Ansuz

u r my favrit web log bout publing. Owen - 2010-10-21 16:02

Have you considered that instead of listening to rubbish about laws of attraction and the like, you could look for patterns in the sites that _you_ read, and which also appear to have large readership? Maybe also look at a few of your friends' more heavily read sites (since you probably want to attract/retain people who are like your friends). Surely you could apply your comp-ling-fu to come up with all sorts of interesting metrics that you could check for correlations with readership. (Once interesting thought that I just had is that simply publishing which of these factors are and aren't correlated would probably make for some pretty interesting content; I mean look at how much attention OkTrends gets). kiwano - 2010-10-22 13:04

<I>Have you considered that instead of listening to rubbish about laws of attraction and the like, you could look for patterns in the sites that _you_ read, and which also appear to have large readership?</I> Yes, I have invested hundreds of hours in that over the course of years. Matt - 2010-10-22 14:45

Could you publish the results of those investigations? (Also, hundreds of hours over the course of years would accurately describe 200 hours over a span of a decade, which averages out to under 3min/day; I'm hoping that you've put more effort into it than that ;) ) Really, I have some hunches that have crossed my mind while reading the blogs that I read, as to why some of them seem to generate a lot of interest; I'd be really curious to see how these hunches line up with the results of your investigations. kiwano - 2010-10-22 15:10

How many hours would you consider sufficient that if I spent them without getting favourable results, you would agree we must discard the hypothesis that the "imitate other sites I read" technique works, and that it is completely inappropriate to recommend spending more hours on it? You seem to be saying that 200 hours wouldn't be enough; so would 500 be enough? Would 2000? If there is no limit, then your claim seems not to be falsifiable, and in that case it's a matter of religion, not science. Matt - 2010-10-22 20:33

The fact that you're talking about "favourable results" and the "'imitate other sites' technique" tells me that you didn't understand my suggestion in the first place. My suggestion was as follows: - Take a bunch of other sites that seem worthy of imitation (i.e. sites that look like they get a lot of the sort of traffic you'd want) - Construct a metric of worthiness for imitation based on measurements that you can take of these sites (e.g. (number of comments)^k * (portion of comments meeting some diction cut-off)^l for values of k and l tuned to your relative (dis)interest in traffic and s/n ) - Go and wget a large corpus of sites, and calculate the worthiness-of-imitation metric for each one - Any time a characteristic of imitable sites appears that seems to be something that calls for imitation develop a metric to measure this characteristic - Once you have such a metric, check to see how strongly it correlates with your imitability metric over your corpus of websites - Publish your results on your website. Even negative results admit posts in the vein of "people have been telling me that I should X to get more visitors, but when I checked for a correlation between X and the number of visitors of the sort that I'd want, there was none. You can stop suggesting that now" As for my comment about the amount of time spent; you might notice that it refers to the density of the time spent. Had you undertake such a program, you wouldn't describe it as "hundreds of hours over the last several years"; you'd say something more along the lines of "I spent a month trying to churn something like that out back in $year". My actual claim is quite falsifiable, in that you could identify half a dozen metrics that correlate with increased traffic. adjust your site to score better on those metrics (testing that the adjustments were actually made correctly by measuring your own site, of course), and fail to see a measurable/significant increase in the desirable traffic metric. kiwano - 2010-10-23 12:30

<I>you could identify half a dozen metrics that correlate with increased traffic. adjust your site to score better on those metrics (testing that the adjustments were actually made correctly by measuring your own site, of course), and fail to see a measurable/significant increase in the desirable traffic metric.</I> And what if I already did that and observed that, many times? Matt - 2010-10-23 13:02

Well you certainly haven't published any information about the metrics used, strength of the correlation, score of your site with these metrics before adapting it, and score of your site after adapting it. Where's your data? kiwano - 2010-10-23 13:17

I'm not willing to share the data, at least not today and with you. Matt - 2010-10-23 13:27

I Love Typography is one of my faves. If you really want to get into it have a look at the Typophile forums: http://www.typophile.com/ Being a Moon-Sun Aquarian I fortunately don't care if I am widely read. Axel - 2010-10-23 15:02