This page is not a place to debate the general philosophical merits or demerits of a personal image filter. This page takes as a given that we ought to do something to help users customize their own experience of Wikipedia with respect to showing images that are not safe for work (NSFW). What I want this page to be, at the end of a week of work, is a clear statement of my own view (clarified and helped by others) about the right way forward.
This is not the time to make the perfect the enemy of the good. I want a simple reasonable design that will be inexpensive to code. (Preferably, a PHP/MediaWiki hacker who finds this compelling will find it easy to implement so that it can be presented to the Foundation as something they just have to turn on.)
Here are some general principles that I think we should assume at the outset. I'll try to refine this over the next few days (with your help) because I'm sure on a first pass there will be things that I overlook.
- The feature will be developed and made available, but will be turned on only upon the request (through a suitable voting process) of individual wiki communities. Why? I want to reduce the possibility that different cultural norms (either in a wider culture or just the internal culture of one Wikipedia language) in some parts of the world will block progress in other parts of the world.
- However, a design goal is that the feature will be so uncontroversial that it will be easily adopted by virtually all language communities.
- To gain widespread support, the feature must be easy to turn on, and easy to turn off. Notice that easy to turn on and easy to turn off also lowers the cost of categorization errors, and lowers the drama around how to categorize. If I'm using the feature and something is hidden which I want to see, it should be one click for me to see it, and only one more click to turn the feature off completely. If I'm not using the feature, and I see something I don't want to see, it should be one click to turn it on. The desires of the reader are paramount.
- [Moved to talk page for discussion; was "It is okay to have a design that commits the Foundation to ongoing editorial support...."]
- ALL complications should be left to a later date. This is supposed to be a useful step forward, not the ultimate solution to all imaginable problems and edge cases. I just want something that works reasonably well. Only after we gather empirical evidence will we even really know what the issues are.