No More More Pages?

Redesign

Google’s good. But it could be better. Chances are that you’ve done a search where you haven’t found what you’re looking for on the first page. If so, then you’ve had to click on the unhelpfully numbered more-result pages:

Google's aging links to get more search results.There’s no semantic meaning in these numbers; there’s no telling what’s lurking behind a representing numeral’s bland exterior. If I find something good on the fourth page, I’ll be unlikely to find it again without aimlessly clicking on random number after random number. Normally, if I don’t find what I want on the first page, I’ll usually just give up.

But it’s not just Google. Alta Vista, Yahoo, Lycos, and all the major search engines conform to the same frustrating way of doing things. Why? Because it was the best solution at the time. A lot of today’s web technologies weren’t around in the mid-1990’s, so designers were forced to place search results on separate pages. But as technology has progressed, no one has thought to go back and redesign.

Slashdot's frustrating links for browsing history. Of course, this page-chunking phenomenon isn’t limited to search sites. It’s used everywhere from blogs to forums, from e-commerce sites to e-mail programs. And it’s surprising how often one finds oneself just giving up and going somewhere else when one has reached the end of a page.

The problem is that every time a user is required to click to the next page, they are pulled from the world of content to the world of navigation: they are no longer thinking about what they are reading, but about about how to get more to read. Because it breaks their train of thought and forces them to stop reading, it gives them the opportunity to leave the site. And a lot of the time, they do.

The take away? Don’t force the user to ask for more content: just give it to them.

At Humanized, we’ve recently developed a solution to the page-chunking problem. We don’t claim that its the end-all solution, but it’s easy to use and easy to implement with current technology. You’ll probably smack your head when you see it because it’s a “no d’uh” sort of thing. It’s called the “Humanized History” and we’ll debut it this week.

[ Update: We’ve released Humanized Reader! ]

This is so exciting to hear. Just today I was absolutely frustrated by an ebook that had medical content that I needed. I could jump using the index on the L side of the webpage or painfully click the right arrow to scroll thru the webpages of the site. Very frustrating and time consuming. I just wanted to read thru the content. Printing was not much better but at least I can refer back to prior sections without waiting for the pages to reload.

I am looking forward to seeing this product in action! Please target electronic textbook publishers.

I’m not convinced page-chunking is really such a bad thing. On the one hand, it is difficult to recreate google searches where the result you wanted was on page 4 or 10 or whatever, but I’m inclined to think the more common reason I don’t read past page 1 is because if what I’m looking for isn’t on page 1, then my search term wasn’t very good in the first place.

I agree that the lack of semantic significance behind numbered search pages is frustrating, though. And I think the point about navigation breaking the reader’s train of thought is a valid one. But on the other hand I think I’d prefer reading a book to a scroll. It seems to me that arbitrary page-chunking is the problem, whereas meaningful chunking of information is quite helpful.

It’s really interesting how people tend to stick to existing solutions and defend them saying “I’m fine with that, you don’t need anything better” :P

I don’t know what you mean exactly, the problem is very broad.

One problem is navigation in page-chunked documents. The solution here might be to make two zoom levels: first readable/scrollable, second: much smaller for “more / next page / prev. page”.. now if you click on the smaller it will enlarge to a readable form…
You don’t show the same information twice here (as in thumbnailed page navigation panel like in Adobe Acrobat) and you don’t need special keywords like “next page” or show the same text but in different zoom level twice (actually sometimes newspapers are repeating and enlarging some key phrases from the article text, thus showing the same information twice, which might be confusing, maybe they should use arrows instead :P).

Another problem is getting rid of chunking at all, here only mapping text to some three-dimensional surfaces comes into my mind as a potential solution… thus the effect is the same: you are basically hiding the information that is not yet available or not yet in the user focus. On the contrary you want to allow user to go forward as fast as he want and easy to remember how to go back into a specific place…

If you still can’t figure out the solution that suits human brain, maybe it’s time to change the information presentation form… I’m not talking about different font, chunking etc.., I’m talking about going beyond text and everything you know.. completely different visual representation of human thoughts…
Actually text is a sort of tree that you read by traversing it left to right (document has paragraphs, paragraphs has sentences, sentences has words, words has letters, letters are segments and joints). While our natural speech is more flattened you can still talk up to 2/3 topics in parallel I guess, and actually there is also some grouping (vowels, words, sentences).
Now when you are just processing information in brain you can’t usually think about whole topic in detail, so you try to create different levels of generalization, still it is hard to think in a opposite order that you remember it (like say digits in your phone number starting from the last one).
Some cultures are writting text right-to-left, some top-to-bottom, we are writting left-to-right… but to make it multi-direction you need to change letter topologies :P
Alright it has now not much to do with chunking.. who cares…
I can only imagine some revolution in visual representation of thoughts if we will create volumetric pens (ability to write information in space, 3d letter shapes, reading documents by not only scrolling and changing pages, but also rotating, etc..)…
so I was just going to say: there are really no limits where human interfaces can go :P

I’m not really arguing for page-chunking–if there’s something better, let’s see it. I just don’t think this article established any good reason to be chastizing Google, for instance.

Nor am I clear how adding more axes of movement is supposed to make information easier to locate. I would think it would do the opposite, adding spacial complexity to what used to be a spacially simple task. Of course I’ve never tried reading anything printed on cubes or icosahedrons, so I guess I could give it a try.

And just because I’m a linguist and am supposed to get picky about such things, saying there’s only some grouping in spoken language is the understatement of the year! Phones get grouped into phonemes, which make up morphemes, which make up words, words group into phrases, phrases into sentences, sentences into dialogues. And at least from words to sentences, the organization is all about tree structures…. The structure on a page may be more immediately apparent in some sense, but it’s no greater than the structure of speech.

Not that this is really relevant to Google ‘more’ pages. Although I do wonder what it would be like if search engines could chunk along linguistic lines. Either by doing some sort of semantic categorization, or even just by sorting an ambiguous search phrase into its possible meanings and chunking that way. Although NLP algorithms always seem to have large margins of error. Meh.

Ha! I think I guessed what the author of this blog is really talking about.
This weblog has 11 entries, they still fit into one big scrollable page without breaking it and there is no need to hide some entries to archive only.. but it can’t be done this way any longer, thus authors need “Humanized History”. I guess they checked out existing solutions for similar problem like Google and Slashdot articles archive etc.. and started to criticize and thinking about improvements.

Ok enough guessing, cause the error of any further conclusion might be too big.

For this weblog I guess they are chunking posts manually, and that is actually the good thing for Mike NLP idea: some algorithm could take that kind of blogs as an input for statistical learning where to chunk.

Mike, I think you bring up some good points. Atul also pointed out that if he generally doesn’t find an interesting link on page one, he refines his search term. I do too (unless I’m searching for something esoteric). However, I think something else is going on here: I scroll to the bottom of the page regardless of whether I’ve told Google to return 10 results per page, or 100 results per page. In other words, the impetuous to do a better search is generally the annoying-ness of going to the next page. It’s just easier to do a more specific search.

Tom, you raise lots of interesting ideas, but we’re thinking much simpler.

Hmmmm… I think at 10 results per page I’d go two or three pages before giving up, but point taken: there is definitely a realm of more pages where even if I do find what I wanted, I wouldn’t be able to find it again.

I’m still not clear on what solution is going to make it that much easier to find and re-find a result embedded deep within a tangled list of useless returns though.

I thought I’d post again now that I’ve seen the Humanized history, and I have to say I have mixed feelings on it.

Firstly, it’s not what I thought it would be given the teaser. It still chunks pages, but appends them rather than giving you a one-at-a-time view. This makes for a neat user experience as another page magically appears, and it holds true to the principle of “don’t make users ask, just give them more”. But at the same time, I wouldn’t want this system as a replacement for Google’s way of doing things.

For one, as I mentioned, I do occasionally go two or three pages into the “mores”. Sometimes I’ll find what I want there, and sometimes I’ll remember that that’s where I found it. So issue one is that the Humanized history has taken away my direct link to page three, where the information I want lies (I think there’s an issue here of browsing versus trying to locate a known item that was maybe glossed over by some false assumptions). Is this a huge problem? No. It’s a fiddly little detail, but then again the accusations leveled against Google could be considered the same.

My second complaint has to do with the havoc the Humanized history wreaks on the scrollbar. I tend to click and drag on scrollbars when perusing long pages of information, which I’m guessing is exactly what you were hoping people wouldn’t do. At the end of the page I was asked to release my hold on the slider, and when I did the slider was taken out from under my mouse and moved! Not only that but my point of reference for where I was on the page was modified in a fairly opaque way. Again, not a huge detail, but about as big a problem in my opinion as having to click “Next” at the end of each page, and not behaviour I would have labeled as “humane”.

All that said, I do like the reader. But I think that’s more for its simplicity than for the Humanized history. The history is fun, and certainly clever, but not a marked improvement over the old way of doing things in my opinion.

What if Google had a “followed links” history… just a page where the links you’ve actually followed from Google searches are arranged by date or something like that. It could be searchable, maybe even with a Google Suggest-style search box that autocompletes with the various search terms you used to generate the return items in your personal history. Relocating search results would be a breeze! Does something like this exist?

The answer, more or less, is yes! Google Search History is part of Google Personalized Search (Beta).

As I understand your critique of Google and many web sites in this article, you claim that chunking results into multiple numbered pages is bad for two reasons: the page numbers have no semantic meaning, and that clicking through pages interrupts a user’s train of thought. These were the problems I was expecting your solution to address.

I don’t see either issue solved in your solution though. First, RSS reading and browsing search results are very different activities, and don’t necessarily have the same solution. That’s another story though. Anyway, instead of having multiple pages of results, now you have a page that increases in length as you get to the bottom. Lets say I scroll what would be a few pages worth of results if it were chunked. If I saw something good on what would have been the fourth page, I might remember where on the scroll bar I was when I read it on this giant page. However, if I continued reading past that point so that more content was loaded, this position would no longer be the same! Instead of clicking on previous pages (where I had a hope of remembering which page had the result I wanted), I instead would have to either aimlessly scroll through a giant page, or attempt to estimate where to scroll to based on how much the scroll bar would have changed from the loading of content.

So how does this address the problem of page numbers not having meaning, and how does it interrupt a user’s train of thought less? Maybe I’m just too set in my ways, but I find it harder to find things on a giant page that dynamically changes length, and I find the behavior of the scroll bar interrupts my train of thought much more than clicking through pages of results.

Also, I find it a bit ironic that you criticize Google in particular for their page chunking. When I saw the Humanized Reader, I realized that I’d seen AJAX used before to dynamically load more content as a user scrolls toward the end of what has currently been loaded — by Google! Google also has an RSS reader: http://www.google.com/reader . The RSS headlines are listed in a box on the left hand side of the screen. At the bottom of this list, it say something like “Item 1 of more than 11″. As I scroll through the headlines and get towards the end, it loads more results dynamically, and increases the size of the list. However, because there is a separate list of items on the page, my scroll bar is not affected by the appearance of new items.

So I can’t help but wonder: what exactly is the contribution here?

Some comments on pagination:

I don’t think it’s fair to criticize search engines for the way they paginate (particularly google, who pays a great deal of attention to their User eXperience). I say this because search results, by their very nature are heterogenious (you don’t know if the results being returned are relevent to the user’s desired goal), and as a consequence, chunking by number is by definition non-semantic. Unfortunately semantic chunk would require you to have a postiori knowledge of the set of objects you are listing (since chunking would then be relative). On top of that, search engine indexes change. So a result on day X found on page 4 could have wandered to page 2 by day Y or to page 10 by day Z, adding to the complication of attempting to chunk results semantically. More or less, you would have to have solved Search in order to remedy this problem. So i think it’s extremely unfair to criticize search engines for the non-semantic way they list results.

Now as a tangent before i defend pagination in general, long lists, as Alex pointed out, do not solve the issue of being able to algorithmically find a query that you discovered previously. Regardless of of whether you’re dealing with a continuous list of results, or a paginated list of results, you have two choices, to refine your query or dig through each individual result. As this discussion has previously noted, most of us (geeks that we are) refine our search queries. For google, that means typing in a new query and fetching a new list of results. For something like the Humainzed Reader, it means trying to figure out an accurate description to type into your browser’s search function.

So for the second method, back to pagination. What does pagination buy you? If you are going to plow through each result summary, pagination basically serves as semi-randomly placed indexes in a list of results. Instead of having to scroll through all of the results, you can jump straight to the middle of the list of results. This can be good, this can be bad. I find Flickr’s pagination system extremely irritating, not because they paginate, but because their pagination system abbreviates. if you’re on page 600 of 1000, they only give you access to pages 1-5, 595-605, and 990-1000. They’re basically failing the purpose of pagination (So you have to result back to searching through tags or looking at every 5th page, to get more indexes further in the list).

There is no similar possibility with a continuous list. You have your single start index, and then you’re boned. You’re back to either doing a direct traversal of each result, or searching. This problem is exacerbated by the loading mechanism in the Reader (and don’t get me wrong, i think it’s clever, i just don’t think it solves any the problems under discussion), which forces you to load new chunks of data (please, note these chunks are semantically meaningless as well) before you have the full list to be able to jump around or search for your result.

Regards,

-T

I’ve posted a response to these fabulous comments at No More More Pages? Part 2. Take a look, and let me know what you think.

I’ve seen this never-ending scrollbar trick on a number of sites before and hate it.

One, the standard scrollbar is often replaced with an image-based scrollbar that isn’t as smooth, sometimes quite quirky (start dragging and the browser sometimes thinks you’re trying to drag and drop an image to your Desktop or something instead of moving the scrollbar image).

The worst part is what Paul mentioned above–you have to remember exactly what position in the scrollbar you saw a result you want to go back to.

And all this ignores the accessibility issue–the standard pagination system works everywhere even on mobile phones while this scrollbar technique requires a modern browser on the computer and a mouse (yes, there are people who browse with a keyboard-only).

Jon,

After we came up with Reader, we were forwarded to Microsoft’s live.com which implements a similar concept of infinite history. We found that their poor implementation got in the way of the concept for the reasons you mention. However, just because one implementation is poor does not mean that the concept is poor.

Live.com focused on solving the problem of how a scrollbar should behave in an infinite history environment. The result was painful. Humanized History makes use of the browser’s scrollbar, rather than re-implementing the wheel without improving it.

We’ve been using Reader for a while now, and we’ve found that most of the problems we and other’s anticipated weren’t really problems. Pontification is fine, but user testing is better. We are still working on the concept of visual landmarks to help mitigate the “lying scrollbar” problem that we are stuck with. We recently added large in-line dates to help you find your place. But it is worth pointing out that if you use the keyboard or scroll wheel to navigate (our preferred methods) you’ll almost never even notice the scrollbar jumping around.

The problem here isn’t that there is a lack of a solution to this problem, but rather we lack the bandwidth to be able to load enough data to see beneath?

I made a tutorial using the plainest of plain HTML to demonstrate the concept. I was surprised; you can navigate search results that give hundreds of hits without getting too lost.

What do you think about the pageless imagesearch from Windows Live Search?

For the first time it was a litte bit confusing, but then I loved it more and more.

Greetings from Germany.

Christoph

Well I created a mod to phpBB3 that used this idea for topics and silently implemented it on my forums (http://www.thetangledweb.net/viewtopic.php?f=29&t=23) and thus far I have not heard any complaints from my users, my guess is they dont even notice the change

interesting site

Supberb idea, easy and yet noone ever thaught to create such a thing.

The only minor thing: when I use paging as a user, I sometimes now where the content is: for example in the middle, so on page 40-50 somewhere.

With your solution, I’d have to go through pages 0-30 before I can go to page 40.

Maybe combine it with a “jump to # page” textbox

This would become the future

However this effect the nos. of page views … right. The stuff on which the www is surviving ?

why not give the whole net on one page. this is just plain stupid. go back to being a taxi driver

Browsers can handle 100KB of search results. Users can scroll. Paging 10 by 10 is not technically necessary.

On our site, we’re limiting the number of search results to 250, handing all of those over at once, but never more.

See for example

http://stbgo.org/mapping/search?centerlat=31.952162238025&centerlng=-15.46875&zoom=1&minlat=-23.241346102386135&maxlat=67.33986082559095&minlng=-180&maxlng=180&query=rice

Thanks —

Stephan

I use repaginator addin in Firefox to get more on one page.

So therefore the *only* solutions is to offer endless scrolling with permalinks and/or vertical pagination… The best of both worlds… the permalinks would address the issues of finding your results back the next time and the vertical pagination gives you the control of navigation back…

What also could be implemented is ‘Save to favorites’ link where your favorite results (wether they’re search results, blog results or rss feeds) can be saved with one click and can be found later on…

Love the idea.. but it needs perfection!