[osol-mktg] OpenSolaris Download Statistics ?
Stephen Harpster
Stephen.Harpster at Sun.COM
Tue Jul 12 11:01:46 PDT 2005
Why do it then? This just seems ripe for a PR disaster. I know people
are asking for numbers, but explain to them that there's no way to track
them and let it go at that. We don't have to do everything people ask
for, you know.
Claire Giordano wrote:
>Simon and I did talk about the number of downloads just before the break, and...
>
>We agree it makes sense to turn to the community to ask for ideas/suggestions
>about how to track the number of OpenSolaris downloads (of source or binaries or
>both?) given the use of BitTorrent.
>
>We're both ok sharing the numbers we do have, but we don't want to share them in
>a way in which people will just take the numbers and run if they scan the email
>quickly. Providing the current numbers (which could be way low) in a easily
>scannable table or list seems like a recipe for misinterpretation. So, we want
>the numbers we have to be embedded in a question. This needs work, but here's a
>feel for what we're thinking of:
>...............................................................................
>People have been asking about the OpenSolaris download numbers (and about other
>metrics as well.) Patrick Finch will be following up about metrics for
>OpenSolaris in general, and we plan to provide useful metrics as part of the
>website moving forward. In the meantime, we wanted some input about how to
>track downloads given that we're using BitTorrent, a P2P system that makes it
>more difficult to track how many people have actually downloaded the source and
>binaries. We know that there have been xxx downloads from SDLC and yyy
>downloads from Genunix.org, but for BitTorrent the numbers are more spotty --
>currently we have detected zzz. This means that the current total is AAA, but
>of course, that could be way low, and perhaps the real number is significantly
>higher.
>
>Suggestions????
>...............................................................................
>
>Patrick - can you pls regroup with Derek and figure out how to proceed? We do
>need to follow up with the community on this, somehow, this week...
>
>Thanks,
>Claire
>
>patrick finch wrote:
>
>
>
>>>BTW, we learned today in the bloggers meeting is that we should NOT
>>>
>>>
>>publish number of "hits" to a website but rather should publish the
>>number of "page views". My mistake on my Highlights report last week.
>> I got the number right,
>>
>>
>>
>>>but I should have referred to them as "Page Views".
>>>
>>>
>>That's my fault - the numbers I gave you were "page views"...but I
>>didn't relate to that name.
>>
>>I don't know what the outcome of your discussion with Simon on the
>>downloads number is.
>>
>>From my perspective, think that the question is motivated by curiousity
>>and it is not driving any behaviour other than giving the audience a
>>sense of momentum. Or to put it another way, the answer to the question
>>"do you know how many downloads there have been?" is "no", but
>>explaining the reasons for this is presumably of interest to the audeince.
>>
>>For those reasons I like Derek's suggestion below, but I would like
>>better to understand the BT figure: how is it arrived at? I'd want to
>>call it an estimate if that is what it is, and estimate for all BT
>>downloads, not just those hosted by Sun, stating assumptions and
>>reasoning. As long as your thought process has integrity, I think that
>>this is entirely justifiable. (to give an idea of what I mean, could we
>>identify if we simulate a BT download how many times we touch a
>>Sun-served file? If we could, we could extrapolate from the Sun-served
>>number, etc.)
>>
>>Derek -we'll talk early after the break, let's cover this too?
>>
>>regards
>>
>>Patrick
>>
>>
>>
>>Claire Giordano wrote:
>>
>>
>>
>>>Simon and I plan to talk about this a bit this afternoon. Please hold off on
>>>publishing the numbers UNTIL he and I have touched base.
>>>
>>>However, if anyone wants to go ahead and post the question - about how to track
>>>BitTorrent downloads - that would be great.
>>>
>>>BTW, we learned today in the bloggers meeting is that we should NOT publish
>>>number of "hits" to a website but rather should publish the number of "page
>>>views". My mistake on my Highlights report last week. I got the number right,
>>>but I should have referred to them as "Page Views".
>>>
>>>Claire
>>>
>>>Derek Cicero wrote:
>>>
>>>
>>>
>>>
>>>
>>>>Simon is correct. We can know how many people we have serviced, but the
>>>>appeal of BitTorrent is that people may not get the file from us, and
>>>>therefore can't be tracked by us. The community undestands this and I
>>>>think we provide the numbers that we have.
>>>>
>>>>I would propose posting this to the 'discuss' list:
>>>>
>>>>The number of downloads for the OpenSolaris Source, as of 06/30/2005:
>>>>
>>>>SDLC: 4,605
>>>>GENUNIX: 1,585
>>>>BitTorrent: 1,000+ *
>>>>Total: Over 7,000 files
>>>>
>>>>*Files served by Sun
>>>>
>>>>Derek
>>>>
>>>>Sara Dornsife wrote:
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>>That needs to be explained to the community then. We can't just not
>>>>>tell people the download stats. Especially after being asked
>>>>>specifically for them. In fact, maybe we should ask the community to
>>>>>help us solve our counting issue? Would that work?
>>>>>Sara
>>>>>
>>>>>
>>>>>Claire Giordano wrote:
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>To be clear - I do not want to publish download numbers at all,
>>>>>>unless the numbers are complete.
>>>>>>
>>>>>>Unfortunately, if we publish a subset of the numbers and state they
>>>>>>are a subset, most people will not remember the caveat and yet will
>>>>>>remember the numbers, or their "small|medium|large" perception of the
>>>>>>numbers. It won't be accurate and it could create a misperception
>>>>>>about the momentum and interest. I don't see what's in it for us.
>>>>>>
>>>>>>Of course, if we could track BT and the mirrors, that would be great.
>>>>>>
>>>>>>Claire
>>>>>>
>>>>>>Karyn Ritter wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>>Claire Giordano wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>I don't want to distribute download numbers that are potentially
>>>>>>>>very misleading. I'd rather not share download #s all if they're
>>>>>>>>not complete and if we don't know if they represent 10% or 90% of
>>>>>>>>the actual number of downloads.
>>>>>>>>
>>>>>>>>Derek - do you agree with Simon's assessment below? Since you were
>>>>>>>>proposing to share download numbers (and had provided them last
>>>>>>>>week to Laura), I'm wondering if you know of some magical way to
>>>>>>>>track downloads from mirrors and BT? Do you?
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>Just my $.02... I agree that we should only publish the numbers we
>>>>>>>can track and state that they are a subset. Should we also provide a
>>>>>>>pointer to the download numbers the mirrors publish?
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>p.s. Just saw email from Tom Gryder complimenting on a great
>>>>>>>>report about OpenSolaris download statistics. Who provided that,
>>>>>>>>and when? I just searched my mail for downloads in the topic and
>>>>>>>>glanced thru emails from Derek and Karyn and didn't find it. I'm
>>>>>>>>sure it's there, but would be much obliged if you could forward a
>>>>>>>>copy? And, we need to figure out who in engineering Patrick should
>>>>>>>>confer with on stats - don't want our teams to be sending out
>>>>>>>>conflicting info.
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>We directed him to get his download statistics from SDLC, and I'm
>>>>>>>sure that is what he did.
>>>>>>>
>>>>>>>- Karyn
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
--
Stephen Harpster
Director, Open Source Software
Sun Microsystems, Inc.
More information about the opensolaris-mktg
mailing list