>
> A. 'Enabled' user preference
> Provide a preference checkbox with Media Viewer enabled by default (e.g.: 'Show images in Media Viewer'). To disable MV, users can uncheck this preference.
> • Pros: preferable from a UX point of view, indicates this is our recommended option, more user-friendy than JS gadget option below
> • Cons: this approach has caused problems before[..]

Really? What problems?

> B. 'Disable' user preference
> Provide a preference checkbox where Media Viewer can be disabled (e.g.: 'Disable Media Viewer'). To re-enable MV, users can uncheck that preference.
> • Pros: addresses user concerns about pre-selection, more user-friendy than JS gadget option below
> • Cons: unclear what Media Viewer is, confusing because you have to uncheck the preference to re-enable Media Viewer, adds to preference bloat issue

My understanding is that negations are harder to understand at a glance. Check pref to not do X is more confusing than check pref to do X no matter what the default state is.

I find it hard to believe users would object to having something checked by default but not object to having its meaning reversed and unchecked by default. If people are going to object its going to be to enabling the feature. Phrasing it as a double negative wouldn't change that. Otoh humans are silly sometimes.

> C. Javascript gadget or script
> Provide a site-wide gadget (or personal JS script) that would let users disable Media Viewer.
> • Pros: no preference bloat, no cache fragmentation, can simply ride on #263 and provide example JS code.
> • Cons: not user-friendly (the gadget has to be installed manually), the bootstrap script would still get loaded.

(Parser) Cache fragmentation should not be an issue either way. This sort of pref should not need to vary the parser cache. (If it was implemented in that way, i would predict strong objections in gerrit. We have only a few cache varying prefs, and my understanding is we dont want to add any more without a very very good reason).

-bawolff