There’s been a lot of talk about ‘gaming’ Memeorandum‘s algorithm lately, much of it linked to from the ‘Hack 2.0 Workgroup’ (a take-off on the Web 2.0 Workgroup collection of blogs.) I can’t speak to the effectiveness or ineffectiveness of any such gaming – I suspect any too obvious gaming will produce a less-than-interesting post and therefore not lead to traffic over the long run. However, my post on Amazon’s Mechanical Turk did (completely inadvertently) end up at the top of Memeorandum for almost a day and I’ve been wondering why it happened – since I’m hardly an A-list blogging rock star. Happily, Memeorandum archives itself every five minutes, so I was able to recreate how my post was ranked over time and come up with some theories.
The Mechanical Turk first appeared on Memeorandum on November 4th, at 12:25 PM – scroll down to find it, since it started near the bottom. Philipp Lenssen‘s post leads the discussion, with posts from four other blogs subordinate. Two of those link into Philipp’s post so this seems fair; perhaps being first to post also helped. I haven’t written my own post yet, so of course it didn’t show.
I break for lunch at my workplace and do a little blog-reading. Between learning about the Turk and deciding to write my own post I forget who tipped me off to the Turk in the first place. I do a bit of hunting and search Technorati but come up empty. I therefore don’t link back to Philipp (the most likely source) in my own post.
At 1:05 PM, my post on the Turk gets scraped by Memorandum and appears subordinate to Philipp’s. Interesting – there are still only four subordinate posts. My own takes the spot of Metafilter‘s. Is this because Metafilter’s post was short? Because Metafilter’s treated as a place for reposting? Because Memeorandum values new posts disproportionately? I don’t know. Whatever the reason for this bump, it’s definitely not due to traffic totals, since Metafilter towers over me in every metric I can think of. I didn’t take Metafilter’s spot for long – by 2:35 PM, Memeorandum’s decided to give the topic six subordinate links instead of four, and the Metafilter post returns.
At 5:50 PM, Memorandum moves the link over to the ‘earlier picks’ section in the lower right. It’s still got six subordinate links. At 5:55 PM, the subordinate links increase to eight. At eight links the topic has more activity than most of the stuff on the main section of the page, so it’s not surprising to see it jump to top of the main page at 6:20 PM.
When the topic first leads Memorandum, Philipp Lenssen’s post still heads the topic. Since folks like Greg Linden linked to my own post, I’ve become a subsection of the main topic, with my own subordinate posts. However, subordinate posts don’t require direct links – just like my post was originally under Philipp’s without linking to him directly, Paul Kedrosky‘s post soon appears under mine without any link from him to me. Once again illustrating that organization on Memeorandum has nothing to do with traffic. Why Paul’s post was considered subordinate to my own rather than Philipp’s – for this I have no clue.
At 7:35 PM Erin Bradley linked to me. More importantly, she doesn’t link to anyone else currently discussing this topic on Memeorandum – just me, Slashdot, and the source material. This puts my post over the top; I get the headline while Philipp’s post becomes a subsection. For unclear reasons, I stay at the top even though Philipp soon has more links in his discussion section, but after TechCrunch links to my post at 10:35 PM my lead seems more rational and I stay there until the topic itself is downgraded. Eventually Philipp’s post disappears from Memeorandum entirely. I suspect that wouldn’t have been the case had I just remembered what blog told me about the Mechanical Turk and linked to it, since my post would’ve then been identified as a continuation of an already-underway conversation rather than the start of a new one – by forgetting to link out, I think I hijacked the ‘spot’ of someone else.
If I was looking to replicate what happened with my Mechanical Turk post, I’d probably try the following:
- identify a very new post on Memeorandum – one on a topic I suspect will have legs;
- quickly write my own post on the subject, linking to the same source material but deliberately neglecting to link to the inspiration of the original post – this way I look like an independent root of the conversation;
- as new posts come in, I’d try to get them to link to my own post, maybe by commenting on theirs – any means of getting their attention without linking to them from my own post directly, which might risk its status as the conversation starter;
- finally, if I really needed some help getting over the top, I’d ask a friend or two to write a post linking to my own, but not any of the other posts on the topic.
I don’t know if that strategy would work while still seeming like natural behavior, but I strongly suspect that it might. Of course, that’s a bit problematic, since the strategy deliberately involves not linking to others, which seems more than a bit antisocial and selfish. In the end, I suspect the risks of being so manipulative aren’t worth the very-much-not-guaranteed rewards – during the day my blog led Memeorandum, I got about five hundred additional visitors and probably only a couple of extra links. Nice, but certainly not enough to put one’s reputation at stake. If I could lead Google for the keyword ‘home refinance’, on the other hand…
{ 2 comments… read them below or add one }
This is an excellent post. Reads like a sports game play by play.
I love memeorandum and the traffic is great when you are a headline (and much of this traffic converts to RSS, unlike digg and slashdot). However, it penalizes short posts (Dave Winer is often not shown because of this) and it penalizes too many links.
The second point is more troubling to me personally because I like to link to others that have joined in the discussion, so I tend to have a lot of links in my posts.
But there is a way to have both. Memeorandum already has the discussion from many blogs, and a permalink to that discussion. So more and more often I link to the Memeorandum permanent link (telling Memeorandum in the process exactly where my post should be grouped in case there is any doubt). I then link to other posts I’ve found that aren’t in Memeorandum for whatever reason (post too short, too many links, or they just aren’t in the Memorandum seed). It seems to work perfectly.
Oh, that was a fun read. Parts of your hypothesis are off, but overall a nice effort! Odd, you might have been the first to write about the continuous archiving…it’s escaped the attention of most people.
{ 1 trackback }