Wikimedia monthly activities meetings/Quarterly reviews/Wikipedia Zero/September 2014
The following are notes from the Quarterly Review meeting with the Wikimedia Foundation's Wikipedia Zero (Mobile Partnerships) team, September 29, 2014, 2:15PM - 3:45PM PDT.
Present (in the office): Tomasz Finc, Yana Welinder, Carolynne Schloeder, Kevin Leduc, Yuri Astrakhan, Rachel diCerbo, Toby Negrin, Kevin Leduc, Damon Sicore, Lila Tretikov, Erik Moeller, Tilman Bayer (taking minutes), Jeff Hobson, Adele Vrana, Dan Foy, Adam Baso; participating remotely: Smriti Gupta, Jorge Vargas, Brandon Black
Please keep in mind that these minutes are mostly a rough transcript of what was said at the meeting, rather than a source of authoritative information. Consider referring to the presentation slides, blog posts, press releases and other official material
NB: not all our partnerships are WP0, we are also working with OEM's
Team is growing
Jorge (formerly legal intern) now as contractor from Colombia, working on policy issues & partnerships
still missing SE Asia partnerships manager
couple of team members out in Q1; still short-staffed
we hit our big goals:
migrate 52.5% of partners to IP (whitelisting, instead of URL whitelisting)
second SMS pilot deprioritized
Q1 was a lot of OEMs and policy
Lila: are we seeing a lot of traction with OEMs?
Carolynne: yes, making good progress
Lila: Tracking uniques yet?
(team:) do not want to track unique users [for privacy reasons], because we would need to give them cookies
Lila: talk to Toby
I want to compare apples to apples
(team:) currently UVs via ComScore
next q via our own site
Erik: definitions differ though
Lila: OK, but I want us to be able to attribute uniques [by either def] to WP0
Erik: team should be involved in LUCID [Limited-duration Unique Client Identifiers] discussions with Oliver from Analytics
Carolynne: concerned about scaling partners
Communications is still a gap
need to work specifically on Global South UX: support for languages, feature phones
thrid party use of content (mediawiki): great, but need to bring users back in
Lila: waiting on the Comms hire?
Carolynne: yes, and in the meantime working with Katherine and Minassian on comms
Minassian don't understand us enough for this, need in-house
Lila: what is your objective for communications?
Carolynne: visibility of programs
marketing efforts with partners present lots of opportunities (need to raise awareness in target segments)
Lila: try measure demand based on that?
e.g. when publishing a blog post, have a call to action in that?
Carolynne: yes, e.g. the recent post on Chile
Yana: and the petition
Lila: need more strategy
marketing vs. classical Comms (press) work
generate specific interest (via media) vs. just know how to write press release
Carolynne: agree we need "marketing"
since last qr, built pipeline (e.g. onboarded Smriti)
Reach: 603M subscribers
with internet-enabled devices: 400M (back of the envelope)
pageviews: Yuri worked with Analytics
on old dashboards, missed HTTPS traffic
which we just turned on
Lila: so you were blind to that?
Yuri: yes, did not collect it
Carolynne: but HTTPS was only just turned on
Q1 was a bit flat, but did IP migration - might get like 10-20% bonus traffic
are on track for December goal
will set aim on stretch goals
steadily increasing #operators
Four launches in Asia in Q1
Smriti: Asia update
Q1 pipeline: started with 14 telcos, many new
Existing telcos: saw pageviews dip in summer and pick up again starting Aug
checked: is the integration optimal? can we improve PVs?
e.g. started Opera support
M-dot push: convert from zero.wikipedia.org to m.wikipedia.org
one rationale: zero-dot does not get organic traffic
[slide 12]Asia pipeline
One large telco in India: talking about a test, hope they will expand it
Carolynne: Kul and Amit had tried to crack India before - difficult market; it's significant that Smriti got this going
Lila: how long to close?
Carolynne: we can still speed that up, three months is minimum
Lila: could we put these numbers in relation with population and subscriber numbers (penetration rate)?
Carolynne, Tomasz: sure
Lila: will help us focus
Africa & Middle East update
in Africa, we prioritize "big fish"
context for numbers:
- contract delays
- some had problems implementing IP whitelisting (instead of URL)
growth in Rwanda, Nigeria, Botswana
things are changing there
getting important operators on board
Lila: I know, just doing contracts can be very slow in Brazil ;)
I've been supporting the team on policy side for 6-8 months, excited to be on partnership side now too
work with Edu team
Lila: what does subscriber number mean? pilot itself or reach in general?
Carolynne: subscribers of that operator
Jorge: Chile - see blog post on why it's an important country
this is preliminary work
[slide 15]IP migration
Lila: do they choose to activate HTTPS?
challenges: lot of back and forth with partners, find out what they are capable of doing
so far offered choice of whitelisting by URLs or IP
now with HTTPS, IP whitelisting is required
some even don't know immediately if they can support HTTPS
this is going to be focus in the next q
Lila: so for new partners, it's required?
mixed column in table (IP/URL) contains some of the biggest partners
Lila: what about those who can't do it?
Dan: sometimes we go back after a while to see if their hardware has been upgraded
recall we work in developing world with lots of old hardware; sometimes it makes sense to wait for upgrade
ongoing support for URL
we offer them assistance (on HTTPS) - we actually request manuals for their systems sometimes ;)
Tomasz: it happens ;)
Dan: but some of them really can't, because of hardware is too old
Erik: so the "can't" category contains 2.3% of PVs?
Erik: need to define a threshold
recall that these are people who won't be able to edit
Carolynne: looking at it differently, it's 5% of PVs (including an estimate of non-IP in "mixed")
agree that the confused message is a problem (people not able to edit on WP0)
Carolynne, Lila: there's also value in making pages available for viewing only
Yana: they might be motivated [by reading Wikipedia on Zero] to get on Wifi to edit
Erik: agree, we can accept some of this, just need to develop a sense where that threshold is, monitor development
Health in Africa
Language in South Asia
Education in SE Asia
education: help our partners support our community - cool stuff going on
Pageviews in India
growing, but want to find out where PVs are coming from
India: growth via local languages
disconnect between how much of the people need acccess in local languages, and how many English (not local language) WP page-views we see
task not just for us, but also for community and Language Engineering team
local languages come up in every meeting with partners, important topic there
Lila: can we attribute pageviews to WP0? yes
Carolynne: we do reach people who don't care if it's free (middle class)
Yuri: that graph (Aircel) is WP0 only
Dan: many older cellphones can't display complex fonts for Indic languages; for Android, only newer models can
Carolynne: that's where relationships to handset manufacturers become important
working with Anasuya and Asaf from Grantmaking, language engineering and product/UX
Toby: historically we haven't geocoded by region in e.g. India, new package can do that[?]
Erik: not clear that this will shift even if we do everything right
Lila: Grantmaking is focusing on India, opportunity
e.g. largest non-English languages
Smriti: local lang speaker might need to go through 10 steps to just configure their phone
Lila: UX is really important
understand how phones are configured (get that info from handset manufacturer, or operator)
then, check how does font look like etc. - need to look at that directly
investigate that, as timeboxed project
even UNICEF etc. do their campaigns only in English
community has translated WP article in local languages
suggested to operators to send SMS with article in local languages
Lila: should have actionable information, not just e.g. history of the virus
e.g. how to do burials
partnerships to distribute health content
(general, e.g. about hand washing - not just Ebola)
we were clear that WMF doesn't touch content (work with editors/Wiki Project Med foundation instead)
partner with GSMA
for distribution on feature phones
include content on SIM cards
Lila: total population affected?
Adele: drives syndication
preloaded on Samsung devices
Tomasz: app or content preload?
Firefox OS already does preload, Windows too
Erik: need to be careful of branding around health
have general disclaimer
Yana: if they curate something based on our content, perhaps it should not be called Wikipedia
Toby: with handset preload, still decided by carriers?
Carolynne: it depends. Some are open handset markets.
(Adele:) OEM partnerships
We have commitment from one OEM to preload Wikipedia, more in pipeline
search integration, preload (seed content) - there is interest
Lila: (get mobile search right)
Tomasz: careful not to get into curating contenet
Carolynne: we're not touching content, Yana watches us ;)
Lila: Qualcomm has this nonprofit arm...
Lila: what's your thinking if OEM takes off?
Carolynne: want to do both OEM and Zero
Dan: Q1 results vs. Goals
hiring took longer
product & engineering Q1 results vs. goals:
Portal [for partners]: entire interface is done and mocked up
Jeff just started
begin implementation work tomorrow
building blocks were put in place by Yuri
portal mockup: 3 tracks legal/marketing /tech
Yuri: posted link for demo (defunct)
Work with Analytics, but also do our own analysis (lots of domain knowledge required)
curently from logs, but need full access to Hadoop
would like to collecct more data
new Graph extension for MediaWiki, Fundraisinig team is interested too
bonus: will be usable by Wikipedia editors as well, we would remove necessity for images in these cases
already have data for many partners, can just switch over from Limn to this new extension, it works dynamically
demo (just as an aside) Visualized edit history for Obama & D Adams, 500 last edits - could be added as a template to every article's talk page. (Adams spike from Google Doodle)
Toby: main thing you get [compared to Limn] is access control
Dan: slow getting data, neither Ops nor Analytics can measure
Tomasz: image compression?
Dan: can do on page by page basis internally, see 30% reduction, but needs to be verified externally
Erik: active where?
Dan: all new partners
will move existing ones too
Erik: evidence that compression has increased PVs?
Dan: working on analyzing that
Damon: In what layer is the compression happening?
Adam: on app servers
Dan: (Q2 focus)
portal - communicates [exchanging data] with CRM
add banners to sister projects, e.g. Wiktionary
make sure Yuri's Graph extension is consistent with previous
started to see questions regarding net neutrality
worked to explain that WP0 does not violate NN
started in Chile, because of new order there, with media reports that it would prohibit WP0
worked with them, to not undermine NN but not prevent WP0 either
regulator in Chile told us he does see difference between WP0 and the prohibited services
and other educational initiatives - important there
direct carrier outreach
Adele: Chile precedent has helped me in Brasil
Yana: presentation at Internet Governance Forum (IGF)
focus on digital divide
also highlighted our operating principles
very positive discussion at IGF, saw reports from conference that acknowledged that
and blog post on WP0 and NN
focus on A2K (access to knowledge): Wikimania talk with Yale ISP scholar, IGF panel discussion, meetings with advocates to collaborate on A2K
Lila: IMO it's a separate issue, WP is a public service. focus on A2K
need to work on messaging
Wikipedia Zero brand:
decide whether to retain term "Zero"
target mobile world congress
Lila: worry about expressing the goal of A2K [strongly enough]
should be a priority for Comms
Ecconet piloting free access to 50 educational services in Zimbabwe
in Zambia PVs, we see effect of internet.org launch
Lila: they have an army compared to us ;)
Carolynne: we let the portal slide a bit and still have hiring to do, but otherwise good on Q2 goals
Erik: likes seeing PVs climbing up ;)