10 November 2005 · matching rem
When designing software that displays lists, remember that you're trying to optimize for the user's experience, not the page size or the roundness of numbers.
1. Scrolling is far easier than paging, so generally the more you put on each page, the better off anybody is. For brief items like query results, think 100 instead of 10, for example, and adjust from there. Preferably up.
2. Never orphan a page with less than half as many items as the page size. That is, if you have less than 1.5x items to show, put them all on the same page. 104 items with a nominal page size of 100 items should produce a single 104-item page, not a 100-item first page and a 4-item second page. 428 items should be shown on 4 pages, not 5. And that's 1-128, 129-228, 229-328 and 329-428, not 1-100, 101-200, 201-300 and 301-428. A little more information never hurt anybody much.
3. After spending 5% of your effort on paging, put the other 95% into obsessively tweaking the format of the results so that they convey as much information as can be conceivably clearly expressed. If you haven't read Tufte, read Tufte. If you've read Tufte but think his advice doesn't apply to your case, read him again.
1. Scrolling is far easier than paging, so generally the more you put on each page, the better off anybody is. For brief items like query results, think 100 instead of 10, for example, and adjust from there. Preferably up.
2. Never orphan a page with less than half as many items as the page size. That is, if you have less than 1.5x items to show, put them all on the same page. 104 items with a nominal page size of 100 items should produce a single 104-item page, not a 100-item first page and a 4-item second page. 428 items should be shown on 4 pages, not 5. And that's 1-128, 129-228, 229-328 and 329-428, not 1-100, 101-200, 201-300 and 301-428. A little more information never hurt anybody much.
3. After spending 5% of your effort on paging, put the other 95% into obsessively tweaking the format of the results so that they convey as much information as can be conceivably clearly expressed. If you haven't read Tufte, read Tufte. If you've read Tufte but think his advice doesn't apply to your case, read him again.