Culture of humiliation

I have learned a lot recently about online humiliation.

Monica Lewinsky's Vanity Fair piece was very insightful. Indeed, if anyone would know about being the victim of online shaming, it would be her. Interestingly, she shares that a new stage of horror for her was achieved when she understood the impact his scandal had on her family.

In September of 2010, the culmination of these experiences began to snap into a broader context for me. A phone conversation with my mother shifted the lens through which I viewed my world. We were discussing the tragic death of Tyler Clementi. Tyler, you will recall, was an 18-year-old Rutgers freshman who was secretly streamed via Webcam kissing another man. Days later, after being derided and humiliated on social media, he committed suicide by jumping off the George Washington Bridge.

My mom wept. Sobbing, she kept repeating over and over, “How his parents must feel … his poor parents.”

It was an unbearably tragic event, and while hearing of it brought me to tears, too, I couldn’t quite grasp why my mom was so distraught. And then it dawned on me: she was reliving 1998, when she wouldn’t let me out of her sight. She was replaying those weeks when she stayed by my bed, night after night, because I, too, was suicidal. The shame, the scorn, and the fear that had been thrown at her daughter left her afraid that I would take my own life—a fear that I would be literally humiliated to death.

—Monica Lewinsky, Shame and Survival

(Click to read the rest of the article)

Building trust in #massiveteaching, through visuals

As part of the module on trust in Connected Courses, I would like to share some tricks that I have used to visually build trust across cultures. It was especially powerful for me to watch Jonathan Worth's video on Leveraging your why [1].

Worth offers that his "Why?" for #phonar is actually citizen journalism, and discusses a photo by John Stanmeyer, winner of the World Press Photo 2013 award.

So this brings me back to, this is my ‘why?’ So this image is the winning image from the World Press Photo this year, by a photographer called John Stanmeyer, and this is why it annoys me so much. Because it’s by a photographer called John Stanmeyer. These are African immigrants standing on a beach, trying to get a phone signal before they jump in the water. Let me just get this right, “on the shore of Djibouti City at night raising their phones in an attempt to capture an inexpensive signal from neighbouring Somalia”. These are the people that aren’t in my class. They’re holding these smart devices which can all take pictures, and yet it takes a photographer from New York to bring it back and we can now talk about it. What these people need is to be able to be trusted and to be heard, and these are the people that aren’t in the class. So this is my why, and this is my way marker, so this is definitely the best and most exciting thing that’s come out of my classes.

Jonathan Worth

Part of my efforts with my Coursera course Massive Teaching: News skills required were some interesting blend of Worth's efforts in #phonar and in #connected_courses.

(Click to read the rest of the article)

Security theatre, in airports and academia

When I was in Oxford, I had the best view in the world from my apartment. Off Merton Field, above Deadman's Walk onto Christ Church Meadow. Uninterrupted grass landscape into the river, with rolling mist up to the river most mornings, Lord of the Rings-style. In fact, Tolkien's office was the apartment upstairs, and I like to think the view inspired him. How come, in a cramped city like this, I could enjoy this same view decades later, centuries after the construction of the building I was living in? Well, because in the byzantine world of Oxford real estate, one college (ChristChurch) owned that land, while another owned grazing rights for cattle on it (forgot which, presumably Merton itself). Not that tourists would see cattle in that field: every year, the college with grazing rights would rent sheep for just one day, and move it there. The theory was that if they didn't exercise their right, they would legally lose it, and the other college could build whatever they wanted there and ruin the view (Wikipedia mentions a road that was defeated). The pace at Oxford allowed to think this through: custom is built over centuries, and preserved with yearly activities.

Why this long preamble?

I was just in Goteborg for the #Chals14 conference, where a big part of my schpiel was that we should not succumb to the free-is-a-lie mantra of MOOCs and resist narrow views of the "Internet happening to education" (quoting George Siemens) based around technological determinism (quoting Evgeny Morozov). The afternoon session was in Swedish, so I was distracted and looking at Twitter. One person that I am following, Aral Balkan, tweeted this:

Balkan is a privacy advocate, who has been trying to raise the attention of people to the Spyware 2.0 model employed by many companies now: in his Orwellian view, by offering free services, large corporations manage to convince us to give away all our data, so they can spy on us.

(Click to read the rest of the article)

U Colorado system Coursera contract

Over at his blog, Jonathan Rees discusses the decision of the Colorado State University in Fort Collins to switch LMSes, from Blackboard to Canvas.

He makes sly references to the fact that this blog post starts hitting very close to his own "home", since he is working himself at Colorado State University - Pueblo.

In a previous post on the Coursera contracts, I gathered together all the Coursera contracts that I could find online. This was in fact partly a response to repeated appeals he has done in the past for faculty to speak up across universities (see first line of this post).

One of those contracts was from the University of Colorado system, it is here. I was wondering how the decision was taken in Colorado of joining Coursera in the first place, and if faculty were consulted on this. The comments are below, feel free to contribute (even anonymously if you would like).

Now, I was also wondering how he (and other professors and administrators) would react to this present post, over which he has no control (and we haven't discussed this beforehand).

To increase likelihood that someone would respond, I also posted a comment on his blog sending readers here.

MOOCs as inventions #chals14

I just finished giving my talk on MOOCs as inventions: opportunities and risks in Goteborg, at the #chals14 meeting. This is a reunion of Swedish universities convened by Jonas Gilbert (Chalmers University Library) to discuss MOOCs.

My slides are available here.

In the talk, I discussed the idea that MOOCs are social machines put in the hands of professors (thanks to my PhD student Helen Riedtmann for boiling down my ideas to this simple statement), and quoted Evgeny Morozov, which I am reading at the moment.

In retrospect, it was much better to be last speaker of the morning, to offer a (hopefully constructive) balance to some of the other points made earlier. It also lead naturally to interesting discussions (including with Karin Markides, who asked me a few questions above my pay grade. It is much easier to convey these ideas offline than online, but indeed the audience is smaller and different). Still, the digital world was interesting too: I was glad to see how Twitter added to the talk for me afterwards (something I am not used to, since very few mathematicians use Twitter), which helped me assess how my ideas were received. In a talk about MOOCs' risks and opportunities, someone immediately called me up on the risks of the opportunities, with a reference to Jaron Lanier:

Now adding this to my reading list, since in fact he seems to talk a lot about mathematics too! It was also good to meet Jorg Pareigis, a fellow #ccourses participant.

Two quick mementos for myself:

  • I have now traveled to the first 15 countries to constitute the European Union.
  • I have never been before in such last minute trouble with computers: still private Google Slides locked out of my account, requiring Google two-factor authentication through a dead cell phone... Literally managed to get to the slides as I was walking to the stand.

Foundations of mathematics


The ETH Zurich hosts every year the Paul Bernays lectures. This the type of lecture where people expect a Nobel prize winner to show up. Well, unless it is a mathematics one (the topic rotates among a few sciences), in which case it is OK if the speaker only has a Fields medal.

This year, it was the latter. The speaker was Vladimir Voevodsky from the Institute for Advanced Study in Princeton (the same institute where Einstein and many other brilliant scientists were coopted following great successes).

Voevodsky's initial success was in proving the Milnor conjecture, but he hasn't stopped there. He has announced a proof of the Bloch-Kato conjecture as well, and is now working on Homotopy Type Theory.

This theory is tied to the foundations of mathematics. To the layman, it really asks deep questions about what mathematics is, and how mathematics is constructed and improved, both by computers and humans. In some ways, he is trying to scientifically map out mathematics, and during the talk suggested a 500 year horizon to do that (part of many bombastic and exciting claims made by very smart mathematicians, such as Thomas Hales at the end of his lecture here, who are willing to challenge the statu quo). Of course, one needs to start "small", and this is what Voevodsky explained in Zurich, his vision of the Foundations of Mathematics.

For this he had three times one hour, which he sensibly divided into: past, present, and future.

I want to focus here on the "past" talk. He started it by quoting wikipedia's definition of Foundations of Mathematics:


and highlighted that there were two parts to this definition: the mathematical concepts and the metamathematical concepts (the structures that form the language). He then went on to explain the presence of this dichotomy through the ages: discrete vs. continuous, pure vs. applied, theorem vs. problem (an inquiry the answer to which is a construction of an object), and eventually touched on this account of what mathematics is:


Voevodsky's interpretation of what Grassmann is saying is thus: mathematics is seen as rising from many independent roots and acquiring wholeness only through the intertwining of the lines arising from these roots.

Coursera contracts of Stony Brook, Roma, UNC, Colorado, Kentucky, Michigan, UToronto, UCSC (and your university?)

Scholarship of the XXIst century will include a lot of research on teaching, learning, and citizen science. MOOCs offer fantastic opportunities in this area. The XXIst century should also probably include scholarship of its own labor practices, or at least some concerted effort to stay afloat of developments in the profession. Asymmetric information can hurt badly in negociations. For this reason, it could be helpful to gather all Coursera Framework Agreements that have landed on the internet (so far) in one place. Those agreements bind the university administration and Coursera, and package within them the actual contract to be signed later by the instructor (and guests), both with the administration and with Coursera. On top, any user of Coursera requires an account which is bound by its own Terms of Service [1].

By simply googling around (mostly for Coursera Framework Agreement), I was able to find the following eight contracts:

Not all of those are signed, some are just preliminary and might not match the final version. Some of those were originally obtained through Freedom of Information requests, while others were obviously misplaced. Some seem to have been removed from the web since, so I put them back here. The more interesting aspects here are the subtle variations between these contracts.

Should you have one that you want to share, feel free to send it to me.

I also include the following Course Development Agreement for UNC. This contract is packaged within the Framework Agreement, and one is signed for each course at UNC, according to the CFA. It is relatively short, but refers to external guidelines located on the Coursera Partner Portal (i.e. online, inaccessible without an account with different permissions than a regular student account).

It includes the rollout policy for a particular course: 1.d.i and 1.d.ii dictate that the content should be uploaded one month ahead of the course start for the first week of content, and two weeks ahead for the second week, but "[these] will not apply to the University Courses beginning in the Fall of 2013 if the achievement of these requirements is not practical or not feasible". UNC joined Coursera on February 2013.

The CDA also includes the following:

  1. Course reusability: Ensuring that the content can be easily reused in subsequent offerings of the course for which the instructor might not be involved. This includes:

a. Avoiding references to particular dates in lecture videos (e.g., "The homework will be due next Monday, March 4" or "It’s a beautiful summer day today");

b. After the first offering of the course, systematically editing/removing references to dates and episodes that are specific to the first offering of the course (e.g., an announcement about correcting a typo on the final exam, or about extending due dates because of Hurricane Sandy), as well as references to activities for which instructor involvement is necessary (e.g., "We hope to see you at the next live Google+ Hangout").

—Coursera Development Agreement, to be signed by Coursera instructors at UNC (emphasis mine)

I have read through quite a lot of this, so if you have any question feel free to ask!

[1] As indicated by Jonathan Mayer, since this is what gave him sufficient credentials to potentially dump the Coursera database of 9M students' names and emails.

Edtech policies (part II)

For the first post in my series on edtech privacy policies, terms of service and contracts, click here.

In his recent post You are not alone on the Salaita affair, Jonathan Rees comments on the social media policies enacted at the Colorado, Kentucky and New Mexico state systems. He encourages other professors to speak up:

Few of us can be sure whether or not our campus is really the canary in the higher education coal mine. However, better mine safety everywhere eventually benefits everyone with a pick and shovel. The more we professors talk to each other, both on social media and off, the better off we’ll all be when bad ideas surface that may someday affect us all.

—Jonathan Rees

The evocation of the canary is very beautiful. It resonates quite a bit: ETH Zurich was after all founded to help the Swiss develop their own railway systems. In any case, looking back at his writing, Rees has really dug in the Coursera labor practices. This seems like a good opportunity to continue my edtech policies series.

After some googling, one can easily find the Coursera Framework Agreement between Coursera Inc., a Delaware corporation [..] and the Board of Regents of the University of Colorado [..] for the benefit of the University of Colorado Boulder.


Independently of any stereotype that as an academic I hate capitalism, another stereotype could be that I also quite like to read contracts. Of course this one makes for a particularly interesting read: it is supposed to hint at how MOOCs will be sustainable for the UC system and how Coursera, Inc. will outlast its 85M dollars "gifts" from venture capital.

The contract details the various rights and obligations of the parties involved, the different Coursera monetization strategies (ten instead of the eight like in the 2012 Michigan contract), the Course Acceptance Procedures, the responsibility of the University Advisory Board when things go wrong, etc. Of particular interest to me is how it already packages in Exhibit G-1 the relationship between an instructor and the university.

It also includes the following gem:

I hereby release, discharge, and promise not to sue Company and its affiliates, successors and assigns from and against any and all claims, demands, costs and/or causes of actions of any nature arising out of or in connection with the exercise of any rights herein granted, including, without limitation, any claim for infringement, right of publicity, libel, slander, defamation, moral rights, invasion of privacy or violation of any other rights relating to any Content I upload, share or otherwise provide in connection with use of the Platform.

—Exhibit G-1, Form of Agreement for Instructors and Guest Presenters

In other words, IANAL but once the instructor signs this, s/he is giving free pass to Coursera for any libel, slander, defamation, invasion of privacy etc against her/himself. I don't know the backstory on how the faculty at Colorado accepted this, but would be very interested. At UCSC, the collective bargaining chapter has on the other hand been very critical, as has been discussed in part I of this series.

Rees' words on this are very apt, to his credit a year ago already:

At some point you have to realize that you do not work in a vacuum. Teaching a MOOC or not teaching a MOOC, speaking out in favor of them or remaining silent – these things all have an effect on the rest of your profession. You can’t just declare that MOOCs are the future and let the chips fall where they may. You have the ability to influence what kind of future we’ll all face.

—Jonathan Rees (Labor Day 2013)

Can we guess what XXIst century education might guarantee in terms of labor rights? Is there an online strike clause in any of them? All I could find is called Force Majeure:

Each party is excused from performance of this Agreement (other than for any payments due) and will not be liable for any delay in whole or in part caused by the occurrence of any contigency beyond the reasonable control of such Party. These contigencies include, without limitation, war, sabotage, insurrection, riot or other acts of civil disobedience, act of public enemy, failure or delay in transportation, act of government or any agency or subdivision thereof affecting the terms of this Agreement or otherwise, judicial action, labor dispute, student disorders, accident, fire, explosion, flood, severe weather, natural disaster or other act of God, shortage of labor, hardware failure, interruptions or failure of the Internet or third-party network connections or incapacity of an instructor.

Force majeure clause 19.6 of University of Colorado Boulder contract (emphasis mine)

I don't know if those risks are sorted in any way, but can't help but notice that act of God sits right between labor dispute and failure of the Internet.

Apparently Rees has a show on the road. The title is Educational Technology, Budgetary Priorities and Academic Freedom. Should be good.

The academic (social) machine (part I)

There was recently a picture circulating on Twitter, like pictures do.


This is, to say the least, a skewed view of academia, although I am certainly not the best placed to say that. I tend to have a beard, use big words, have recently started blogging and did wear robes at some point in my academic career. This is however a good opportunity to show how algorithmic bias works.

First off, where does the bias originate here? As I explained before in my post on social teaching machines, autocompletion surfaces information collected previously. The information in this case is collected in various ways, most notably by looking at previous searches. It is difficult to make broad statements on why people make a Google search. It's an act that always occurs in a certain context, for a ton of different reasons, and Google optimises for an average that is unclear. Who else is making the same search will undoubtedly have an effect on how Google sees the context of a search. Consider the search Why do English people have British humor? and its first answer: Google obviously got the context wrong in my case.


I am a bit at a loss to say more on this, so feel free to comment.

In any case, some of those biases are substantially more serious, of course. For this, you simply have to enter a search of the form "Why do A people B", where A can be any of {asian, white, black} and B any of {look, like smell} to realise that autocomplete is powerful to surface common stereotypes. Not all those autocompletes work though (presumably because the output is too vile and has been hand blocked). So we humans enter our biases, and Google actually amplifies them.

Let's see where this leads.

In the case of autocompletion, the impact is certainly weak, but it might correlate with other biases (or cause them?), underscoring a more ingrained problem. Let's go back to Google's view of academia: what does the output of a Google Image Search of academics faculty return? You can try to use this link, which is user agnostic (but its output will be personalized by Google once you click, unless you use privacy conscious tools). Here is the view I get, when logged in:


Yours should be different: most probably, Terence Tao, the short-sleeved mathematician in the middle row, is further down in yours. This is reasonable, and explained by Christian Sandvig in a beautiful post called Show-and-Tell: Algorithmic Culture: since I am a mathematician, Google gives Tao a bump [1]. And beyond that? Well, Google really thinks that academics wear robes half the time, and perpetuates this bias also visually, not just in autocomplete.

Is this really a serious problem? Well, one consequence is that when humans need to illustrate something (a blog, an educational resource,...) it actually requires effort, judgement to accurately assess the potential bias and not succumb to it. For robes it is of course very easy. For skin color or diversity, as we know, it can be harder and thus requires training and conscious effort. In its outputs, Google is potentially already biased, in ways that are hard to assess for all of us who don't know the Google secret sauce.

Automation is moving rapidly in many areas. Online advertising could easily remove any intermediate human step, refeeding to us our own biases for commercial gain. Services like Sketch2Photo offer the promise of automating the illustration processes, for instance for lecture notes, which would perpetuate this same effect over more vulnerable populations.

[1] The fact that Tao is bumped higher than average for me is probably good. It makes the output more relevant to me. If overdone it could also lead to some form of filter bubble.


Well, no language, as far as I know, has a single word for that chin-stroking moment you get, often accompanied by a frown on your face, when someone expresses an idea that you’ve never thought of and you have a moment of suddenly seeing possibilities you never saw before.

In Ithkuil, it’s ašţal.

John Quijada, Department of Motor Vehicles

That's what we are going for here.