[Hpr] Hpr Digest, Vol 69, Issue 3

Ken Fallon ken at fallon.ie
Wed Jun 4 08:35:38 PDT 2014


On 2014-06-04 15:35, Carl D Hamann wrote:
> On Tue, Jun 3, 2014 at 9:06 PM, Fifty OneFifty
> <fiftyonefifty at linuxbasement.com> wrote:
>> I think it could help to encourage people to contribute shows, if they
>> knew approximately X people would download them.
> 
> I do see some value in this, but unique per-show stats might not be
> necessary. We could simply take the aggregate numbers we already have
> and post a per-show average. That removes the question of why any
> particular show got more downloads, because we wouldn't be displaying
> that.
> 

I have been asked this question often myself and I addressed this in
detail on episode hpr1351 (http://hackerpublicradio.org/eps.php?id=1351,
see attached image)

In that show I described the tool I wrote and pointed out that the data
is kept updated on the server at
http://hackerpublicradio.org/report.bz2.
Dave Morris mentioned that tracking the subscribers to the RSS feed
might be a better way of tracking downloads.

Therefore the report for May 2014 is as follows

Average RSS Subscriber Check in: ......................2,562
Average Episode Download: ............................ 1,420
Average Daily Download All shows: .................... 2,410
May Downloads: ...................................... 74,778
Average Download * 1519 Episodes:................. 3,664,122
Actual Downloads Since 2010-09-25:................ 3,499,611
Extrapolated Download (1,344 days => 2,347 days ): 6,111,300
Extrapolated Average Episode: ........................ 4,023

So depending on the marketing spin you choose to use you can answer:
- "OUR SHOWS ARE DOWNLOADED OVER SIX MILLION TIMES AND COUNTING"
- "We get about *75 thousand* downloads a month"
- "Most shows get downloaded well over 4,000 times"
- "It might get downloaded 1,400 times but there is no guarantee that
anyone will ever listen to it"

And this is my problem with parsing logs, you get data but you're never
sure if it's correct or not. You can improve your results to remove
spammers, crackers, spiders or bots, and still there will always be some
other thing that throws it off. I consider it unfair to report a show as
popular one month, the discover that it's gotten on some sharing site
only to have to reduce it's count 90% the next. And it's a constant
battle to check and modify the collection algorithm.

So to answer my own question from the original email "Should we include
the number of downloads an episode gets on the website ?" the answer is
simply *NO* as we have no way of reliably calculating that number in a
manner that doesn't require constant maintenance.

There is still a need to provide some number. I would like that to be an
honest number and provide it in the stats page and on the about page.
(http://hackerpublicradio.org/stats.php)

I'm open to suggestions, but from the numbers above the RSS *feels* to
me like it's the most accurate.

As you probably can tell, I find this extremely frustrating. So
apologies to all who have experienced the "Wrath of Ken", on this thread.

-- 
Regards,

Ken Fallon
http://kenfallon.com
http://hackerpublicradio.org/correspondents.php?hostid=30
-------------- next part --------------
A non-text attachment was scrubbed...
Name: communitynews-2013-08.png
Type: image/png
Size: 31581 bytes
Desc: not available
URL: <http://hackerpublicradio.org/pipermail/hpr_hackerpublicradio.org/attachments/20140604/21cec6cc/attachment.png>


More information about the Hpr mailing list