UploadWizard roll-out

[Neil Kandalgaonkar]

It's been a long time coming but we're getting close to being able to
release UploadWizard to a broader spectrum of users.

If you are unfamiliar with the project, the idea is to greatly improve
the usability of the uploading experience. We haven't been public for a
while but we are working hard on fixing all the bugs you've reported.
You should see more regular updates from us over this month.

We're pretty sure this will be a giant improvement over the current
experience, although we're aware that there's a degree of skepticism out
there about this, and some questions about whether this will increase
copyvio or have other bad effects. This is where you all come in.

Questions we're considering:

- How best to roll this out -- it should be phased, but how exactly?

- How to monitor how this changes what gets uploaded, does it increase
copyvio, etc. Obviously if this starts to increase we want to pull back
and reassess.

We already add every image uploaded with UploadWizard to the category
"Uploaded With UploadWizard". What tools exist to survey the copyvio
rates, let's say of one category versus all categories? If this requires
coding, does this sound like a cool project to you?


--
Neil Kandalgaonkar     <[hidden email]>

_______________________________________________
Commons-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/commons-l

[Siebrand Mazeland]

Op 03-03-11 22:25 schreef Neil Kandalgaonkar <[hidden email]>:

>- How best to roll this out -- it should be phased, but how exactly?
Can the roll-out be geo based? This will allow you to slowly grow usage
across different countries, and make it possible to most likely get
feedback from for example "language X" speakers. It would get you across
project communities, too.

Another option would be a language/project based incremental deployment.

>
>- How to monitor how this changes what gets uploaded, does it increase
>copyvio, etc. Obviously if this starts to increase we want to pull back
>and reassess.

You have to take a "0 measurement" now to be able to compare. Define
critical success factors, measure, deploy, measure again. Copyvio may be
hard to measure (additional tools/dev/work required). Survival rate of one
month would for example be an easier indicator.

Soiebrand



_______________________________________________
Commons-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/commons-l

[Neil Kandalgaonkar]

On 3/3/11 2:12 PM, Siebrand Mazeland wrote:
> Op 03-03-11 22:25 schreef Neil Kandalgaonkar<[hidden email]>:
>
>> - How best to roll this out ?
> Can the roll-out be geo based? This will allow you to slowly grow usage
> across different countries, and make it possible to most likely get
> feedback from for example "language X" speakers. It would get you across
> project communities, too.

I'm not sure I understand. What's the advantage of this? Will Germans
use it very differently from Americans? Does a publicized "now available
in your country" help create the nucleus of a community?


> Another option would be a language/project based incremental deployment.

Well, this is not slated for any Wikipedia (yet). We're focusing on
Commons, which is in theory multilingual.


>> - How to monitor how this changes what gets uploaded, does it increase
>> copyvio, etc. Obviously if this starts to increase we want to pull back
>> and reassess.
>
> You have to take a "0 measurement" now to be able to compare. Define
> critical success factors, measure, deploy, measure again. Copyvio may be
> hard to measure (additional tools/dev/work required). Survival rate of one
> month would for example be an easier indicator.

Our real critical success factor is increasing participation, the most
obvious would be a simple increase in volume of work submitted that
survives after a month.

One worry that people have voiced is that this will increased
*undetected* copyvio. The theory is that people who copyvio are apt to
be dissuaded by Commons' "traps" for the unwary. I am personally really
skeptical of that theory, but I recognize that we may be making it
easier to get through the process for copyvio people too.

Perhaps we need to take a sampling of photos and ... try harder to
determine copyright? Maybe contact the uploaders in question, if
possible, and just ask them what they thought they were doing? (I'm
assuming most copyvio is done in good faith).


--
Neil Kandalgaonkar     <[hidden email]>

_______________________________________________
Commons-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/commons-l

[Gerard Meijssen-3]

Hoi,
I am pretty sure that when we announce that the easy upload facility becomes available in Indian languages once the software is localised there will be a friendly competition what language gets the new functionality first.

<grin> when you add a competition that showers recognition on the first Indian photographer who uploads a new picture and gets it featured on Commons from the date of the availability in his or her language, we will not only gain many localisations but also potentially many photos. Many great photos even :) </grin>
Thanks,
       GerardM

On 3 March 2011 23:29, Neil Kandalgaonkar <[hidden email]> wrote:

On 3/3/11 2:12 PM, Siebrand Mazeland wrote:
> Op 03-03-11 22:25 schreef Neil Kandalgaonkar<[hidden email]>:
>

>> - How best to roll this out ?

> Can the roll-out be geo based? This will allow you to slowly grow usage
> across different countries, and make it possible to most likely get
> feedback from for example "language X" speakers. It would get you across
> project communities, too.

I'm not sure I understand. What's the advantage of this? Will Germans
use it very differently from Americans? Does a publicized "now available
in your country" help create the nucleus of a community?

> Another option would be a language/project based incremental deployment.

Well, this is not slated for any Wikipedia (yet). We're focusing on
Commons, which is in theory multilingual.

>> - How to monitor how this changes what gets uploaded, does it increase
>> copyvio, etc. Obviously if this starts to increase we want to pull back
>> and reassess.
>
> You have to take a "0 measurement" now to be able to compare. Define
> critical success factors, measure, deploy, measure again. Copyvio may be
> hard to measure (additional tools/dev/work required). Survival rate of one
> month would for example be an easier indicator.

Our real critical success factor is increasing participation, the most
obvious would be a simple increase in volume of work submitted that
survives after a month.

One worry that people have voiced is that this will increased
*undetected* copyvio. The theory is that people who copyvio are apt to
be dissuaded by Commons' "traps" for the unwary. I am personally really
skeptical of that theory, but I recognize that we may be making it
easier to get through the process for copyvio people too.

Perhaps we need to take a sampling of photos and ... try harder to
determine copyright? Maybe contact the uploaders in question, if
possible, and just ask them what they thought they were doing? (I'm
assuming most copyvio is done in good faith).


--

Neil Kandalgaonkar     <[hidden email]>

_______________________________________________
Commons-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/commons-l

_______________________________________________
Commons-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/commons-l

[Platonides]

Neil Kandalgaonkar wrote:

>>> - How to monitor how this changes what gets uploaded, does it increase
>>> copyvio, etc. Obviously if this starts to increase we want to pull back
>>> and reassess.
>>
>> You have to take a "0 measurement" now to be able to compare. Define
>> critical success factors, measure, deploy, measure again. Copyvio may be
>> hard to measure (additional tools/dev/work required). Survival rate of one
>> month would for example be an easier indicator.
>
> Our real critical success factor is increasing participation, the most
> obvious would be a simple increase in volume of work submitted that
> survives after a month.
>
> One worry that people have voiced is that this will increased
> *undetected* copyvio. The theory is that people who copyvio are apt to
> be dissuaded by Commons' "traps" for the unwary. I am personally really
> skeptical of that theory, but I recognize that we may be making it
> easier to get through the process for copyvio people too.

Same upload #, less copyvios => Undetected copyvios
Less uploads => UploadWizard failure
More uploads, same copyvio rate => Probably good
Other numbers => ???
The problem when the number of uploads increase is that you can hardly
determine if that came from the good or bad pools. The bad percentage is
likely to be higher on the difference group than in the previous one.
But it could also be that more people is now able to do things
correctly, so you can't assume that all those are copyvios either.


> Perhaps we need to take a sampling of photos and ... try harder to
> determine copyright? Maybe contact the uploaders in question, if
> possible, and just ask them what they thought they were doing? (I'm
> assuming most copyvio is done in good faith).

I haven't done such reviewing recently, but there's usually no feedback
to messages. "Workaround the software to upload the image (no matter
that it says it is not allowed)" might be a good description.



_______________________________________________
Commons-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/commons-l