Hey Magnus,
There are a few other factors to consider in addition to those listed. For
example, development cost. Our maps tile service is not compatible with the
iOS app out of the box. This isn't surprising; Apple wants you to use
things like Apple Maps rather than your own solution. Android is, by its
nature, a more open platform, so I am not too surprised that it was easier
to integrate our tile server into the Android app than the iOS app. Sadly,
it's not as simple as just switching over to OSM-based tiles; on the
contrary, it's a significant amount of work.
Now, using our tile service would also have required the iOS app to use the
MapBox SDK. This is the size of all their other third party libraries
combined, significantly increasing app download size. The size of your app
can significantly reduce downloads [1]. Switch a single feature over to a
different set of map tiles, and possibly decreasing downloads of the app,
seems like a dangerous and counterintuitive tradeoff to me.
So the question is, given all this, is switching over the nearby feature to
use OSM-based tiles instead of Apple Maps worth it? In the long run, if
these problems could be solved, I'd say it absolutely is worth it. But, in
the short term, the work would take significant time and effort, and could
actually decrease app usage by decreasing the app download rate; that
tradeoff doesn't seem worth it to me.
Thanks,
Dan
Disclaimers: These are my opinions only. I worked on the apps in the past,
but haven't for two years; my statements about development costs may be
wrong, and the apps folks may well disagree with me about things. I work in
the department responsible for Wikimedia maps, but have only worked on the
team working on maps for a couple of months.
[1]:
On 15 March 2017 at 09:25, Magnus Manske <magnusmanske(a)googlemail.com>
wrote:
Hi Josh, all,
I am not one hell-bent on "FOSS or death"; I tend to use whatever works
best.
That said, the cost-benefit analysis of using Apple Maps seems to boil
down:
* Apple Maps has slightly better rendering (didn't check, but I assume)
* Apple Maps uses less mobile bandwidth
* Apple Maps is not free (as in freedom)
Now, looking at these points:
* Somewhat better quality is not an argument. If it were, we would have
stayed with Britannica, and skipped that whole Wikipedia nonsense.
Wikipedia became better, in part, because people actually used it, saw the
issues, and fixed them. And OSM rendering might be not quite en par with
Apple Maps, it is quite usable, in my experience.
* Less bandwidth usage is not an argument either. I doubt we are talking
about a significant percentage of an average users' data volume here. If
Android users can afford the bandwidth, so can people who buy an iPhone
(source: used to have iPhone).
* The price tag is the "non-freedom". As far as I can tell, this would be
the very first Wikimedia "product" that incorporates non-free technology
and data. It sets a precedence. It also has the potential to poison the
otherwise great relations between the Wikipedia, Wikidata, and OSM
community. It says "OSM is not good enough (at least for Apple users)"
quite plainly. How would we feel if OSM started to remove Wikidata tags and
replace them with Britannica links?
All in all, IMHO, the cost is too high for the (at best) flimsy benefits.
Cheers,
Magnus
On Wed, Mar 15, 2017 at 12:52 AM Joshua Minor <jminor(a)wikimedia.org>
wrote:
Hello,
My name is Josh Minor, and I am the Product Manager for the Wikipedia iOS
app. I wanted to speak to a couple specific issues and misunderstandings
raised by this email thread.
First, please take a look at
https://www.mediawiki.org/wiki/Wikimedia_Apps/Team/iOS/Maps_service which
provides some background on this decision.
Jonatan linked to it, and it
covers several of the concerns raised on the thread and gives our
reasoning. I'd also suggest subscribing to this ticket:
https://phabricator.wikimedia.org/T157763 which Jonatan filed, and where
you can track efforts and issues with replacement maps.
A few clarifying points:
1. The Places tab[1], and its use of Apple’s maps tiles, is not part of
the
articles or article display, it is a navigational
aid to help you find
articles. This doesn’t mean it’s exempt from considerations raised here,
but just want to clarify that this is not about editor created maps in
projects, but rather an app-specific discovery mechanism.
2. The feature doesn’t violate our privacy policy[2] and was reviewed by
Wikimedia Foundation's Legal department before entering beta. The App’s
access to the users’ geolocation to recommend nearby articles, with the
users’ explicit consent, is already part of both apps. The new feature
merely adds a different way to visually view nearby articles - the user
must, as before, still provide explicit consent for the App to access
their
geolocation. Users can always turn on or off the
provision of their
geolocation via their iPhone location settings.
The feature also makes requests to Apple’s map tile servers for display
on
the App. These tiles may or may not be near the
actual location of the
user. It doesn’t involve sending Apple the articles you read or anything
about your Wikipedia usage. Apple has public statements and documentation
to explain[3] how their maps service preserves privacy by using a
randomized and frequently changing device ID to request the maps, by not
tracking users over time, and by not building map usage profiles of
users.
Overall, Apple’s data collection practices are
governed by their privacy
policy [4], which users must agree to order to use their iPhones.
We plan to further expand the explanation in the FAQ/privacy section of
the
https://www.mediawiki.org/wiki/Wikimedia_Apps/Team/iOS/Maps_service page
in
the next day or so.
3. As stated by others on this thread, the issue at hand is the
feasibility
and usability of a libre maps tile server, and
impacts on users and how
it
reflects (or doesn’t) the values of Wikimedians.
The rest of the work on
this feature (such as the time spent on search, visually clustering items
on the map, a list view of nearby landmarks, and the Wikipedia article
pins) will be applicable, independent of the map provider. In fact, I’d
estimate the engineer doing the work spent more time on hacking to try to
make a combination of MapBox and Wikimedia tiles work, than he did/will
on
integrating/removing Apple maps.
4. This feature was announced on the Wikimedia Blog[5], described in an
initial
MediaWiki.org page[6], all work was documented and tracked on
Phabricator (including an initial tech investigation, the request to
remove
Apple Maps during development, and the overall
feature[7]) and then the
decision to push into beta with Apple Maps further documented on
MediaWiki.org[8].
In conclusion, I would like to thank you for the feedback and the
opportunity to engage in a civil discussion about these important issues.
Again, if you are interested in the next steps, I’d invite you to
subscribe
Maps_service#Privacy
_______________________________________________
Wikimedia-l mailing list, guidelines at:
https://meta.wikimedia.org/
wiki/Mailing_lists/Guidelines and
https://meta.wikimedia.org/
wiki/Wikimedia-l
New messages to: Wikimedia-l(a)lists.wikimedia.org
Unsubscribe:
https://lists.wikimedia.org/mailman/listinfo/wikimedia-l,
<mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe>