On feedback. Particularly feedback on {feedback, donation, annotation} and similar meta-tools, which define and characterize participation and membership and engagement.
- We need a more nuanced spectrum of rollout scope: ways to make things visible to all testers / all newbies / all logged-in users / N% of readers.
- We need a standard model for presenting / discussing / analyzing / reviewing features. Both those that are definitely going to be rolled out on a known schedule, and those that are being developed in community collaboration and will be rolled out to those who want them.
- We need a standard set of options for trialing something. "A 1-day 1% trial" or "A 2-wk 10% logged-out trial" should be possible to specify in a discussion, and reliably implement. Without worrying that someone would forget to close the trial, or to share the resulting data, or to carry out a follow-up recap <fixed timeframe> afterwards.
- We need a constellation of community test / feedback groups. Does this exist in persistent form? It's not ideal for tests to be run by "whoever is agitated about / excited by" a feature: that leads to always having highly polarized debates between lovers, haters, and coders. And it's helpful to have groups that care about specific topics; not just i18n, and browser test groups, but also design & usability, multimedia, and workflow/workload groups.
On AFT
editFrom the AFT5 RFC, though many of these comments apply to all of AFT, or all AFT-style "any reader can post" comments.
- Some don't want 'any reader' to post comments easily. That level of input is reminiscent of early sandbox inputs -- though the vast majority self-corrected then.
- Some are mainly concerned about scale of review work. Tied to the % of rollout, amt of work needed to post, and whether an acct or other info is requested.
- Many want better granular data. acct-creation conversion, data from each of en/de/fr, better table of summaries of data from the past month.
Proposed features or options
- opt-in feature per-article. [or feature for certain types/categories/popularities of article] only implement where there is at least one editor willing to monitor the feedback [sadly, the same can be said of talkpages]
- post to / integrate with talk pages [to work with current review/curation tools, avoid splitting the stream of comments, show commenters where they can look for replies and updates]
- combine with streams from feedbackdashboard and talk-page.
- Minimally intrusive design on main article page
- [anything on "Feedback from my watched pages"?]
- Use smaller sampling to allow a broad/faster sampling of opinion
- Alternatives.
- Make talkpages more reader-friendly. New comments at top.
- Eval
- Do an impartial evaluation of whether the tool is worth its cost. banal contents, random order. Make watchlist-visible
- Hypothetical concerns
- Don't have comments show up at bottom
- Negative concerns
- No demonstrably positive results
- Weed out useless features and complexity.
- Positive concerns
- Might be useful on some wikis [or subsets?] but not on enwp
- Feedback box is disruptively large
- [active editors] find the nonsense, or even normal requests that are against our policy (for (c) or other reasons) actively discouraging. [sadly applies to talk pages]
- Wikipedia already has too many features that provide little value to readers but consume editors' time (like category tags, sorting tags, stub tags, editorial tags, navboxes, wikiprojects, article evaluation, citation templates, ...).