[Hpr] We should publish the HPR downloads stats (in some form)
ken at fallon.ie
Tue Jun 3 06:00:49 PDT 2014
-----BEGIN PGP SIGNED MESSAGE-----
On 2014-06-02 18:31, epicanis+hpr at dogphilosophy.net wrote:
> On Monday, June 02, 2014 01:38:26 PM Ken Fallon wrote:
>> Hi All,
>> Should we include the number of downloads an episode gets on the website ?
>> Some points to note:
>> - The logs are generated every month so it would be from month to month.
>> - hits are defined as 1 download per show per ip per day.
>> We don't have any logs before Sep-2010 so do we
>> a) ignore shows before that date even though they have been downloaded
>> since then ?
>> b) calculate some guesstimate for a base and add the downloads we do
>> know about.
>> My own main objection to doing this is that downloads are not a measure
>> of worth of the show.
> I would suggest yes, though that's partly because I tend to always lean
> towards having more information more readily available for just about any
> context anyway. I'd even be interested in seeing both "downloaded" and
> "listened from the web page" if it's possible to meaningfully estimate this
> distinction. (I would assume both would show up in a total count of the log
> anyway, but I'm not sure how feasible it would be to track those separately
> unless we can assume a useful majority of downloaders send a "referer"
What benefit would this provide ?
> If not directly on the website, I would be very interested in having some way
> to examine that information myself (i.e. if we don't want to put the number on
> each episode's page, it would be nice to at least have downloads-per-episode
> data updated regularly and posted somewhere for download and examination).
The monthly report is kept on the ftp server
That list is based on show/ip/day which results in a total download of
all shows as 3,499,611. If we base that on episode GET's the number is
6,626,895. If we base it on HITS the number jumps to 42,277,625 hits.
> The more detailed, the better (it would be interesting to see if some shows
> have a "spike" in downloads sometime after release due to being mentioned
> somewhere else, for example).
> I'd hypothesize that the main drivers of initial downloads would be the
> apparent subject (as expressed by the title), followed by the show notes.
> ("Show Quality" can only affect downloads from referrals after release, since
> you obviously have to download it to find out what the quality is like).
> Having the data would (I further hypothesize) help a lot with "marketing" of
> HPR by helping me and anyone else who's interested to figure out what kinds of
> titles and what kinds of show-note information makes a show more likely to be
> downloaded (plus what kinds of topics attract the most attention.) In the long
> run, this might help "market" HPR better overall, attract an even bigger
> audience, and by extension, more contributors, eventually culminating in
> growing political power and wealth until the HPR elites destroy the world over
> an argument regarding which among them get to rule Australia. (I may be
> getting a little carried away there, but you get the idea.)
It is notoriously difficult to derive this from the download logs.
total hpr1135 2206
total hpr1136 2395
total hpr1137 13160 <-- big spike
total hpr1138 2377
total hpr1139 2203
total hpr1140 2179
total hpr1141 6411 <-- small spike
total hpr1142 2588
total hpr1143 10699 <-- big spike
total hpr1144 2301
total hpr1145 2044
So what's causing the spikes, better than normal shownotes ? Yes for
1137 and 1143 but 1141 only has only one liner. Further analysis shows
that their popularity is because they all got indexed by mp3 share sites.
1137 contains the host "New Age Techno" Hippie
1143 contains the word "Christmas"
1141 contains the word "intro" for the song "Unendlich Intro- Matthias
With so many of our episodes containing the word "intro", it renders any
information gleaned from the logs of questionable usefulness. We could
go down the road of using Google Analytics, but that raises many privacy
> I'll even go a step further and say if the information is made available we
> can get at least one HPR episode out of analysing it (possibly more than one
> if anyone is interested in collaborating).
I don't think this is useful or productive as all the information we
need to improve the quality and reach of our shows is already known.
Namely provide good content, and describe it with complete and detailed
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
-----END PGP SIGNATURE-----
More information about the Hpr