At some point, I suspect those leader boards are never going to change. I wonder what it would be like to hold a "response drive," like some sort of Blogging Marathon akin to PBS fund raising. Imagine the pageantry! :p
See, now you're just asking for trouble. Back when this meme was new, whenever someone posted it would people flood it with responses (sometimes posting each individual word of their response in a separate comment) just to improve their standings. The first time I posted a chart like this, several years ago, I got hundreds of responses, mostly from just a few people trying to jostle each other for position.
Technically, kithan and mst3kforall are both tied for tenth place. That's why I don't do this meme more often; the last time I did it I manually combined the numbers for people who had posted under more than one name, and gave equal weight to the people who had tied. I also dropped the deleted accounts and my own accounts, and resized the horizontal lines to make them properly proportionate. As you might imagine, that was a huge pain in the ass.
That's still an option. The actual processing is not difficult, it's just that there's a large quantity of data to go through. Of course, my version would have all sorts of extra bells and whistles: It wouldn't be limited to just the top 100 posters; it would process all comments, and then let you decided whether or not you wanted to stick with a subset or post the full list. It would allow you to select which people to exclude (yourself, your sock puppets, anonymous posts, deleted accounts) and which people to combine (other people's sock puppets). It would have a few different options for how the results are displayed (in particular, how to handle ties). Optimally, it would not only provide HTML output (for posting in LJ) but would also save the results in a comma- or tab-delimited text file - this would allow you to import the data into something like Excel if you wanted to do further processing, make charts, or whatnot; it would also allow the program to automatically look for changes between runs (and tell you who went up, who
( ... )
From the FAQ: Why is the report limited to 100 commenters? LJ entries are limited to a finite number of characters. Any charts exceeding ~140 commenters will exceed the alloted number of characters, resulting in a truncated entry. Truncated entries have the potential to corrupt the remainder of your journal page (and your friends page). Therefore, the limit was set to a nice round logical number of 100. Note that all comments are counted, but only the top 100 commenters are displayed.
Something to keep in mind (I actually did write a post with so much HTML once that it overflowed LJ's limit, but I can't remember what the limit is); but if the chart is too big all you have to do is drop the red bar graphs. I'd still like to be able to see a list of all of the people who've ever commented in my journal. The bars in the chart don't seem proportional to the numbers! Why? They are not proportional linearly, but they are proportional logarithmically. This design decision was made for aesthetic reasons; using a linear scale will result in a very
( ... )
There are a lot of variables there. How much I post, how many folks comment...etc. I mean, one person on there could change radically in one day, they just have to comment a whole bunch.
That's true also. There are several friends journals where I have gone back to the beginning and read through. (Although, I haven't done that in a while.) But I rarely, if ever, commented on those old posts. It always felt weird to do so...
Comments 21
Reply
See, now you're just asking for trouble. Back when this meme was new, whenever someone posted it would people flood it with responses (sometimes posting each individual word of their response in a separate comment) just to improve their standings. The first time I posted a chart like this, several years ago, I got hundreds of responses, mostly from just a few people trying to jostle each other for position.
Reply
Reply
Reply
Technically, kithan and mst3kforall are both tied for tenth place. That's why I don't do this meme more often; the last time I did it I manually combined the numbers for people who had posted under more than one name, and gave equal weight to the people who had tied. I also dropped the deleted accounts and my own accounts, and resized the horizontal lines to make them properly proportionate. As you might imagine, that was a huge pain in the ass.
Reply
Reply
That's still an option. The actual processing is not difficult, it's just that there's a large quantity of data to go through. Of course, my version would have all sorts of extra bells and whistles: It wouldn't be limited to just the top 100 posters; it would process all comments, and then let you decided whether or not you wanted to stick with a subset or post the full list. It would allow you to select which people to exclude (yourself, your sock puppets, anonymous posts, deleted accounts) and which people to combine (other people's sock puppets). It would have a few different options for how the results are displayed (in particular, how to handle ties). Optimally, it would not only provide HTML output (for posting in LJ) but would also save the results in a comma- or tab-delimited text file - this would allow you to import the data into something like Excel if you wanted to do further processing, make charts, or whatnot; it would also allow the program to automatically look for changes between runs (and tell you who went up, who ( ... )
Reply
From the FAQ:
Why is the report limited to 100 commenters?
LJ entries are limited to a finite number of characters. Any charts exceeding ~140 commenters will exceed the alloted number of characters, resulting in a truncated entry. Truncated entries have the potential to corrupt the remainder of your journal page (and your friends page). Therefore, the limit was set to a nice round logical number of 100. Note that all comments are counted, but only the top 100 commenters are displayed.
Something to keep in mind (I actually did write a post with so much HTML once that it overflowed LJ's limit, but I can't remember what the limit is); but if the chart is too big all you have to do is drop the red bar graphs. I'd still like to be able to see a list of all of the people who've ever commented in my journal.
The bars in the chart don't seem proportional to the numbers! Why?
They are not proportional linearly, but they are proportional logarithmically. This design decision was made for aesthetic reasons; using a linear scale will result in a very ( ... )
Reply
:)
Need improvement though.
Reply
Reply
Reply
Reply
Reply
Once Daniel is born, I suspect my comment/posting frequency will drop, at least temporarily.
Reply
Reply
Reply
Reply
Leave a comment