[Wikimedia-l] Let's accept Bitcoin as a donation method

Thu Jan 9 23:44:24 UTC 2014

That very rough number that Matt threw out there has far less to do with
the cost of applying human brainpower than it does with the cost of taking
the available brainpower away from things we know are going to
significantly increase our efficacy. We have several of those things
looming on the horizon, and we choose to concentrate new development on
what we know will be the biggest earners out of those.

My understanding (I am no analyst) is that we continue to have a difficult
time finding hard evidence that bitcoin is currently anywhere near the
other top candidates, so it remains off the roadmap in favor of
concentrating on solid numbers. If anybody would like to supply us with
hard figures, we’d certainly be interested in seeing them.

The main reason the expected earnings > one dude’s salary calculation of
worthiness doesn’t work here, is that there are four people in fundraising
engineering. The four of us support and maintain all existing payments
functionality, ensure integrity of the donation pipeline, and do all new
code development and review. For the sake of the foundation and the
movement, each one of us has to do significantly better than individually
break even.

As the fundraising tech lead, I definitely appreciate any outside interest
in potentially helping us out by modifying fundraising code in order to
support more payment methods, and I would be happy to outline the general
process of integrating with a new gateway in a way that is consistent with
our current code.

Before I get in to the nitty-gritty, though, I want to be completely clear
on this one point: Even if I had the authority to do so (I do not), there
is no universe in which I am willing to enable new functionality simply
because the switch exists. Matt has already done a pretty good job
outlining the scope of the collective distraction that bitcoin represents,
and that scope extends well beyond tech. In fact, it seems to me that
producing the actual integration code is the most trivial issue regarding
bitcoin integration that has been brought up thus far, and I would not be
pleased to see well-intentioned volunteer time go to waste over hastily
dismissed blocking issues which exist well outside the purview of the
fundraising tech team.

That said, here is a very general 30,000 foot view of a typical new gateway
integration from a purely technical standpoint:

* Donation Interface[1]: This is the mediawiki extension that initiates
payments. A new gateway adapter child class will need to be created, which
will run in parallel to the existing enabled gateway adapters, and not
short-circuit any of the class constraints that have been deliberately
built in to the gateway adapter parent class. Then, an appropriate form (or
redirect) should be created to handle the user experience, which uses the
RapidHTML templating system. At the end of it all, after a successful
donation has been made, an internal donation message should be queued.
Happily, examples of all the things I just mentioned already exist in other
gateway adapter objects; New gateways are rarely so unusual that we haven’t
nearly done it before.
* Payments Listener[2]: Most payment gateways worth even brief
consideration, have an optional near-realtime notification system. This
system tells us when we receive new payments, and existing payments change
status (cancels, refunds, chargebacks). We would need to create a listener
to receive realtime payment updates, process them securely, and queue
donation messages when appropriate. Though a realtime message listener is
usually not strictly required in order to get paid through a new gateway
integration, I have recently decided to require them wherever possible.
* Nightly reconciliation / auditing[3]: Every payment gateway we integrate
with provides a daily downloadable list of all the transactions we should
have on record. So, a job needs to be created that will download the daily
file and chew through our records to make sure we have all the relevant
data, and rebuild anything we may have missed. This job needs to be set up
to run daily.
* Queue consumer module for civicrm integration[4]: The donations queue
consumer will need to be modified, to accept and correctly process donation
messages from the new gateway, in a way that is consistent with our
existing data.

Of course, all of this work will require pretty consistent code review,
which will bottleneck on the same four fundraising engineers. As it
happens, the number of non-fundraising engineers who are willing to code
review for fundraising without special encouragement, has historically been
so close to zero it’s almost not worth mentioning.

In the event that any volunteers are willing to take this on, I have three
things to say:
#1 - There are no guarantees that your code will ever be enabled by the
WMF. Ever.
Yes, I already said it. It seems important enough to mention twice. Even if
the tech team decides your code is beautiful and flawless and we love you,
it is still likely to end up an overblown expression of futility that spans
four codebases.
If this isn’t a problem for you, by all means: Go for it.
#2 - To greatly increase the likelihood that your code will be reviewed,
looked on favorably, and eventually merged (not enabled, though. Not my
call. See #1), you should probably get in contact with the fundraising tech
team and keep us informed about what you’re up to. The best way to do that
is to get in contact with us on IRC. There are usually a few of us tech
types in #wikimedia-fundraising, so that’s probably the best place to start
looking for specific guidance.
#3 - Just kidding; It’s #1 again.

-Katie Horn
Fundraising Tech Lead
Wikimedia Foundation

[1] - Donation Interface:
https://gerrit.wikimedia.org/r/#/admin/projects/mediawiki/extensions/DonationInterface,branches
[2] - Listener:
https://gerrit.wikimedia.org/r/#/admin/projects/wikimedia/fundraising/SmashPig,branches
[3] - Reconciliation and Auditing: At the moment, our auditing code is not
centralized. Some of it is in the CRM repo (below), and some of it lives in
https://gerrit.wikimedia.org/r/#/admin/projects/wikimedia/fundraising/tools,branches
[4] - Donations Queue Consumer:
https://gerrit.wikimedia.org/r/#/admin/projects/wikimedia/fundraising/crm,branches

On Wed, Jan 8, 2014 at 9:31 PM, John Vandenberg <jayvdb at gmail.com> wrote:

> On Jan 9, 2014 11:38 AM, "Matthew Walker" <mwalker at wikimedia.org> wrote:
> >
> > I will probably regret saying this[1] -- but the figure we like to throw
> > around here in fundraising tech is that a new payments gateway [2] is not
> > even worth considering unless it is likely to make us at least 500K USD a
> > year[3].
>
> Thanks for putting a number on the table.
>
> It is a tad higher than I expected, being more than several very highly
> paid person years, but it is a starting point.
>
> In case an enthuiast who can code is reading this thread, which repository
> needs bitcoin support?
>
> --
> John
> _______________________________________________
> Wikimedia-l mailing list
> Wikimedia-l at lists.wikimedia.org
> Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l,
> <mailto:wikimedia-l-request at lists.wikimedia.org?subject=unsubscribe>
>