Eric has a new book you should buy, despite the fact he’s all wrong about weblogs. His thesis is that it’s easy to read chronological items that relate to each other from top to bottom on a page. It’s not that Eric has his facts wrong, he’s just looking at the wrong facts.
Weblogs are not 20 chapter books. Eric’s entries refer and develop more than the average weblog’s, he is no exception to the rule that most weblogs would be just as intelligble if they randomly ordered the entries you haven’t read yet than if they presented the newest ones at the top. For some reason Eric insists on using monthly archives for his permalinks, which isn’t helping his visitors or Google (or the bandwidth he’s concerned about). Anchored monthly archives should never be used for permalinks. The only people who still use these are those who are technically hindered from or too lazy to implement post-level permalinks.
On monthly, weekly, or daily archives it makes perfect sense for the entries to be ordered chronologically, because the defining characteristic of the posts is their chronological relation to that date range. On a search result page, I want to see the results ordered either in reverse chronological order or by some search relevance. Both Technorati and Feedster get this, and both default to reverse chronological order of their results. On the front page, which is where Eric has his main beef, the most relevant thing to at least 95% of visitors is going to be the latest items. Anything else is going to be making the site less useful to the majority of users for the benefit of a few. There is a much larger precedent for putting the most recent items at top than just weblogs: most email and webmail software I’ve used, every press release page, news sites like CNN, Zeldman, the Board of Governors, the White House. Breaking the convention of thousands of familar sites for the benefit of the occasional reader who checks back every week or so but is really annoyed at having to scroll funny to what they’ve missed breaks the Hippocratic Oath of design. Weblogs aren’t ordered the way they are because of some freak historical accident, they’re ordered this way because it works.
(A sidenote about the bandwidth issue: it’s dead. Of course pages should be made as small as possible using standards and efficient markup to help them load quickly. No one is going to argue that. However an optimized 10K page loaded thousands of times a day is still a healthy chunk of bandwidth. If you are adjusting your content or paying more for the popularity of your website, find a better provider. The server that this site (and others) is on is allocated 1.2 terabytes of bandwidth transfer per month. At worst it uses a third of that. Server bandwidth should not be an issue anymore these days.)
I mostly disagree with his post, however I feel Eric’s pain. I’ve had to do the scroll-catchup thing before, and it was annoying. Without breaking the website for the vast majority of my users, I can offer some relief:
- Eric is a busy guy and I’m guessing he uses a newsreader (like NetNewsWire) to keep track of who has updated. If you click through to the permalink of a new item it will take you to the individual archive for that page, complete with comments. I have always had intra-post navigation at the top of the page, but after reading Eric’s screed it occurred to me where that would really be useful is after you’ve read the article and want to move on to the next thing. So now I have post navigation above the article, after the article, and after the comments. If you’re a few posts behind, navigate the individual archives instead of a monthly archive or the front page. Enjoy.
- One benefit to a completely dynamic system is that you can change views on the fly pretty easily. So if the reverse chronological thing bothers you that much, at least with WordPress blogs you have a easy fix. Simply add
?order=ascto any WordPress URI and it will order the dates ascending instead of descending. Example: my March archives, my March archives descending. It would also be pretty trivial to set a cookie to allow people to see things ordered different ways, and I imagine within a few hours of writing this there will be a new hack or plugin on the forums. If only everyone used WordPress.
“It would also be pretty trivial to set a cookie to allow people to see things ordered different ways, and I imagine within a few hours of writing this there will be a new hack or plugin on the forums.”
You were practically begging for it. 😉
“The only people who still use these [monthly archives with anchors] are those who are technically hindered from or too lazy to implement post-level permalinks.”
Or have links blogs and feel that having individual entries is a little much for one-liners. But you’re otherwise right: a site of mine wouldn’t be #1 for a cell-phone product offered locally (an entry which has dozens of comments, by the way, of people reviewing that product!) if it weren’t for individual archives.
“If only everyone used WordPress.”
Then WordPress would be the Windows of personal publishing software, and not MT. Do you really want WordPress to be the Windows of blogging?
Interesting points Matt; #1 up there is something I hadn’t thought of, being an old school blog readin kinda guy (ie, I don’t use a newsreader right now) I never thought that many people may enter a blog directly to a post. Something to keep in mind while developing the blog at fiftyfoureleven (though I suppose having an RSS feed would be a good start…)
Personally, I agree with you. New visitors to a weblog aren’t walking into the middle of a book (or movie). Weblog posts are rarely written to maintain continuity from one post to the next. (If there’s any place for continuity in a weblog’s archives, it would make the most sense in a categorical archive.) Visiting a new weblog is more like walking into the middle of a conversation; forcing someone to start again from the beginning (for the benefit of the new guy) is just a hindrance to the discussion. The great thing about non-linear text (i.e. hypertext) is that you can choose to go back in the conversation if you want to and review what you missed. It may require a little extra work, but at least the option is still there.
“The only people who still use [monthly archives] are those who are technically hindered from or too lazy to implement post-level permalinks.”
Im my case, for the moment, it’s both, plus one more point. I view archives as a backup way to catch up with a site you’ve not visited in a while, or that you’ve only just found. Monthly archives are easier to read through than clicking from post to post. (I do fall down in that the months don’t link together, but that’s another reflection of laziness.)
“For some reason Eric insists on using monthly archives for his permalinks, which isn’t helping his visitors or Google (or the bandwidth he’s concerned about).”
Actually, it can save bandwidth over time. For someone reading through the archives, downloading the month of March once (with a single set of navigation links, footer, masthead, etc.) requires less total bandwidth (and server hits) than clicking through the archives at one page per day (eahc of which has all the extra stuff surrounding the post content). The flip side, of course, if that a person who wants to see a single post has to download a whole month’s archives. Figuring out which is more popular is, well, not easy. I suppose that an archiving system that allowed for both would be ideal, but creating it would probably take more time than I have at present.
“On the front page, which is where Eric has his main beef, the most relevant thing to at least 95% of visitors is going to be the latest items.”
Even if we substitute “important” for “relevant” (because the most relevant thing is whatever will be of the most benefit to the user, no matter when it happened) I don’t agree. But if you’re serious that the most recent entry is that important, then I can’t wait for you to fix your comment ordering so that the most recent comment is first, and the first comment made comes last. The most recent reaction to your post would be the most relevant, after all– wouldn’t it?
“There is a much larger precedent for putting the most recent items at top than just weblogs: most email and webmail software I’ve used…”
And Usenet software. I hate top-posting; I’d meant to mention that in my post but forgot. I have a signature file that reads:
Yes.
>Are you sure?
>>Because it reverses the logical flow of conversation.
>>>Why is top-posting so annoying?
I find the same true of reverse-order weblogs.
“Weblogs aren’t ordered the way they are because of some freak historical accident, they’re ordered this way because it works.”
From where I sit, it doesn’t work as well as would chronologically ordered posts, assuming a good UI can be found. (And there’s the rub, at least for the moment.) There are a lot of examples where something was widely adopted and everyone just went with it, later claiming it was superior because it “just works”– table-and-spacer layout comes to mind right away.
“A sidenote about the bandwidth issue: it’s dead.”
If the bandwidth issue really is dead, then I need to stop the whole CSS thing and get back to promoting table-driven spacer-GIF layout, because frankly it works more consistently than CSS-driven layout across multiple browsers. A main reason I’m still pushing for the use of CSS is that it saves bandwidth, and (as you say) makes the user experience better. Without it, things may be a little slower, but at least it’ll be more consistent for the users.
That probably sounds like a hyperbolic piece of rhetoric, but it isn’t: I’m serious. Bandwidth efficiencies are a big reason for using standards-oriented markup. If that concern is dead, then so is a big chunk of the rationale behind the standards push, and I don’t have the time to devote to losing causes. As it happens, I don’t agree with you, so I’ll keep soldiering on for now. Perhaps I’m looking at the wrong facts here as well.
I happen to agree with both you and Eric… and of course fall back on the old “it depends on the purpose/content/audience” argument.
Some weblogs may simply be short blurbs of thought or a few comments on a handful of links. For these, the subject and content of the posts usually don’t depend on any other post. An anchored monthly archive page is fine as it allows you to quickly read many posts at once, and reverse chronological order also works because it gives you the most recent (and topical) content first. (Imagine a blog started September 11, 2001 but whose content covers anything in the news. Would prefer your first reading on this blog be of 9/11 commentary, or of commentary on the presidential race/mars rovers/superbowl controversy?)
Other blogs, with lengthy paragraphs of posts, may be more like journals. For these, understanding of each new post may rely on knowledge gained from previous posts. In this case, page-per-post archives work best because they allow you to read a whole post without being distracted from chronologically previous (though not necessarily subject-related) posts.
Perhaps what is needed for the home page of these types of blogs isn’t the same as what is needed for the short-post types of blogs (newest posts on top order). Perhaps these need a layout where the most recent post is the only one shown on the home page, with short summaries and links to other subject-related posts below. This way, new visitors to the site can get an idea of the writting/subject of the site, and if they want to read more, they can quickly find the start of the “story”.
While well written, I somehow get the impressions that Eric has issues. What gets me is the bandwidth arguement. It just doesn’t fly with me. If he is worried about bandwidth, then you DO want the newset entries at the top where they are easy to find. Other wise, one would have to wait for the “old” content to load, then click a link (presumably) and wait for the new content to be sent. Twice now, the user used bandwidth that could have been saved by loading the newer content in the first place.
Secondly, the arguement regarding cross linked posts. In my personal opinion, that’s a lazy blog writter. Usually, when I post something that is related to another post of mine, I will include the link. It save the reader the hassle of trying to find it, and saves on the bandwidth as well. I have to admit, I don’t always get it right, but I do my best. I also agree with your point about the bandwidth issue being dead. With all the AOL Optimizers and NetZero Accelerator and DSL/Cable modems out there, the only time it’s an issue is at the webhost. If you are getting that much traffic, well, either the page itself needs to be optimized, or you need to spring for that extra capability. If only I had that problem. But, for the 90% of the bloggers out there, I don’t think it’s an issue.
Another precident to the newer first, older last convention is the use of forums. Nearly every forum I’ve seen has some sort of listing where new posts are listed first over older posts. Does that mean they too are broken? No, it’s a feature. In fact, I’d be lost in most forums with out it.
I’m not sure just what set off Eric, or what he’s smoking, but I think things are fine.
TG
Sorry for the double comments, I had moderation turned on without meaning to. Of course since the first and second comments aren’t identical, I’m not sure whether to drop the first one, which presumably was less edited or thesecond one to preserve temporal integrity. Or reorder the comments with the newest on top. 😉
Stephen, you’re too cool. Interesting approach altering the array and not the query. That actually works better with
posts_pagedenabled.Joe, I think even for weblogs that focus on longer entries, like this one, reading things strictly chronologically doesn’t give you a lot of benefit. When a weblog entry speficially relates to a previous one, I link it, and the old one is usually updated to link to the new one. If I blog about Lockergnome, post a bunch of silly jokes, and blog Lockergnome again, reading chronologically isn’t going to help you follow the thread much better. I’m about to post again about Dvorak, which I haven’t written about in months, should you have to read through those months of archives to follow the thread? I don’t think so.
Eric, monthly archives are good for catching up, though honestly when I get that far behind I just let it go. Ideally you should have several different representations of your posts (by month, by category, etc) and I think that’s one of the defening characteristics of a good weblog. It is impossible to know whether someone is on your archive page for a single post or for the whole month the way you have it set up because people don’t have an option either way. If you offered both you could look at the stats to see which people use more. For me, it’s individual archives by a landslide.
Furthermore, although in the situation you set up a monthly archive would save bandwidth, I would say that’s the exception to the rule. Surges in traffic usually come from Slashdot or Instapundit or Zeldman sending a couple of thousand new friends your way to check out an entry the found worth linking. This is very common and I think that is the source of most bandwidth anomalies, and it’s percisely the type of situation that overwhelmingly favors individual entry permalinks.
“I can’t wait for you to fix your comment ordering so that the most recent comment is first, and the first comment made comes last.”
I know of sites that do this, and it would be useful if every comment was a reply to the post itself, and there was no other metric for ordering the comments (like quality). However as this very comment illustrates, mine tend to be more conversations, and putting them in reverse order would be like top-posting.
“I hate top-posting; I’d meant to mention that in my post but forgot.”
As do I, to an extent. However I wasn’t referring to the conversation within the email itself, but rather many email/webmail programs organize the inbox with the newest items on top by default. This is true of every webmail I’ve used, but not so much for desktop clients. My other examples are stronger.
“If the bandwidth issue really is dead, then I need to stop the whole CSS thing…”
You know we’re in complete agreement here. Page should be as small as possible for every user, but on the aggregate the bandwidth savings isn’t a user issue but a publisher issue. Bandwidth savings on the aggregate can present an easily-quantifiable business argument for smaller pages, and I’m sure that has convinced many of your clients, but I still think the biggest benefit (though not an easily quantifiable one) is to the users for whom the page responds faster. Also we’re talking issues of scale. If AOL or ESPN comes to you, they talk in scale of millions of users and my measly terabyte would last a few minutes on their sites. That aside paragraph was more directed at the people I know who end up writing checks of $50–100 to their hosting company just because they had a few popular posts. Hosting is a commodity, and they should shop around. The large companies should still hire you, and I have every confidence that the “CSS thing” will continue to work pretty well for you in the future. 😉
Oh, the irony…
Umm… Who wouldn’t want to be the Windows of blogging?
“On the front page, which is where Eric has his main beef, the most relevant thing to at least 95% of visitors is going to be the latest items. Anything else is going to be making the site less useful to the majority of users for the benefit of a few.” (emphansis mine)
According to Jakob Nielsen, entries in his blog receive most page views after they are archived and removed from his home page.
I blogged another possible solution: http://richardathome.no-ip.com/index.php?article_id=178
In a nutshell – just display a breadcrumb list along with every article that shows the chain of articles in full.
It does mean some minor modifications to your blogging software though…
First of all, Matt, if you’re going to link to previous and following posts, why not add tags describing these relations to the header aswell? Enables me to use my link toolbar :).
Second, I agree that reverse chronological order is the way to go for weblogs (and yes, Eric’s different ordering for his frontpage and his archives are confusing). As has been said before, and something which I very much agree with, not all posts are about the same topic, so forcibly having to read them while they have nothing to do with the actual entry you’re interested in… doesn’t make sense to me. Reading previous posts on a certain topic can in some cases surely be interesting, and for that a well-placed link can point your readers to it. However there are also a lot of cases (especially for incidental visitors) when you don’t want to do a 2-hour reading job of previous related posts before you can get to the actual message of your interest (which you for example read about on say, slashdot).
There is no ‘universal truth’ about chronological vs. reverse-chronological order, it depends on the context. Blogs are not a forum, nor a newsgroup, nor blog comments for that matter, which contain a list of replies on one topic, each strongly referring to previous replies. In such cases it’s just not do-able to read the last replies without the context of the previous, so yes, you should never use reverse chronological order. However, you probably know those 10-page topics which you are interested in, and which take hours to read through before you caught up… Though interesting, otherwise you wouldn’t be reading it at all, it’s a bit of a waste of your time. So you are presented with a choice, am I going to read all of it, which is quite excessive in comparison to my interest in the topic, or will I just not read it at all, even though I am interested in it?
Fortunately, weblogs are not written like forum posts (meyerweb.com included). They are usually written like seperate entries, with hyperlinks in them to refer to previous context. what I (and I guess everyone else) do if I read this sentence “While I knew I was staking out a position that was likely to cause some controversy, I’ve been rather surprised at the response to my post on Weblog Weirdness.” is that when I don’t know what this is referring to exactly, I immediately click on that link to catch up with the context (well, maybe giving it a quick scan first).
I think this is a good way of writing weblogs, a blog entry should be like an individual text, which refers to the older entries. This is also how most scientific documents are written, because they *are* usually presented as individual texts, published in different magazines, different issues of magazines, or different books. Out of nessecity, one might say, but I don’t think it’s a bad thing, it’s just a different style of writing, and a nice one if you ask me. And fortunately this is the internet, we are blessed with hyperlinks, so all this becomes a lot easier.
Maybe something that Eric’s entry could be improved on a little is to include a small introductionary text for the new readers, who don’t know the context of the entry. A nice example of this you can see in this weblog entry by Matt is the sentence “His thesis is that it’s easy to read chronological items that relate to each other from top to bottom on a page.” That immediately brings your readers up to speed with what it’s all about, and if someone wants to know more details about what Eric said, he or she can follow the link. A similar sentence could also be written when referring to one of your own web log entries.
The same goes for the top posting vs. the bottom posting issue by the way. In my opinion, it is not true that top-posting is always wrong. Your ‘signature’ is a wonderful and very funny example of where this is the case (I’ll remember it so I can quote it sometimes :)). But when a reply is written like a letter, in prose, maybe even having a small introduction and conclusion and all… I do not see any reason to bottom-post! And as it happens I know several people who reply that way, so I wouldn’t call it uncommon practice, if the situation calls for it I myself sometimes do it aswell. It’s a very interesting way of writing emails, not the point by point replies you’re used to but a whole new ‘story’ each mail. Ofcourse, in that case quoting really wouldn’t be nessecary at all, and on newsgroups you can always read up the thread, but I guess that in case you’re interested having some context of the previous discussion at the bottom of the post could come in handy (especially on a mailing list where people might join in on the list any time).
My email client sorts chronologically, and moves to my last read message somewhere at the bottom of the list. This is a preferrence which can be set in basically every email client, and also in most webmail accounts, so apparantly many people have a different preferrence. However, this really doesn’t matter. I read all messages which just came in and are marked as unread, and I read the one with the oldest date first. My chronological ordering is just a matter of preferrence but if it was the other way around, it wouldn’t alter the order in which I read them, nor would it make it any more difficult to do.
One final example… This big comment basically draws only on Eric’s and Matt’s blog entries… Would it really be nessecary to have this chronologically ordered? If you ask me, it would be easier if I just linked (Eric’s, Matt’s) to them. Because reading all these other chronologically-ordered comments really isn’t nessecary.
Conclusion: when writing hyperlinked texts, the ordering does not matter for the content itself (as long as it is applied with some consistency). Ordering is meant to give users easier access to the content. As I said, my email client sorts chronological, but moves the focus down to the bottom of the list, so that I have easy access to my latest emails. As moving focus down the list is somewhat hard to achieve in web browsers (and when done through javascript may be found annoying), combined with the fact that it’s not trivial to remember what the user read last, reverse-chronological ordering makes sense. Another option such as Eric discussed in his latest weblog entry, as long as the end result is user friendly, is just ‘another option’.
~Grauw
Matt-
You can display posts in chronological order even easier in .htaccess. Just add “&order=asc” at the end of the relevant rewrite rules. That way it’s invisible to users and doesn’t make your URIs longer. I’m sure you knew that, but I thought I’d post it for people like me who got here via google looking for a solution to the sorting issue.
Shouldn’t the “my March archives descending” link read “my March archives ascending“?