[Foundation-l] Experiment: Blurring all images on Wikipedia

[Phil Nash]

Bjoern Hoehrmann wrote:
> Hi,
>
>   A while ago I made a bookmarklet that blurs images in articles on
> the english Wikipedia and reveals them when the user hovers over the
> image. I now had a chance to test this as a skin.js extension.

For a start, users would have to opt in to this, which may not be 
appropriate for casual readers brought to us from Google and other external 
links. I'm not sure it's a good idea to make it a default for unregistered 
users, many, if not most, of whom, might not want to be presented with a 
pre-filtered version of Wikipedia, and would be surprised to be so 
presented. It also presents a "slippery slope" argument in that nobody is 
realistically qualified, nor would want to be tasked with, drawing the line 
as to what images should and should not be treated thus. A similar argument 
applies to textual content of articles; however we try to achieve 
neutrality, it seems that there will always be some POV-pushers who will 
argue the toss ad infinitum, and we don't accommodate them. Neither should 
we accommodate those who do not understand that a value-neutral, world 
value, is not the same as their value. These people have their own texts, 
and I think that our response should be that they are welcome to them. 
Nobody is being forced to use Wikipedia, after all.

>   http://en.wikipedia.org/wiki/User:BlurredImages/vector.js
>   http://en.wikipedia.org/wiki/User:BlurredImages/vector.css
>
> To try this out you would have to copy or import this code into your
> own skin.js and skin.css files which are available e.g. under
>
>   http://en.wikipedia.org/wiki/Special:MyPage/vector.js
>   http://en.wikipedia.org/wiki/Special:MyPage/vector.css
>
> This only works in recent desktop versions of Opera and Firefox and
> only on devices where you can easily hover. It may show some images
> that it ought to blur for boring reasons. Spoilers ahead if you want
> to try it.
>
> Browsing around with that is quite interesting. Some findings: it is a
> bit annoying when UI elements (say clipart in maintenance templates)
> are blurred. The same goes for small logo-like graphics, say actual
> logos, flags, coat of arms, and actual text, like rotated table
> headers. I did expect that blurred maps would be annoying, but I've
> not found them to be. Take http://en.wikipedia.org/wiki/Dagebüll as
> example, the marker and text are overlayed so they are not blurred,
> and I can recognize the shape of Germany fine.
>
> I note that there is a perceptual problem if you click around to
> explore how blurring affects the experience as that does not reflect
> what a user would do. I noticed that my impression changed a lot when
> switching from actually paying some attention to the articles to
> randomly moving to the next article just looking at the images.
>
> Pages, or parts of pages, that largely lack content (say all you get
> on a screen is lone line of lead, table of contents, and image plus
> map on the side, or a stub that has four sentences and an image).
> There it's a bit odd, in stark contrast to an article like BDSM where
> I felt blurring is very unobtrusive.
>
> Another thing I've noticed is that I pay a whole lot more attention to
> the images when I focus them, decide to hover over it, reveal it, and
> then look at it, maybe read the caption and so on. I also noticed I do
> not really bother to read the captions before I hover and rather
> decide based on the blurred picture itself (I ignore most captions
> usually, so this is unsurprising). There are also many surprises,
> where images do not come out in the clear as you would have expected
> from the blur.
>
> My impression is that it actually makes it much easier to think about
> if an image is well placed where it is. If there are several images,
> you can focus more easily on just the one, and you remove to some
> degree the "status quo" effect, where you may be biased to agree with
> the placement because someone already placed it there.
>
> Images where red tones are used a lot seem to be rather distracting
> when they are blurred. Blue and green and yellow and black and white
> and so on are no problem, and the red tones are no problem when the
> image is crisp. Not sure what's up with that, I have not noticed this
> before. It would of course be possible to manipulate the colours in
> addition to the blurring.
>
> Largely black and white bar charts and tree diagrams and illustrations
> of data like them are also annoying when blurred, in part because
> there is inconsistency as some of them are not blurred because they
> are made not as image but using HTML constructs. They are perhaps too
> much like text so unlike a photo with many different colors they are
> harder to just ignore using one's banner blindness skills. There is
> also a noise factor to this,
> http://en.wikipedia.org/wiki/Electromagnetic_induction for instance
> compared to http://en.wikipedia.org/wiki/Morse_code -- in the former
> the graphic in the infobox is fine blurred while the latter irks me
> when blurred.
>
> Generally though the added nuisance is hardly worth mentioning, it
> works surprisingly well (well, this was the first thing I thought
> about when I learned of the image filter, but it does work a bit
> better than I had expected initially, and some issues would be easily
> fixed, like blurring only images larger than 50x50 would take care of
> most of the UI graphics for instance). So having conducted this
> experiment, I think the need to have some images hidden while having
> others in the clear, where the com- munity as a whole decided to show
> rather than hide, as in omitting them for all users, is not a
> legitimate need.
>
> regards,