Free Software Foundation president Richard Stallman (RMS) was in
Canada in April and in May 2018 to participate in a couple of
On April 30th, he was in
Ottawa, to support an initiative to create a free
software solution to the Canadian government's employee payroll
debacle. He and Joseph Potvin, executive director of Xalgorithms
Foundation,1 led a breakfast discussion titled "Two lessons
from the Phoenix payroll puzzle: Software freedom & algorithm
Phoenix is the Canadian government's new payroll system, which was
supposed to provide "an employee self-service vehicle to decentralize
data entry and increase access to information." Since its rollout in
2016, however, it has been plagued with malfunctions, which have led
to under-, non-, and over-payments to over 200,000 federal
The resulting stress and hardship for all affected has been
considerable and, more than two years later, in spite of national
outrage and ballooning costs ($1.2 billion and counting), the system
is still not fixed.
As RMS points out, "Phoenix shows that state use of nonfree
software can create a continuing disaster from which the only escape
is a free replacement."
On May 29th, 2018, Canada's
auditor general reported
on the enormity of the failure. In the search for the causes of
the problem, few have considered a practical software solution; in
their event on 30th April, however,
Potvin and RMS did just that. They proposed that (a) nonfree software
being used by a government and (b) inaccessible rules are the two root
causes behind the fiasco.
The auditor general "concluded that the Phoenix project was an
incomprehensible failure of project management and oversight."
However, Potvin, who for six years led IT expenditure analysis and
reporting for the Chief Information Officer Branch of the Treasury
Board of Canada Secretariat, says, "It's only incomprehensible if the
essential questions are not asked."
Because the Phoenix pay system relies on nonfree software, according to Potvin, the first root cause is that
"the Canadian government does not have the source code for it and, due
to restrictive licensing, can't change suppliers. Nobody beyond the
original contractors are allowed to run a copy of the system
independently, to study how it all works, to run tests on it, or to
adapt it with improvements."
Government exists for the people, and does its computing—all of
it—for the people. So every public agency's responsibility to
the people includes maintaining control of that computing. Any time
government uses a non-free computer program, then logically, its owner
controls that part of government's computing. We the people must
insist that our government maintain full control of its computing,
and must not let any other party take control of it.
Potvin went on to explain how the second root cause was inaccessible
payroll rules: not only are they idiosyncratic, poorly-documented, and
convoluted, but in the Phoenix system, each rule is a hard-coded
procedural step instead of being dynamic and declarative. He said,
"They famously claim there are around 80,000 rules. But we say: How
many rule patterns are there? Less than a hundred? Maybe two dozen?"
RMS added that, when it comes to data-driven decision algorithms
that a government uses, there is no accountability unless the
algorithms are published.
Going well past merely analyzing the problem, Xalgorithms
Foundation plans to fix the Phoenix problem with an initiative
called MyPayChecker (CheckerMaPaye, in French), which "can provide
those employees and payroll officers the most consistently accurate
and transparently documented pay information, while also respecting
According to Potvin,
The MyPayChecker Working Group will enable the world’s first
digitally automated collective labor agreement. This means every
clause, deduction and entitlement, and every priority notification,
would be expressed in a transparent computable form in a free/libre
schedule (i.e. annex) to the agreement, in addition to its
conventional expression in natural language text. And each computable
clause would be automatically and accurately discoverable on the
Internet, to be efficiently retrieved without restriction for use in
applicable transactions by at least three independent payroll
We're going to throw down the gauntlet and issue a clear challenge to
the companies behind the Phoenix project. We claim that a free
software community can independently finance, create and operate
within a year, a more respectful, more accurate, and more economical
self-service portal for pay validation to support employees and
payroll officers, than those companies can provide within a year.
According to Xalgorithms Foundation, the "Paycheck Target
Challenge" (« Défi des salaires ciblés »), which Potvin announced
during the breakfast discussion, will begin officially on June 1st, 2018:
Some lead time is required for the rules of this unusual competition
to be discussed and finalized with stakeholders, and to invite other
investors, companies, and technically inclined individuals, including
Canadian Government employees, to collaborate with Xalgorithms Foundation in
making MyPayChecker the service to beat. [...] And if the free
software community wins this bet, the companies behind Phoenix can
adopt it, since MyPayChecker will 100% shared.
Which Potvin jokingly referred to as being something akin to
"wearing the other team's hockey jersey" and, more seriously, as a
huge incentive.3 As the reporting
past week has shown, the Phoenix failure is currently a huge issue in Canada, and Xalgorithms Foundation's
challenge could not be coming at a more opportune time.
Some photos from the breakfast discussion…
(Copyright © 2018 Mike Gifford (mgifford). Photos licensed under CC BY 4.0.)
…and, clockwise from top right, RMS checking email above the Ottawa
River after the event, RMS and Joseph Potvin, and David Graham, a
member of the Canadian Parliament for the riding of
Laurentides-Labelle and an outspoken supporter of free software, who had
attended the breakfast discussion. Three weeks after the event, on May
22th, 2018, in an appeal in favor of
net neutrality in the House of Commons, he commented,
"In the words of Richard Stallman, the father of the free software
movement, either the user controls the program or the program controls
(Copyright © 2018 Joseph Potvin. Photos licensed under CC BY 4.0.)
After Ottawa, RMS then headed to Montreal for the
5th free colloquium of the Adte (an
association for free software, free science, free educational
resources and free data for colleges and universities). This was
RMS's second appearance at the event, and he gave his speech "Logiciel
libre et éducation" ("Education and freedom"1).
(Copyright © 2018 Adte.ca. Photos licensed under CC BY-SA 4.0.
For all the photos from the event, along with comments, see here.
Thank you to everyone who made all these trips possible!
Please fill the linked contact forms so
that we can inform you about future events in and around Ottawa
Please see www.fsf.org/events for a full
list of all of RMS's confirmed engagements,
and contact firstname.lastname@example.org if you'd like
him to come speak.
Foundation convenes the primary Internet of Rules Working Group to
design, produce, test, maintain and document Xalgo, XalgoAuthor,
Lichen, and Interlibr, and to ensure that they operate well as a
loosely-coupled system to express, publish and fetch computational
algorithms on the Internet. As of May 2018 six Working Group charters
are being drafted and discussed with sector stakeholders:
Follow these links to see its source code, architectural
documentation, and organizational
2. See here
for the event's presentation file.↩
3. For more on this, see Xalgorithms's media
4. Slides available here.↩
- Xalgo4Trade (for automated inter-jurisdictional trade agreements),
- MyPayChecker/CheckerMaPaye (automated labour agreements),
- Benchmarks (for dynamic pricing linked to verifiable reference data),
- Reciprocal Data (for data empowerment of individuals, and "mutual loyalty"),
- Industrial Control Systems (for dynamic manufacturing process control),
- AlgoRisks (for management of algorithm errors and omissions liability risk).