acknowledge it isn’t being perfectly followed.

I guess my question is whether there’s a concrete need for more TL3 “regulars”?

  • TL3 grants the ability to create your own wiki page and to mark a TL0’s post as spam and have it automatically become hidden (in effect, with TL3 you are granted some moderation abilities). Other than that, TL3 doesn’t give you extra abilities.
  • TL2 members see notices about new users or returning users (i.e., they have a feel for the community).
  • TL1 is where you gain the ability to essentially use the forum regularly and most of the rate limits are removed
  • TL0 is where we haven’t seen enough of the user, and so rate limiting is applied (on links, images, attachments, and so on).

Again, I’m not sure we need tweak the algorithms to specifically increase the number of TL3 users. Perhaps you could elaborate.

If anything, I think we need to focus on all the TL0 users, and see if we can find ways to promote their usage of the forum by reading and replying so that they become TL1 users.

Is there a Rubin Staff group? I don’t recall seeing it. But if there were, then I could subtract Rubin staff count from the number of trust level 3+4 people to understand if there are any organic trust level 3 people on the site.

I think my unease at the double power law is that over the longer run it should become more like a single power law. My concern is that our site isn’t a single-interest chat site, it’s more siloed. The people that come from the science collaborations are generally going to be focused on only the topics in their science range and the answers to technical questions that apply to them. That kind of user is TL2. I’m doubting that we’d ever (strong word, more than 3?) people who meet the TL3 requirements without being paid to do so.

If the CET vision of Community is of a larger community self-supporting, seeded by the Rubin staffers is to come into being, we’ll have to allow that larger community to grow here. I’m speculating the TL3 requirements will be an impediment.

I do think that getting more TL0 users and transitioning them to TL1 is important. At the moment, though, there’s equal numbers in TL0 and TL1, so the transition to TL1 is pretty effective.


Yes, the “lsst” group has about 150 members.

Again, I’m not convinced that trust level 3 will be an impediment. Aside from the moniker “member” versus “regular” there isn’t a real difference (outside moderating TL0 members, but again, there are already so many moderators on this forum I don’t think that’s an issue) as I wrote above, and I don’t think it substantially makes a difference to how people will feel about their belonging to this community.

I think the CET has an opportunity to grow this forum’s community and can do so by providing useful content and provoking interesting conversations around Rubin science, data, and technology. I think that tweaking trust levels is a tool of last resort to promote the community, and wouldn’t be something that I would personally be focused on during an initial revamp of our forum.

I am very confused as to what the problem that we are trying to solve here is aside from the user trust level distribution not adhering to some pre-conceived notion of what it should be, the effect of which is… I’m not sure what.

You don’t grow community regulars by finetuning trust level parameters, you grow community regulars on a forum for Rubin data by having Rubin data for people to talk about, a situation that we are not yet in.

The trust level system defaults are set because of experience the developers have with providing Discourse as an open forum. It was certainly my vision that we too would operate an open forum, allowing anyone who wanted to engage with our project - amateur astronomers, schoolchildren, software developers and of course the science collaborations would be able to use it to communicate, as well as retaining it as a valuable in-project communication tool. You are not going to get a single distribution of users for all those groups. Some will consume the forum unregistered, some will lurk, others will be frequent posters. That’s okay! I think being concerned about metrics before (a) the largest userbase is even onboard even and (b) without specific user population surveys to understand in what way the various userbases interact with the service is putting the cart before the horse.

My personal advice having run digital spaces in a variety of contexts is that the primary attractor to any forum is the quality of the content creation, the rate at which new content is available, and the moderation of bad actors. If you have that, the users will follow.

1 Like

Dreaming about the future- in a few months we’ll have 300-odd motivated people playing with the DP0 data on the cloud science platform. Watching how that bump of people moves through the trust levels will be interesting.

I imagine the groups LSST + LSSTDM add together to reach the 430 people in TL 3+4? Or is there really a significant contribution from the Science Collaborations?

Looking at (only visible to admins) I see a lot of people who do not appear to be in LSST or LSSTDM. I don’t know if they received their trust level manually or automatically. But again, there’s no huge benefit to being TL3 vs. TL2.

Those folks are (generally) members of science collaborations.

Jim - I agree that it will be interesting to observe what 300-odd motivated people will do with all our services - that, of course, is the actual point of getting 300-odd motivated people :slight_smile: Let’s take advantage of having that userbase and survey them about their experience so we can make well informed adjustments after having data to serve as a control. Right now if you make any service changes, you won’t have a real baseline to compare it with.

It’s also worth remembering that even our full complement of users is nothing to what eg. udacity are doing with this platform, though, so we can also look to the discourse community for best practices in community management.

1 Like