Forum Moderators: open
Of course, this may depend on many considerations, visual appearance to name one.
When the page limit is reached the category must be branched into subcategories, each of which may carry an additional nLimit pages before it in turn has to be branched. In this way, a uniform page limit number will be a measure of the capacity of a directory at a given number of sublevels. The page limit should thus be an important characteristic of a directory.
Does ODP have a defined page limit? What about Yahoo? What are your suggestions for an optimal page limit, i.e. one that provides enough capacity and still keeps the user scrolling to the end of the category?
But that gives fairly broad numbers in between. But there are other factors.
Some categories may have many logical subdivisions (such as dog breeds) but the resulting categories may vary a lot (more dalmation sites than mastiffs?).
Even dividing surname categories by alphabar means a large "B" and a small "U".
And there are other considerations. It's important to be consistent; so each town (I'm guessing here) might have a cinema subcat, if there's two - or 52.
In other words ... it all depends!
In the early days of dmoz, according to rumor, the boss knew there was huge potential growth - so editors were encouraged to think about subcats if the number was greater than - I think - a dozen. Now, with mere categories than Dewey Classifications, editors can create new subcats as required - but are encouraged to think first - to avoid duplication or overlap (Pets/Dogs/European Breeds .... Europe/Animals/Dogs/Breeds).
In some areas, editors are expected to offer proposals for discussion before creation (or deletion).
Capacity is not an issue at all; content is infinitely more important - and there's easier ways of counting than totting up categories :)
* 100
* 25
* 12
However, there's an exception to every rule as Quadrille points out:
"Some categories may have many logical subdivisions (such as dog breeds) but the resulting categories may vary a lot (more dalmation sites than mastiffs?)."
I do agree that the real world is much more complex than a web directory can depict and that any rule on page limit must have exceptions where appropriate. Still I think a "statistical" rule is useful to define both as a measure of the capacity (calculation below) as well as the usefulness of a directory.
BTW, I just had a look at ODP, at the bottom of the page it says: "over 3.8 million sites - 51,500 editors - over 460,000 categories", which calculates to an average of 8.26 pages per category.
--- Suggestion for capacity calculation of a directory with uniform dimensions ----
Let:
T = total number of pages in directory
L = number of hierarchical levels
S = number of subcategories in one category
P = number of pages in one category
T = S*P + S*S*P + ... + (S**L)*P = P * sum (S**N), where 0 < N <= L
The capacity of a directory will thus be directly proportional to the page limit for a given configuration of levels and branches.
In other parts of a directory you may have sibling cats. One branch of the directory may push another part into creating subcategories.
If none of the above applies I think a dozen sites would be too few, but dividing a category makes sense when you've got a score or more. At about 50 sites it's starting to be painful, and by the time it's 100 I believe you're heading for trouble.
Division by alphabet and numbers is a last resort. To avoid tiny letter and number categories the threshold for dividing a category this way is usually far higher than for example a topical/thematic division.
Google yields 10 per page by default. Yahoo...a little more. MSN about 15 or so. The max in Zeal at a node is supposed to be 10 (not including subcats.)
I've never like alphabars. If you need an alphabar...you need to rank the sites in a different way. (by order of submission, by town size, by website size, by state, by your opinion of importance or many other ways. Just my opinion)
... but I always crank it up to 50 hits, so that I get a nice long list to look through, rather than having to hit 'next' every few seconds like you do when going through the list 10 sites at a time.
I've never like alphabars. If you need an alphabar...you need to rank the sites in a different way.
There are two circumstances where an alphabar is the perfect solution:
1. Where the category is 'names' - once it gets big, an alphabar is entirely logical.
2. Where you have exhausted other methods - eg you have a huge 'Miscellanous / general / varied content category' - then anything except an alphabar is illogical and confusing - and looks desperate.
Alphabars would rarely be a first choice - but if the alternative is artificial differences or huge categories, then they do the job.
I end up taking hours to figure out a better way...it's just a personal preference.
(But I agree with you on the fact that when you've exhausted other methods..)
