Re: [Wikitech-l] [Analytics] Unique Devices data available on API

[Nuria Ruiz]

>Sure, and browses which rejects or periodically delete cookies will be
counted multiple times
Actually no, they will be counted only once towards the period in which we
are counting them (daily or monthly) as long as the IP of the device is the
same. Not multiple times. In mobile IPs are shared due to NAT-ing and thus,
this method undercounts.
This is explained in detail here:
https://wikitech.wikimedia.org/wiki/Analytics/Unique_Devices/Last_access_solution#Nocookie_Offset



>a device might use multiple browsers (especially the embedded browsers on
mobile devices)
This is correct, multiple browsers on one device will be counted multiple
times. This is less of a common browsing pattern than using the same method
of access every time.

On Tue, Apr 19, 2016 at 3:12 PM, Gergo Tisza  wrote:

> On Tue, Apr 19, 2016 at 11:52 PM, Erik Zachte 
> wrote:
>
>> Like Nuria said: this is unique devices, not unique people. Many people
>> in the Global North use more than one device to access Wikipedia (desktop,
>> tablet, phone).
>>
>
> Sure, and browses which rejects or periodically delete cookies will be
> counted multiple times, and a device might use multiple browsers
> (especially the embedded browsers on mobile devices)... even taking all
> that into account, a factor of four seems like a large disparity between
> this data and the audience panel numbers.
>
> ___
> Analytics mailing list
> analyt...@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/analytics
>
>
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Re: [Wikitech-l] [Analytics] Unique Devices data available on API

[Gergo Tisza]

On Tue, Apr 19, 2016 at 11:52 PM, Erik Zachte  wrote:

> Like Nuria said: this is unique devices, not unique people. Many people in
> the Global North use more than one device to access Wikipedia (desktop,
> tablet, phone).
>

Sure, and browses which rejects or periodically delete cookies will be
counted multiple times, and a device might use multiple browsers
(especially the embedded browsers on mobile devices)... even taking all
that into account, a factor of four seems like a large disparity between
this data and the audience panel numbers.
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Re: [Wikitech-l] [Analytics] Unique Devices data available on API

[Erik Zachte]

Like Nuria said: this is unique devices, not unique people. Many people in the 
Global North use more than one device to access Wikipedia (desktop, tablet, 
phone). 

Also I'd like to add a caveat: any long term change will have to factor in that 
this ratio of devices owned per user isn't fixed over time.

Erik Zachte

-Original Message-
From: Wikitech-l [mailto:wikitech-l-boun...@lists.wikimedia.org] On Behalf Of 
Gergo Tisza
Sent: Tuesday, April 19, 2016 22:54
To: A mailing list for the Analytics Team at WMF and everybody who has an 
interest in Wikipedia and analytics.
Cc: Wikimedia developers; Research into Wikimedia content and communities
Subject: Re: [Wikitech-l] [Analytics] Unique Devices data available on API

Very interesting, thank you!

Do you have any estimate of how much this overcounts? I checked the monthly 
uniques for huwiki 
,
and it's about 5.8 million, which is a bit higher than the total number of 
internet users in Hungary (estimated to 5.2 million). This Gemius analyis 
 from a 
year ago claims a 30% reach for Wikipedia, which would be about 1.5 million. 
They use a software panel (a demographically representative group of volunteers 
who installed tracking software) so they might be inaccurate (and they only 
count traffic originating from Hungary I think) but probably not by a factor of 
four.


On Tue, Apr 19, 2016 at 9:17 PM, Nuria Ruiz  wrote:

> Hello!
>
> The analytics team is happy to announce that the Unique Devices data 
> is now available to be queried programmatically via an API.
>
> This means that getting the daily number of unique devices [1] for 
> English Wikipedia for the month of February 2016, for all sites 
> (desktop and
> mobile) is as easy as launching this query:
>
>
> https://wikimedia.org/api/rest_v1/metrics/unique-devices/en.wikipedia.
> org/all-sites/daily/20160201/20160229
>
> You can get started by taking a look at our docs:
> https://wikitech.wikimedia.org/wiki/Analytics/Unique_Devices#Quick_Sta
> rt
>
> If you are not familiar with the Unique Devices data the main thing 
> you need to know is that is a good proxy metric to measure Unique 
> Users, more info below.
>
> Since 2009, the Wikimedia Foundation used comScore to report data 
> about unique web visitors.  In January 2016, however, we decided to 
> stop reporting comScore numbers [2] because of certain limitations in 
> the methodology, these limitations translated into misreported mobile 
> usage. We are now ready to replace comscore numbers with the Unique Devices 
> Dataset .
> While unique devices does not equal unique visitors, it is a good 
> proxy for that metric, meaning that a major increase in the number of 
> unique devices is likely to come from an increase in distinct users. 
> We understand that counting uniques raises fairly big privacy concerns 
> and we use a very private conscious way to count unique devices, it 
> does not include any cookie by which your browser history can be tracked [3].
>
>
> [1] https://meta.wikimedia.org/wiki/Research:Unique_Devices
> [2] [https://meta.wikimedia.org/wiki/ComScore/Announcement
> [3]
> https://meta.wikimedia.org/wiki/Research:Unique_Devices#How_do_we_coun
> t_unique_
> devices.3F
>
>
> ___
> Analytics mailing list
> analyt...@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/analytics
>
>
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l


___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Re: [Wikitech-l] [Analytics] Unique Devices data available on API

[Kevin Leduc]

Here's another useful link to a form that helps you construct the API call:
https://wikimedia.org/api/rest_v1/?doc#!/Unique_devices_data/get_metrics_unique_devices_project_access_site_granularity_start_end


On Tue, Apr 19, 2016 at 12:17 PM, Nuria Ruiz  wrote:

> Hello!
>
> The analytics team is happy to announce that the Unique Devices data is
> now available to be queried programmatically via an API.
>
> This means that getting the daily number of unique devices [1] for English
> Wikipedia for the month of February 2016, for all sites (desktop and
> mobile) is as easy as launching this query:
>
>
> https://wikimedia.org/api/rest_v1/metrics/unique-devices/en.wikipedia.org/all-sites/daily/20160201/20160229
>
> You can get started by taking a look at our docs:
> https://wikitech.wikimedia.org/wiki/Analytics/Unique_Devices#Quick_Start
>
> If you are not familiar with the Unique Devices data the main thing you
> need to know is that
> is a good proxy metric to measure Unique Users, more info below.
>
> Since 2009, the Wikimedia Foundation used comScore to report data about
> unique web visitors.  In January 2016, however, we decided to stop
> reporting comScore numbers [2] because of certain limitations in the
> methodology, these limitations translated into misreported mobile usage. We
> are now ready to replace comscore numbers with the Unique Devices Dataset .
> While unique devices does not equal unique visitors, it is a good proxy for
> that metric, meaning that a major increase in the number of unique devices
> is likely to come from an increase in distinct users. We understand that
> counting uniques raises fairly big privacy concerns and we use a very
> private conscious way to count unique devices, it does not include any
> cookie by which your browser history can be tracked [3].
>
>
> [1] https://meta.wikimedia.org/wiki/Research:Unique_Devices
> [2] [https://meta.wikimedia.org/wiki/ComScore/Announcement
> [3]
> https://meta.wikimedia.org/wiki/Research:Unique_Devices#How_do_we_count_unique_
> devices.3F
>
>
> ___
> Analytics mailing list
> analyt...@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/analytics
>
>
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Re: [Wikitech-l] [Analytics] Unique Devices data available on API

[Gergo Tisza]

Very interesting, thank you!

Do you have any estimate of how much this overcounts? I checked the monthly
uniques for huwiki
,
and it's about 5.8 million, which is a bit higher than the total number of
internet users in Hungary (estimated to 5.2 million). This Gemius analyis
 from
a year ago claims a 30% reach for Wikipedia, which would be about 1.5
million. They use a software panel (a demographically representative group
of volunteers who installed tracking software) so they might be inaccurate
(and they only count traffic originating from Hungary I think) but probably
not by a factor of four.


On Tue, Apr 19, 2016 at 9:17 PM, Nuria Ruiz  wrote:

> Hello!
>
> The analytics team is happy to announce that the Unique Devices data is
> now available to be queried programmatically via an API.
>
> This means that getting the daily number of unique devices [1] for English
> Wikipedia for the month of February 2016, for all sites (desktop and
> mobile) is as easy as launching this query:
>
>
> https://wikimedia.org/api/rest_v1/metrics/unique-devices/en.wikipedia.org/all-sites/daily/20160201/20160229
>
> You can get started by taking a look at our docs:
> https://wikitech.wikimedia.org/wiki/Analytics/Unique_Devices#Quick_Start
>
> If you are not familiar with the Unique Devices data the main thing you
> need to know is that
> is a good proxy metric to measure Unique Users, more info below.
>
> Since 2009, the Wikimedia Foundation used comScore to report data about
> unique web visitors.  In January 2016, however, we decided to stop
> reporting comScore numbers [2] because of certain limitations in the
> methodology, these limitations translated into misreported mobile usage. We
> are now ready to replace comscore numbers with the Unique Devices Dataset .
> While unique devices does not equal unique visitors, it is a good proxy for
> that metric, meaning that a major increase in the number of unique devices
> is likely to come from an increase in distinct users. We understand that
> counting uniques raises fairly big privacy concerns and we use a very
> private conscious way to count unique devices, it does not include any
> cookie by which your browser history can be tracked [3].
>
>
> [1] https://meta.wikimedia.org/wiki/Research:Unique_Devices
> [2] [https://meta.wikimedia.org/wiki/ComScore/Announcement
> [3]
> https://meta.wikimedia.org/wiki/Research:Unique_Devices#How_do_we_count_unique_
> devices.3F
>
>
> ___
> Analytics mailing list
> analyt...@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/analytics
>
>
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Re: [Wikitech-l] Unique Devices data available on API

[Andrew Otto]

Ha, perhaps Nuria’s quote should read:

> it does not include any cookie by which your brows*ING* history can be
tracked [3].

s/browser/browsing/

On Tue, Apr 19, 2016 at 4:38 PM, bawolff  wrote:

> > it does not include any
> > cookie by which your browser history can be tracked [3].
>
> Umm, it involves a cookie, which tracks whether you have previously
> browsed the site. While I applaud the analytics team in using such a
> privacy friendly method, I'm not sure its an entirely truthful
> statement to say that it does not involve a cookie in which your
> browser history can be tracked. The date you've visited previously
> sounds like a part of your browser history to me.
>
> To be clear, this is not meant to be a criticism. I think the approach
> that is being taken is really great.
>
> --
> -bawolff
>
> ___
> Wikitech-l mailing list
> Wikitech-l@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/wikitech-l
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Re: [Wikitech-l] Unique Devices data available on API

[bawolff]

> it does not include any
> cookie by which your browser history can be tracked [3].

Umm, it involves a cookie, which tracks whether you have previously
browsed the site. While I applaud the analytics team in using such a
privacy friendly method, I'm not sure its an entirely truthful
statement to say that it does not involve a cookie in which your
browser history can be tracked. The date you've visited previously
sounds like a part of your browser history to me.

To be clear, this is not meant to be a criticism. I think the approach
that is being taken is really great.

--
-bawolff

___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

[Wikitech-l] Unique Devices data available on API

[Nuria Ruiz]

Hello!

The analytics team is happy to announce that the Unique Devices data is now
available to be queried programmatically via an API.

This means that getting the daily number of unique devices [1] for English
Wikipedia for the month of February 2016, for all sites (desktop and
mobile) is as easy as launching this query:

https://wikimedia.org/api/rest_v1/metrics/unique-devices/en.wikipedia.org/all-sites/daily/20160201/20160229

You can get started by taking a look at our docs:
https://wikitech.wikimedia.org/wiki/Analytics/Unique_Devices#Quick_Start

If you are not familiar with the Unique Devices data the main thing you
need to know is that
is a good proxy metric to measure Unique Users, more info below.

Since 2009, the Wikimedia Foundation used comScore to report data about
unique web visitors.  In January 2016, however, we decided to stop
reporting comScore numbers [2] because of certain limitations in the
methodology, these limitations translated into misreported mobile usage. We
are now ready to replace comscore numbers with the Unique Devices Dataset .
While unique devices does not equal unique visitors, it is a good proxy for
that metric, meaning that a major increase in the number of unique devices
is likely to come from an increase in distinct users. We understand that
counting uniques raises fairly big privacy concerns and we use a very
private conscious way to count unique devices, it does not include any
cookie by which your browser history can be tracked [3].


[1] https://meta.wikimedia.org/wiki/Research:Unique_Devices
[2] [https://meta.wikimedia.org/wiki/ComScore/Announcement
[3]
https://meta.wikimedia.org/wiki/Research:Unique_Devices#How_do_we_count_unique_
devices.3F
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

[Wikitech-l] CentralAuth timing security

[Chad]

Hi all,

We've just landed a patch to the master branch of CentralAuth[0] that
helps avoid timing attacks with token comparisons. The Phabricator
task is T125290[1].

If you are are using CentralAuth (which is probably not many of you), please
update your installations with this patch.

-Chad

[0] https://gerrit.wikimedia.org/r/#/c/284237/
[1] https://phabricator.wikimedia.org/T125290
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

[Wikitech-l] Help review/test AuthManager

[Gergo Tisza]

Hi all,

AuthManager [1][2] is getting close to being completed. It will replace
AuthPlugin as the interface for authentication extensions, and involves a
rewrite of most authentication code in core (password handling in User,
login/signup/password change/password reset APIs and special pages). It
will be more flexible than AuthPlugin, allowing multiple authentication
extensions to live together, providing a proper interface for
redirect-based authentication, and much more.

If you would like to make sure that some feature you are interested in will
be supported, or would just like to help along, this is the right time! The
gist of AuthManager is in these patches:
* https://gerrit.wikimedia.org/r/#/c/195297 Add AuthManager
* https://gerrit.wikimedia.org/r/#/c/240052 Use AuthManager on special pages
* https://gerrit.wikimedia.org/r/#/c/265201 API changes for AuthManager
* https://gerrit.wikimedia.org/r/#/c/282202 AuthManager release notes

If you would like to try it out, there is a test installation at
http://authmanager.wmflabs.org/
All feedback welcome, code reviews especially so.


[1]
https://www.mediawiki.org/wiki/User:Anomie/SessionManager_and_AuthManager
[2] https://www.mediawiki.org/wiki/Requests_for_comment/AuthManager
___
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l