Estimating contributors by looking at wiki accounts?

[Rob Weir <ro...@apache.org>]

I recently saw another open source project claim that they had over
3000 contributors.  They derived this estimate by looking at the
number of user accounts they had in their wiki.

That is quite clever, I thought.  Since we use the same wiki software,
I thought I'd check this metric for us.  Our wiki says we have over
58,000 user accounts.

I know we're doing well, but would it really make sense to claim that
we have over 58,000 contributors?  I don't think so.

I suppose we could look only at accounts where the person has actually
contributed edits, or even recent edits. (MediaWiki is a well-known
target of registration spam).  Although the other project did not seem
to filter out inactive or unused accounts, I think the metrics are
meaningless unless we do that.

What do you think?  Or do we even care?

-Rob

Re: Estimating contributors by looking at wiki accounts?

["Marcus (OOo)" <ma...@wtnet.de>]

Am 10/19/2012 05:28 PM, schrieb Rob Weir:
> I recently saw another open source project claim that they had over
> 3000 contributors.  They derived this estimate by looking at the
> number of user accounts they had in their wiki.
>
> That is quite clever, I thought.  Since we use the same wiki software,
> I thought I'd check this metric for us.  Our wiki says we have over
> 58,000 user accounts.
>
> I know we're doing well, but would it really make sense to claim that
> we have over 58,000 contributors?  I don't think so.
>
> I suppose we could look only at accounts where the person has actually
> contributed edits, or even recent edits. (MediaWiki is a well-known
> target of registration spam).  Although the other project did not seem
> to filter out inactive or unused accounts, I think the metrics are
> meaningless unless we do that.
>
> What do you think?  Or do we even care?

Yes, maybe a good chance to tell others some numbers from our project.

However, the wording of the number is (for some people) the more 
important part. So, this should be double-checked.

That means it doesn't make sense to say "hey, we have 58,000+ 
contributors" but more like "... in the last 12 months we got 
contributions from ~3,000 active people (incl. accounts from SVN, BZ, 
Wiki, MLs, etc.).

Marcus

Re: Estimating contributors by looking at wiki accounts?

[Roberto Galoppini <rg...@geek.net>]

Sent from my iPhone

On 19/ott/2012, at 18:37, Rob Weir <ro...@apache.org> wrote:

> On Fri, Oct 19, 2012 at 12:09 PM, Shane Curcuru <as...@shanecurcuru.org> wrote:
>> +1 all around.  This sounds like it would be more interesting on the
>> ooo-marketing@ list, since it's more about telling the story of "who helps
>> make AOO".  With a project with as many different kinds of end users as AOO
>> has, accurate stats like these would be good, if you want to go generate
>> them.  Plus, I like numbers. 8-)
>> 
>> The most useful thing about generating them would be showing exactly how
>> they're generated, with code (if any), and being very clear - as you suggest
>> - at what the specific numbers mean.  Openness in the way you generate the
>> details is key to ensuring people know exactly what you're measuring.
> 
> I think Mwiki has REST API that gives XML out.  But I'd need to check.
> 
>> - Shane
>> 
>> P.S. Is there already a chart of auto-upgrade "downloads" anywhere? Just
>> curious.
> 
> Not yet.  But it is something I've been trying to figure out.
> SourceForge numbers don't report it, but if you correlate the SF
> numbers with the website numbers from Google Analytics (we send users
> to a special update URL) I think we can estimate it.  But getting
> charts means I need to figure how to automate it on both the GA and SF
> sides.

I'll be happy to help with that, early next week I ll have a look at that.

Roberto

> 
> But note that AOO 3.4.0 shipped with auto-update checking *disabled*
> by default (Doh!).  So the AOO 3.4.0 --> 3.4.1 auto update numbers
> there are going to be modest compared to the numbers from OOo 3.3.0
> users upgrading to AOO 3.4.x.  Of course, many users will hear about
> the new releases via other means.  We see that in the strong AOO 3.4.1
> download numbers.
> 
> -Rob
> 
>> 
>> On 10/19/2012 11:38 AM, jan iversen wrote:
>>> 
>>> I think your idea of filtering out account that actually contributed is a
>>> wise thing, especially because our product has many end-users that want to
>>> be informed but do not contribute.
>>> 
>>> As a developer I do not care, but thinking of some of the ongoing
>>> discussions in other forums (like: nearly nobody contributes to AOO
>>> anymore
>>> because Apache rules makes it far to difficult and restrictive), makes it
>>> worth while to publish a figure on our web, especially a figure saying
>>> e.g.
>>> "during the last year we had xxx active contributors and xx active
>>> committers".
>>> 
>>> jan.
>>> 
>>> 
>>> On 19 October 2012 17:28, Rob Weir <ro...@apache.org> wrote:
>>> 
>>>> I recently saw another open source project claim that they had over
>>>> 3000 contributors.  They derived this estimate by looking at the
>>>> number of user accounts they had in their wiki.
>>>> 
>>>> That is quite clever, I thought.  Since we use the same wiki software,
>>>> I thought I'd check this metric for us.  Our wiki says we have over
>>>> 58,000 user accounts.
>>>> 
>>>> I know we're doing well, but would it really make sense to claim that
>>>> we have over 58,000 contributors?  I don't think so.
>>>> 
>>>> I suppose we could look only at accounts where the person has actually
>>>> contributed edits, or even recent edits. (MediaWiki is a well-known
>>>> target of registration spam).  Although the other project did not seem
>>>> to filter out inactive or unused accounts, I think the metrics are
>>>> meaningless unless we do that.
>>>> 
>>>> What do you think?  Or do we even care?
>>>> 
>>>> -Rob
>> 

-- 
====
This e- mail message is intended only for the named recipient(s) above. It 
may contain confidential and privileged information. If you are not the 
intended recipient you are hereby notified that any dissemination, 
distribution or copying of this e-mail and any attachment(s) is strictly 
prohibited. If you have received this e-mail in error, please immediately 
notify the sender by replying to this e-mail and delete the message and any 
attachment(s) from your system. Thank you.

Re: Estimating contributors by looking at wiki accounts?

[Rob Weir <ro...@apache.org>]

On Fri, Oct 19, 2012 at 12:09 PM, Shane Curcuru <as...@shanecurcuru.org> wrote:
> +1 all around.  This sounds like it would be more interesting on the
> ooo-marketing@ list, since it's more about telling the story of "who helps
> make AOO".  With a project with as many different kinds of end users as AOO
> has, accurate stats like these would be good, if you want to go generate
> them.  Plus, I like numbers. 8-)
>
> The most useful thing about generating them would be showing exactly how
> they're generated, with code (if any), and being very clear - as you suggest
> - at what the specific numbers mean.  Openness in the way you generate the
> details is key to ensuring people know exactly what you're measuring.
>

I think Mwiki has REST API that gives XML out.  But I'd need to check.

> - Shane
>
> P.S. Is there already a chart of auto-upgrade "downloads" anywhere? Just
> curious.
>

Not yet.  But it is something I've been trying to figure out.
SourceForge numbers don't report it, but if you correlate the SF
numbers with the website numbers from Google Analytics (we send users
to a special update URL) I think we can estimate it.  But getting
charts means I need to figure how to automate it on both the GA and SF
sides.

But note that AOO 3.4.0 shipped with auto-update checking *disabled*
by default (Doh!).  So the AOO 3.4.0 --> 3.4.1 auto update numbers
there are going to be modest compared to the numbers from OOo 3.3.0
users upgrading to AOO 3.4.x.  Of course, many users will hear about
the new releases via other means.  We see that in the strong AOO 3.4.1
download numbers.

-Rob

>
> On 10/19/2012 11:38 AM, jan iversen wrote:
>>
>> I think your idea of filtering out account that actually contributed is a
>> wise thing, especially because our product has many end-users that want to
>> be informed but do not contribute.
>>
>> As a developer I do not care, but thinking of some of the ongoing
>> discussions in other forums (like: nearly nobody contributes to AOO
>> anymore
>> because Apache rules makes it far to difficult and restrictive), makes it
>> worth while to publish a figure on our web, especially a figure saying
>> e.g.
>> "during the last year we had xxx active contributors and xx active
>> committers".
>>
>> jan.
>>
>>
>> On 19 October 2012 17:28, Rob Weir <ro...@apache.org> wrote:
>>
>>> I recently saw another open source project claim that they had over
>>> 3000 contributors.  They derived this estimate by looking at the
>>> number of user accounts they had in their wiki.
>>>
>>> That is quite clever, I thought.  Since we use the same wiki software,
>>> I thought I'd check this metric for us.  Our wiki says we have over
>>> 58,000 user accounts.
>>>
>>> I know we're doing well, but would it really make sense to claim that
>>> we have over 58,000 contributors?  I don't think so.
>>>
>>> I suppose we could look only at accounts where the person has actually
>>> contributed edits, or even recent edits. (MediaWiki is a well-known
>>> target of registration spam).  Although the other project did not seem
>>> to filter out inactive or unused accounts, I think the metrics are
>>> meaningless unless we do that.
>>>
>>> What do you think?  Or do we even care?
>>>
>>> -Rob
>>>
>>
>

Re: Estimating contributors by looking at wiki accounts?

[Shane Curcuru <as...@shanecurcuru.org>]

+1 all around.  This sounds like it would be more interesting on the 
ooo-marketing@ list, since it's more about telling the story of "who 
helps make AOO".  With a project with as many different kinds of end 
users as AOO has, accurate stats like these would be good, if you want 
to go generate them.  Plus, I like numbers. 8-)

The most useful thing about generating them would be showing exactly how 
they're generated, with code (if any), and being very clear - as you 
suggest - at what the specific numbers mean.  Openness in the way you 
generate the details is key to ensuring people know exactly what you're 
measuring.

- Shane

P.S. Is there already a chart of auto-upgrade "downloads" anywhere? 
Just curious.

On 10/19/2012 11:38 AM, jan iversen wrote:
> I think your idea of filtering out account that actually contributed is a
> wise thing, especially because our product has many end-users that want to
> be informed but do not contribute.
>
> As a developer I do not care, but thinking of some of the ongoing
> discussions in other forums (like: nearly nobody contributes to AOO anymore
> because Apache rules makes it far to difficult and restrictive), makes it
> worth while to publish a figure on our web, especially a figure saying e.g.
> "during the last year we had xxx active contributors and xx active
> committers".
>
> jan.
>
>
> On 19 October 2012 17:28, Rob Weir <ro...@apache.org> wrote:
>
>> I recently saw another open source project claim that they had over
>> 3000 contributors.  They derived this estimate by looking at the
>> number of user accounts they had in their wiki.
>>
>> That is quite clever, I thought.  Since we use the same wiki software,
>> I thought I'd check this metric for us.  Our wiki says we have over
>> 58,000 user accounts.
>>
>> I know we're doing well, but would it really make sense to claim that
>> we have over 58,000 contributors?  I don't think so.
>>
>> I suppose we could look only at accounts where the person has actually
>> contributed edits, or even recent edits. (MediaWiki is a well-known
>> target of registration spam).  Although the other project did not seem
>> to filter out inactive or unused accounts, I think the metrics are
>> meaningless unless we do that.
>>
>> What do you think?  Or do we even care?
>>
>> -Rob
>>
>

Re: Estimating contributors by looking at wiki accounts?

[jan iversen <ja...@gmail.com>]

I think your idea of filtering out account that actually contributed is a
wise thing, especially because our product has many end-users that want to
be informed but do not contribute.

As a developer I do not care, but thinking of some of the ongoing
discussions in other forums (like: nearly nobody contributes to AOO anymore
because Apache rules makes it far to difficult and restrictive), makes it
worth while to publish a figure on our web, especially a figure saying e.g.
"during the last year we had xxx active contributors and xx active
committers".

jan.


On 19 October 2012 17:28, Rob Weir <ro...@apache.org> wrote:

> I recently saw another open source project claim that they had over
> 3000 contributors.  They derived this estimate by looking at the
> number of user accounts they had in their wiki.
>
> That is quite clever, I thought.  Since we use the same wiki software,
> I thought I'd check this metric for us.  Our wiki says we have over
> 58,000 user accounts.
>
> I know we're doing well, but would it really make sense to claim that
> we have over 58,000 contributors?  I don't think so.
>
> I suppose we could look only at accounts where the person has actually
> contributed edits, or even recent edits. (MediaWiki is a well-known
> target of registration spam).  Although the other project did not seem
> to filter out inactive or unused accounts, I think the metrics are
> meaningless unless we do that.
>
> What do you think?  Or do we even care?
>
> -Rob
>