[uf-discuss] rel="nsfw"

Mike Schinkel mikeschinkel at gmail.com
Sat Dec 30 15:52:05 PST 2006

Dougal Campbell
> I disagree. I think that the people who are likely to 
> produce/consume a 'nsfw' tag have a moderately similar 
> (though vague) notion of what is or isn't safe for most 
> people's work places. 

In certain countries, a picture of a topless woman would be "sfw" whereas in
others a picture of woman's uncovered face would be considered "nsfw."  It
is rather myopic and (unconsciously) arrogant to presume other's culture are
moderately similar to one's own.

> any more than the concepts of 'friend', 
> 'acquaintence', or 'spouse' in XFN have to be defined. 

Those concepts are far more cross-cultural than that for offensive material.

> Alice 
> might flag something as 'nsfw', whereas Bob might consider 
> the same content 'sfw'. That doesn't invalidate Alice's 
> personal opinion and her desire to warn others that the 
> destination link might be questionable in some way. In fact, 
> the designation might not even reflect whether or not the 
> content is 'safe' in Alice's workplace, but merely that she 
> recognizes that it might not be appropriate for *some* 
> workplaces.

You need to consider what Microformats are for. They are there to provide
for automated processing. So yes while it is fine for Alice and Bob to write
that things are "nsfw" or "sfw", or send emails to friend with a link where
they mention that it is "nsfw", but I would argue that is not the same as
using markup meant for machine processing. The former allows the human
reader to evaluate the context, the latter has no intelligence with which to
evaluate context. Consequently I would argue that microformats usage should
be as objectively universal as possible.

More simply said, it is fine for people to type "NSFW" next to a link they
put on a web page, but to encode it for machine processing would be a

> Some metadata represents subjective opinions, not objective 
> facts (e.g., hReview). Opinions vary. Ergo.

Reviews are opinions by nature but that which defines something as a review
is rather objective.  Further, one need look at the use case with which the
microformat would be applied.  hReview allows aggregators to find reviews,
"nsfw" would allow system to censor content. Those are two very different
use-cases so even if there were some subjectively in what was considered a
review and what wasn't, someone would get a longer list of reviews where
many are not so good as opposed to content being sensored by "nsfw."

Now if the proposal is instead to include identifiers that are objective,
I'd be far more supportive of that:

	<img src="..." class="nudity" />
	<img src="..." class="violence" />
	<img src="..." class="contains-the-f-word" /> 
	<img src="..." class="sexual-acts-depicted" />
	<img src="..." class="beastiality" />
	<img src="..." class="christian-sacrilege" /> 
	<img src="..." class="islamic-sacrilege" /> 
	<img src="..." class="jewish-sacrilege" /> 
	<img src="..." class="woman-sans-burka" />

Of course this could lead to a long list if we tried to cover all bases, but
"nudity" and "violence" might be a start.  Are there other classes you are
concerned about?

BTW, there is are a few others to specifically consider ;-)

	<img src="..." class="catholic-sacrilege" />  [1]
	<img src="..." class="swearing-in-using-the-koran" /> [2]
	<img src="..." class="pictures-of-mohammed" /> [3] 
	<img src="..." class="goatse-related" /> [4]

IMO, censorship is a very serious issue[5] and we should always err on the
side of censoring less, not more.  Of course if you are of the mind that
censorship is a good thing, then my arguments may not be compelling for you.

-Mike Schinkel

[1] http://www.usatoday.com/tech/news/2002/07/10/italy-porn.htm
[4] http://en.wikipedia.org/wiki/Goatse
[5] http://progressives.typepad.com/broadview/images/justiceDouglas_0.gif

More information about the microformats-discuss mailing list