I've also been thinking about how to limit the reach of "fringe" requests, but have reservations about the methods. Here's my current thinking. First, I'm not counting requests which are against the rules. Just the fringe requests which remain.
The "problem" isn't that I don't like the request, it's the potential harms:
1. Diminished credibility of our
service. This impedes our objectives by: (i) Discouraging
requesters who don't want to be associated with such users. (ii)
Discouraging volunteers. It certainly demotivates me sometimes. (iii) Discouraging sponsors. If the first requests they see are quite fringe, they may think twice. I know this has happened at least once.
2.
Wasted public agency time. But I've talked to a few FOI officers here
and they have told me that requests of this nature from our platform are
a drop in the bucket. They've told me they think the requests will get
sent to the agency one way or another. That makes me feel less bad about
it, so long as we're not encouraging more.
I've steered clear of "backpage" prominence at all because:
- Proportionality: It's quite severe, disappearing even from the user's profile page.
- Fairness: If a user asked why, it's difficult to give a satisfying answer why a request should be buried this way if it didn't break moderation rules. The user should know the criteria in advance. And if it did break a rule, it's not suitable for publishing so why isn't it requester-only?
Inventing lesser measures of "deboosting" (e.g. just using noindex to keep it out of Google) seem quite underhanded. If we're taking manual action on specific requests or users, we should be transparent about it and be able to explain our decisions.
So instead I'm now thinking about how to "surface the good requests" instead. Reorganise lists and search results so that exemplar of good requests are seen first. Somehow work out a way to calculate or crowdsource a quality metric to sort by.
This approach would also help with one of my other goals I've not done enough work on: guiding users to write better quality requests.
Oliver