[uf-discuss] hReview feedback

Paul Bryson paul at msn.com
Wed Jan 18 10:35:42 PST 2006

"Mark Rickerby" wrote...
> In terms of machine readable data, that is an interesting idea. You
> might not even need the [1,100] in this case - a % sign would
> communicate this range directly and is also more readable to humans.

I would think that percents would be much more universal, but it is obvious 
that percents are rarely used in the ratings world.  Is there a single 
website that does ratings that uses a percent instead of some 1 to X scale? 
I wonder why this is.  It could be hold overs from old ways of doing things, 
or it could be a huge caution sign to usability.

> True, but this seems like a slightly separate use case from the
> standard hReview, which is a *specific* review written by an author.

I think I must be the only one that finds this distinction strange.  Perhaps 
someday a different microformat will be thrown together just for aggregate 

> Which of these following ratings is more readable in browser?
> Which has the richest semantics?
> Which is easiest to parse?

<img src="stars-2-0.gif" class="rating" alt="2 [0,5]" />
<img src="stars-2-0.gif" class="rating" alt="2 out of 5" />
<img src="stars-2-0.gif" class="rating" alt="2 (0..5)" />
<img src="stars-2-0.gif" class="rating" alt="2 (out of 0..5)" />
<img src="stars-2-0.gif" class="rating" alt="2 out of 0 to 5" />
<dfn class="rating" title="2 [0,5]"><img src="stars-2-0.gif" alt="* *" 

I think the second is easiest to read, my only real problem with it being 
that it doesn't provide the lower bound.  Number five is much the same but 
with a lower bound, but it doesn't seem very natural for reading. 
Mentioning the lower bound in text just feels so out of place, especially if 
that lower bound is zero.

One quick point about parentheses and math syntax.  [] are used for ranges 
that are inclusive, as opposed to ().  So [0,5] could have a value that is 
zero or five whereas (0,5) couldn't.  Might be an important distinction.

> I don't think there are necessarily clear cut answers, though I would
> definitely tend towards encouraging the alt/title text to be closer to
> plain english.

They should lean towards plain English, but the results just seem very 
dissatisfying.  This is why I pointed to the date-principle where the 
visible text is human readable and the Title attribute contained information 
that was more machine readable than human.  As the Title/Alt attribute is 
likely to almost solely be parsed by machines, it may be prudent to follow 
the same course of action.

 - who thinks he is never going to feel satisfied with whatever answer comes 
out of this. 

More information about the microformats-discuss mailing list