The Audacity: Thrun Learns A Lesson and Students Pay

Sebastian Thrun, founder of Udacity, one of the most high-profile private sector attempts to “disrupt” higher education discovered inequality this week. Thrun has spent the last three years dangling the shiny bauble of his elite academic pedigree and messianic vision of the future of higher education before investors and politicos. He promised nothing short of radically transforming higher education for the future by delivering taped classroom lessons of elite professors through massive open online courses.So what went wrong?

After low performance rates, low student satisfaction and faculty revolt, Thrun announced this week that he has given up on MOOCs as a vision for higher education disruption.    The “godfather of free online education” says that the racially, economically diverse students at SJSU,“were students from difficult neighborhoods, without good access to computers, and with all kinds of challenges in their lives…[for them] this medium is not a good fit.” It seems disruption is hard when poor people insist on existing.

Thrun has the right to fail. That’s just business. But he shouldn’t have the right to fail students like those at San Jose State and the public universities that serve them for the sake of doing business.

It is fine if you missed it but for three years now Massive Open Online Courses from tech giants and start-ups have been selling a solution to all that ails higher education. Two short years ago Thrun declared to Silicon Valley and the traditional universities with the courage to follow him that he could not go back to teaching at Stanford. He’d taken the red pill, massive open online courses pioneered by Udacity, and he couldn’t go back.

In 2013, reality harshed Thrun’s red pill high. The low completion issues he had encountered with his MOOC courses while at Stanford became deeper, more fundamental problems at San Jose State. Single digit completion rates in MOOC courses make for-profit colleges’ dismal completion rates look progressive. Thrun said the courses were the pedagogical best he could make them. Coming from a rock star professor from an elite institution, that suggests the class must have been pretty damn good. But it wasn’t good enough for SJSU students.

The faculty at SJSU handed Thrun’s Udacity a very public flogging. Many faculty members questioned the morality of a publicly funded college with a mission to serve diverse students should spend tax-payer money and invest the hopes of students with fewer options than those at the Stanfords of the world into being Thrun’s guinea pigs.

It is a fair question that in many ways the academic and scientific communities have already answered with a resounding no. When I want to interview students for a research project I have to present a carefully, detailed plan to my University for approval. The plan is vetted by an Institutional Review Board. Every research university has an IRB but they didn’t always. Before 1974 doctors figured out the internal reproductive organs of women by cutting them open without consent or sedation. They observed the effects of untreated syphilis on test subjects — insanity and death — without bothering to inform the participants that there was a known, available treatment. They told volunteers they had electrocuted a stranger to see how human decision-making works.

Basically, before IRB a lot of modern science would have been war crimes had the U.S. been on the losing end of World War II. And because this is America, there was a disturbing pattern among the victims of these kinds of horrific experiments. They were overwhelmingly black, brown, indigenous, poor, and powerless. A 1978 report on regulating research on human beings declared that ethical research has “an obligation to protect persons from harm by maximizing anticipated benefits and minimizing possible risks of harm”. The connection to inequality was clear. The most vulnerable were likely to be prodded, poked and tested because the elite don’t often sign up to risk their lives for little reward. And flagrant disregard for these risks had few penalties because the victims were powerless. The rules governing academic and scientific research recognizes that some groups are too vulnerable to risk the failure that the scientific method requires.

Where was this institutional ethic in what Thrun freely concedes was always an experiment? When Udacity was primarily interested in beaming the erudite countenance of professional smart people out into the world, it can be said that any risk was assumed by the those who chose to sign up. But when Udacity went after formal arrangements with colleges like SJSU to offer courses, for credit, to students enrolled in the University, the risk calculation changed.

Udacity’s partnership with SJSU mostly offered general education courses in things like math. General education courses like English and math fill up fast in most colleges because all students have to take them. At places like SJSU that don’t benefit from Stanford’s highly selective admissions standards to skim the most prepared students, those general education classes have to do double-duty filling in learning gaps. Offering these courses for credit using Udacity significantly increases the incentive for students to take the class and risks for students if the class is a dud. General education courses are path dependent, meaning you fail one course at the beginning of a sequence and you cannot take the next course in that sequence. Research shows that disrupting path dependent coursework really hurts the most marginal students by increasing their time to degree completion, dinging their motivation, and sinking their GPAs.

Udacity always knew that the non-completion rates were high for its courses. They may not have known why, but that was a reason for greater testing, not a reason to roll-out the for-profit product for University clients. With sanction from the California governor on down the political line, Udacity  had to meet no ethical requirement to prove that the risk of failure was worth the promise of rewards. And what was promised? University partners could prove they were innovative, forward-thinking, and cut expensive faculty out of the complex equation of teaching students.

To prove that teachers don’t matter and Stanford knows best what the world needs, a public university gave a for-profit company unfettered authority to experiment on its students without informed student consent or consideration of an ethical threshold. We may need more experimentation in higher education but it should be as explicit and ethical as any other we conduct in the name of science and progress.

Thrun says it wasn’t a failure. It was a lesson. But for the students who invested time and tuition in an experiment foisted on them by the  of stewards public highered trusts, failure is a lesson they didn’t need. Students like those at SJSU tend to know quite a bit about failure — institutional, social, and political. They did not need to learn again what Thrun, a smart guy from Stanford and Google, could have learned from a book.

24 thoughts on “The Audacity: Thrun Learns A Lesson and Students Pay

  1. This was a great read and I shared it with some faculty and my Facebook group PhD. If you are on Facebook I would highly recommend your joining the group Tressiemc. There is some fantastic sharing taking place there. Keep up the good fight and all the best to you and your scholarship!! 🙂

  2. Brilliant stuff. will be retweeting! But ‘single-digit drop out rates’ ? do you mean ‘single digit completion rates’??

  3. [Apologies for the length of this! I’ve had pent up thoughts on the subject for a while, so this ended up more blog-post length, and I have no blog of my own to put it on. Sorry to be That Commenter, post or delete as you see fit…]

    Thank you for this post. I really like the comparative lens of the historical background of the IRB system versus the “failure is a feature” model of the start-up culture. As a recovering doctoral student in education, I know well just how much assurance has to be given, in that first context, that possible harm to participants is limited to the extent possible. Also crucial to the MOOC comparison: human-subjects research requires that:
    (a) The possible harm has to be *disclosed* to the participants.
    (b) Participants must be allowed to stop participating whenever they want, without penalty.

    This is SUCH an important angle on the whole topic, and I’ve never before seen it articulated the way you do here.

    I do ALSO hope you, or someone similarly talented, do a deeper dive sometime into the sentiment in your final sentence:
    “They did not need to learn again what Thrun, a smart guy from Stanford and Google, could have learned from a book.”

    I admit to some professional bias, but more than anything, this point is what bothers me most about Thrun, et al. Those of us who have committed to serious intellectual and personal engagement with the complex empirical realities relevant to education could identify any number of foundational things he and his fellow disruptors have clearly neither learned — or, for that matter, don’t show any real interest in learning. (And that’s not even getting into the many folks in my broader professional community who, in addition, *know*-know those complex empirical realities in a way I — a child of White suburban baby-boomers — do not.)

    Why exactly are they so deeply ignorant in this way? How is it that inequality is something Thrun became aware of such a shockingly late moment?
    (If I’m reading that Fast Company article right, even with some awareness — if we can call it that — it also seems like he only counts it as validated “knowledge” because is personal mini-empire “generated” it.)

    Naturally, part of the answer is that this is nothing new, and education’s status position in the constellation of disciplines is clear: a successful and wealthy engineer who wants to experiment in education will OF COURSE be given an irrationally large benefit of the doubt, if not outright cheered for imposing magically-superior ideas which happen to fit the market-oriented reform zeitgeist. (And the magic gets a doubleplusgood bonus when it comes from Stanford — in many ways, a VC firm and technology-transfer holding company with “university” in its name that happens to have a football team and some educational programs.) I think there’s more to it, though. It’s not a coincidence, I suspect, that:
    (a) The Udacity and Coursera founders are all people who have spent their professional lives in elite US universities, but grew up in Europe, Israel and Hong Kong.
    (b) Academically, they all come from computer science. There are structural “issues” with race, gender, etc., everywhere, but those in CS are among the most epic. It’s hard to find other disciplinary cultures that so vigorously conflate talent with prior preparation or take such a perspective-free view of “innate” suitability for participation in its practices.

  4. The last paragraph was powerful and devastating. It should be tattooed on Mr. Thrun’s forehead so he has to read it every morning in the mirror.
    The bottom line has not changed since the GI Bill made correspondence schools viable commercial entities: They’re in it for the money, and student failure has little effect on the bottom line.
    Thanks for the article.

  5. At any of the six Detroit-area colleges & universities I have taught at, the work of face-to-face, in-class & long hours out-of class prodding, motivating, tutoring, mentoring, cajoling, & enticing a high % of extremely ill-prepared students into the super-extra work they require to succeed is a Herculean effort–for faculty as well as students, with a heart-breakingly low success rate. It’s not easy to bootstrap a 3rd-5th grade set of skills & knowledge from your shitty K-12 schools into an education adequate for computer science, medicine, business, or law, adequate to shine in a job interview, and on the job. But online?
    Online requires extremely well developed reading & writing skills, and extremely well developed skills in self-starting, prioritizing, time management, focus, persistence, . . . . The “computer schooling for everyone” hucksters have sold off our Detroit K-12 pupils/students into the MOOC world already. What a wonderland! Not only will we not need to hire teachers, but we will not need buildings. Students can be issued a laptop, like an ankle bracelet, and stay “home” (who cares where that is, if it is, or if it is a good living/learning environment) and learn–or not. If not, there’s always a computer program for your prison cell. Excuse my cynicism. Our decades of unelected emergency managers have already sold our Detroit children into this morass.

  6. This is just tremendous and I really appreciate it, not least since my own University, with a population very similar to that of SJSU, is just now trying to hop on the MOOC bandwagon in what strikes some of us as a very misguided (and belated, thankfully) move.

    I think your raising the question of IRB in an ethical context is very important and you seem to me very careful to raise the question that way. The ethics of conducting such experiments on a large number of students are very questionable, especially when there is so much money involved.

    From a formal standpoint, I think it’s important to note that direct educational efforts are typically exempt from IRB (I think the exceptions are things that may already violate school policy, such as giving students injections and things like that). My understanding is that this exemption exists to protect academic freedom: having teachers pre-approve their classroom activities through IRBs does not sound much like freedom of thought. I believe this is why Nicholas Negroponte’s abhorrent project of dropping laptops into African villages was able to get IRB “approval,” in some twisted fantasy world (though MIT voluntarily follows its own IRB guidelines for projects that aren’t federally-funded, somewhat differently from public institutions; obviously that project was not part of a standard class on the MIT roster, which is part of the criteria at my school). I am not even uncomfortable with raising the question of whether MOOCs are such a dramatic change in college structure that they require IRB approval; I am just putting in my .02 for the importance of keeping classroom education, if possible, out of the IRB world, for reasons too numerous to mention.

    1. I am pretty intent on saying that there is an IRB ethic of weighting risks and rewards against vulnerability of populations. One need not engage IRB to apply a similar ethic to for-profit, third party experimentation in institutions. But I would argue having an IRB ethic during vendor contract negotiations would be a good way to have a conversation about how programs are rolled out, with what confidence, protections, and compensations for students.

  7. Great article. “Disrupting higher education” is SO much less important than, you know, teaching people things.

    But some kinds of “educational experimentation” are okay, maybe. If I were a math professor, and I had a new idea for teaching calculus, and I tried it in a gen ed course at a public university — would that count as experimenting on students, in the same way as Thrun at SJSU or the Stanford Prison Experiments? Why or why not?

    1. If you were doing this as part of your for-profit company, with a clear goal of making a case for people to hire your company,……

  8. If Thrun failed, it’s because SJSU failed before him. SJSU students after graduating are woefully underprepared. You go to a class, you will pass it. Period. Thrun is expecting these students to actually pass some kind of automated test where you actually need a correct answer.

    In the real world, professors grade on a curve, professors get lazy, and they stand in front of you and talk about their lives. They waste time and they know things are hopeless (because teachers before have failed the students and they can’t catch up in one semester) so they just pass their kids.

    Colleges and real teachers are a mess. And the problem is that kids simply don’t want to be there learning the way they are being taught. Uninspired. Udacity is basically the SAME THING AS COLLEGE. You sit down and watch someone on a video talking and talking and talking. And then you do some problems and take an exam.

    Thrun is not disrupting ANYTHING because just taking college and distributing it in on the web is not a disruption. That was happening with VIDEO TAPES and DVDs a LONG TIME AGO.

    Real disruption is when you inspire and engage in a certain way that the student learns and finds the passion to learn on his or her own, without having grades and checking if you understand like Big Brother.

    Real disruption… nobody knows what that is yet in education.

    1. It’s importamnt to qantify how Thrun failed. And it is quantifiable. This was not a failure, per se on the part of SJSU.

      The courses San Jose provided for the same types of students, in the same subjects, outperformed Thruns courses. They provided face to face courses in which the most vulnerable students hugely outperformed their counterparts in Thrun’s courses.

      I agree with your points about real disruption, and about the compolete lack of innovation that Thruns taped lectures provided. It;s 1950’s tech in 21st century clothing.

      But the fact is that San Jose had been doing a better job than Thrun for those students, and it’s importnat to note that the failure here is Thrun’s.

      This doesn;t meqan San Jose are doing a great job with these students – or a bad one either. It does mean, though, that they get to revert to previous practice knowing they did better than Udacity. By a large, large margin.

      1. I don’t feel like anyone can say that it’s Udacity’s fault or SJSU. What data do you guys have to support this? You can’t just compare drop-out rates or pass/fail rates and say one is better than the other because more students passed. You need to acknowledge that with the high percentages of people in this country who attend higher learning that some are going to fail, and should. Not everyone can be an engineer, a scientist, a lawyer, a doctor. You also have to gauge how perceptive these students were to online learning to begin with. Student A could excel in a classroom environment where there is much more social interaction than Student B who goes to class but is completely self-sufficient. There are different learning types. Perhaps the schools involved did a horrible job informing and preparing students for this method of education.

        I’m reading all these posts and rants but I don’t, and nobody will, see data on about these students to draw any conclusions at all. It’s just “he said” and “she said” crap.

  9. I think your analysis is really important and vital. You’re absolutely right that the IRBs on campuses, especially those that serve underrpresented and at-risk groups, should be consulted before MOOCs are used to “Serve” masses of students as a means to save money. Excellent point and of course now seems obvious since you’ve pointed it out but wasn’t before to me, even as I have struggled to argue about problems with MOOCs. Even if MOOCs are low-cost or no-cost, TIME is MONEY–especially for people who are stretched in all ways.

  10. Thank you for writing this. The discussion about the ethics of experimentation in education is much needed. Personally, I think that SJSU should admit that this experiment was unethical and erase the failing grades from the transcripts.

    I think that this analysis also raises a broader question about online instruction in general. Two state-wide studies of community college students in Washington and Maryland show that the completion rates in online courses is significantly lower than the completion rates for f2f courses. Doesn’t that also raise ethical questions? Why are we promoting an instructional medium that fails students at higher rates?

  11. Thanks for the article.

    I think the thread here – that there is culpability for failing these students – is key. Thank you for raising it.Thrun has other culpabilities as well. There are careful, considered, and thoughful learning professionals trying to make something worthwhile out of MOOCs, open education, and dismantling access barriers, and Thrun has damaged them, their cause, and their intentions. Thrun has also, probably, done damage to professionals in traditional teaching contexts (who, ironically, hugely outperformed him in San Jose). His patter, now patently gibberish, of disrupting education has helped cast educators in a particular light – resistant to change, self ineterestedly defensive, unwilling to embrace his new and vibrant paradigm.

    But the biggest failure here is failing the students.

    The sheer, banal, predictability of it is somewhat depressing. The reseacrh on lots of this had already been done, the pedagogy he put forward was known to be archaic, and anyone who looked at the available evidence, both from the literature, and from the MOOCs already run, was going to have huge concerns.

    We already new that certain populations perform less well in online courses than other students, and less well than they do in face to face. Amongst these populations, already identified, were the ones Thrun was targetting in San Jose.

    These were already known things, which, when ignored, damaged the populations they were supposed to serve. Educators take on a huge responsibility. We help form a huge part of the sense students create of themselves, as capable, or incapable people. As smart and clever, or not. As people whose horizons of choice and possibility expand, or contract. We have an impact. We are part of the shaping process, and a powerful one.

    To fail students, so predictably, and ones who can ill afford such failure, is, I think, culpable.

    Thrun was in sore need of a remedial learning theories class. Some Bandura would have been a good start. Some work on prior knowledge and self-efficacy. But not if the remedial course was delivered by Udacity.

    All that said, Udacity is an organisation, not an individual. I wonder who his consultants were?

  12. Well, yes. But. IRBs will pass any clinical trial you can remotely rationalize to treat patients with the brain cancer glioblastoma multiforme (want to give them the malaria drug hydroxychloroquine cause it mucks up autophagy? Knock yourself out!). Because there is no possible way to make it worse. The tragedy is not wanting to experiment on those the status quo is utterly failing, the tragedy is how racially and socioeconomically stratified our education systems are and how screwed up it is that the status quo fails who it does so spectacularly.

  13. Reblogged this on MOOC Madness and commented:
    So many Thrun reactions ~ they almost need an aggregation page or Storify all their own…for all I know, one or both are in the making. Reviewing the Audacity “lesson” from the perspective of her own research area, Tressie MC shows us what others seem to have missed but we all should have caught right off.

  14. May I ask what the content author has done for higher education, other than write an article criticizing somebody for having an idea and failing?

Leave a Reply

Your email address will not be published. Required fields are marked *


The ‘H’ Is Silent: Historically Black Colleges and Mission Conflict #ecsucuts #highered

Elizabeth City State University (ECSU) is a small, public college in North Carolina. It is also a historically black college. Recently, the mission of the state’s University of NC system, of which ECSU became a part in 1972, has bumped up against the historic mission that most HBCUs share. Apparently, seven degree programs at ECSURead More “The ‘H’ Is Silent: Historically Black Colleges and Mission Conflict #ecsucuts #highered”