Udacity Partners with IBM, Amazon for Artificial Intelligence 'Degree'

http://fortune.com/2016/10/25/udacity-ibm-amazon-ai/

Udacity is now valued at over $1b. This seems a long way from the dream of open (libre and free) learning of the early MOOC pioneers (pre-Thrun):

“Earlier this year, Udacity’s revenue from Nanodegrees was growing nearly 30% month over month and the initiative is profitable, according to Thrun. According to one source, Udacity was on track to make $24 million this year. Udacity also just became a unicorn—a startup valued at or above $1 billion—in its most recent $105 million funding round in 2015.”

This should also be a wake-up call to universities that believe their value is measurable by the employability of their graduates. Udacity has commitments from huge companies like IBM, BMW, Tata and others to accept its nanodegree graduates. Nanodegrees are becoming a serious currency in the job market, at lower cost and higher productivity than anything universities can match, with all the notable benefits of online delivery and timeframes that make lifelong learning of up-to-date competencies a reality, not an aspiration. If we don’t adapt to this then universities are, if not dead in the water, definitely at risk of becoming of less relevance.

I recently posted a response to Dave Cormier’s question about the goals of education in which I suggested that our educational institutions play an important sustaining and generative role in cultures  – not just in large-scale societal level culture, but in the myriad overlapping and contained cultures within societies. Though I have reservations about the risks of government involvement in education, I am a little fearful but also a little intrigued about what happens when private organizations start to make a substantial contribution to that role. There have always been a few such cases, and that has always been a useful thing. Having a few alternatives nipping around your heels and introducing fresh ideas helps to keep an ecosystem from stagnating. But this is big scale stuff, and it’s part of a trend that worries me. We are already seeing extremely large contributions to traditional education from private donors like the Gates and Zuckerberg foundations that reinforce dreadful misguided beliefs about what education is, or what it is for. With big funding, these become self-fulfilling beliefs. As long as we can sustain diversity then I think it is not a bad thing, but the massive influence of a few (even well-meaning) individuals with the spending power of nations is  very, very dangerous.

Original post

‘Rote learning, not play, is essential for a child’s education’ – seriously?

https://www.tes.com/news/school-news/breaking-news/rote-learning-essential-a-childs-education-play-isnt-says-expert

An interesting observation…

Helen Abadzi, an expert in cognitive psychology and neuroscience, who was an education specialist at the World Bank, said that pupils who “overlearn” and repeatedly practise tasks, such as mental arithmetic, free up their working memory for more “higher order” analytical thinking.

Yes, they do, good point. We should not forget that. Unfortunately, she goes way beyond her field of expertise and explicitly picks on Sir Ken Robinson in the process…

“Go out and play, well sure – but is that going to teach me mental math so I can go to a store and instantly make a decision about what is the best offer to buy?” she said.

I cannot be certain but, as far as I know, and although he has made the occasional wild assertion, Sir Ken has never for one moment suggested that overlearning should be avoided. In fact, that’s rather obvious from the examples he gives in what the article acknowledges is the most popular TED talk of all time. I’ve yet to meet a good ballerina that has not practiced until it hurt. When you get into the flow of something and truly play, rote learning is exactly what you do. I have practiced my guitar until my fingers bled. Indeed, for each of my many interests in life, I have very notably repeatedly practiced again, again, and again, doing it until I get it right (or at least right enough). I’m doing it right now. I am fairly certain that you have done the same. To suggest that play does not involve an incredible amount of gruelling repetition and rote learning (particularly valuable when done from different angles, in different contexts, and with different purposes, a point Abadzi fails to highlight but I am sure understands) is bizarre. Even my cats do it. It is even more bizarre to leap from suggesting that overlearning is necessary to a wildly wrong and completely unsubstantiated statement like:

People may not like methods like direct instruction – “repeat after me” – but they help students to remember over the long term. A class of children sitting and listening is viewed as a negative thing, yet lecturing is highly effective for brief periods.

Where the hell did that come from? A scientist should be ashamed of such unsupported and unsupportable tripe. It does not follow from the premises. We need to practice, so extrinsic motivation is needed to make students learn? And play is not essential? Seriously? Such idiocy needs to be stamped on, stamped out, and stamped out hard. This is a good case study in why neuroscience is inadequate as a means to explain learning, and is completely inadequate as a means to explain education.

In the interests of fairness, I should note that brief lectures (and, actually, even long lectures) can indeed lead to effective learning, albeit not necessarily of what is being lectured about and only when they are actually interesting. The problem is not lectures per se, but the fact that people are forced to attend them, and that they are expected to learn what the lecturer intends to teach.

Activity trackers flop without cash motivation – Futurity

http://www.futurity.org/activity-trackers-motivation-1281832-2/

Another from the annals of unnecessary and possibly harmful research on motivation. Unsurprisingly, fitness trackers do nothing for motivation and, even less surprisingly, if you offer a reward then people do exercise more, but are significantly less active when the reward is taken away…

…at the end of twelve months, six months after the incentives were removed, this group showed poorer step outcomes than the tracker only group, suggesting that removing the incentives may have demotivated these individuals and caused them to do worse than had the incentives never been offered.

This effect has been demonstrate countless times. Giving rewards infallibly kills intrinsic motivation. When will we ever learn?

One interesting take-away is that (whether or not the subjects took more steps) there were no noticeable improvements in health outcomes across the entire experimental group. Perhaps this is because 6 months is not long enough to register the minor improvements involved, or maybe the instrument for measuring improved outcomes was too coarse. More likely, and as I have previously observed, subjects probably did things to increase their step count at the expense of other healthy activities like cycling etc. 

What is education for?

Dave Cormier, in typically excellent form, reflects on the differences between education and learning in his latest post. I very much agree with pretty much everything he writes here. This extract condenses the central point that, I think, matters more than any other:

Learning is a constant. It is what humans do. They don’t, ever, learn exactly what you want them to learn in your education system. They may learn to remember that 7+5=12 as my children are currently being taught to do by rote, but they also ‘learn’ that math is really boring. We drive them to memorise so their tests will be higher, but is it worth the tradeoff? Is a high score on addition worth “math is boring?””

This is crucial: it is impossible to live and not to learn. Failure to learn is not an option. What matters is what we learn and how we learn it. The thing is, as Dave puts it:

Education is a totally different beast than learning. Learning is a thing a person does. Education is something a society does to its citizens. When we think about what we want to do with ‘education’ suddenly we need to start thinking about what we as a society think is important for our citizens to know. There was a time, in an previous democracy, where learning how to interact in your democracy was the most important part of an education system. When i look through my twitter account now I start to think that learning to live and thrive with difference without hate and fear might be a nice thing for an education system to be for.”

My take on this

I have written here and there about the deep intertwingled relationship between education and indoctrination (e.g, most recently, here). Most of its early formal incarnations were, and a majority of them still are, concerned with passing on doctrine, often of a religious, quasi-religious, or political nature. To do that also requires the inculcation of values, and the acquisition of literacies (by my definition, the set of hard, human-enacted technologies needed to engage with a given culture, be that culture big or small). The balance between indoctrination, inculcation and literacy acquisition has shifted over the years and varies according to culture, context, and level, but education remains, at its heart, a process for helping learners learn to be in a given society or subset of it. This remains true even at the highest levels of terminal degrees: PhDs are almost never about the research topic so much as they are about learning to be an academic, a researcher, someone that understands and lives the norms, values and beliefs of the academic research community in which their discipline resides. To speak the language of a discipline. It is best to speak multiple languages, of course. One of the reasons I am a huge fan of crossing disciplinary boundaries is that it slightly disrupts that process by letting us compare, contrast, and pick between the values of different cultures, but such blurring is usually relatively minor. Hard core physicists share much in common with the softest literary theorists. Much has been written about the quality of ‘graduateness‘, typically with some further intent in mind (eg. employability) but what the term really refers to is a gestalt of ways of thinking, behaving, and believing that have what Wittgenstein thought of as family likenesses. No single thing or cluster of things typifies a graduate, but there are common features spread between them. We are all part of the same family.

Education has a lot to do with replication and stability but it is, and must always have been, at least as much about being able to adapt and change that society. While, in days gone by, it might have been enough to use education as a means to produce submissive workers, soldiers, and priests, and to leave it to higher echelons to manage change (and manage their underlings), it would be nice to think that we have gone beyond that now. In fact, we must go beyond that now, if we are to survive as a species and as a planet. Our world is too complex for hierarchical management alone.

I believe that education must be both replicative and generative. It must valorize challenge to beliefs and diversity as much as it preserves wisdom and uniformity. It must support both individual needs and social needs, the needs of people and the needs of the planet, the needs of all the societies within and intersecting with its society. This balance between order and chaos is about sustaining evolution. Evolution happens on the edge of chaos, not in chaos itself (the Red Queen Regime), and not in order (the Stalinist Regime). This is not about design so much as it is about the rules of change in a diverse complex adaptive system. The ever burgeoning adjacent possible means that our societies, as much as ecosystems, can do nothing but evolve to ever greater complexity, ever greater interdependence but, equally, ever greater independence, ever greater diversity. We are not just one global society, we are billions of them, overlapping, cross-cutting, independent, interdependent. And there is not just one educational system that needs to change. There are millions of them, millions of pieces of them, and more of them arriving all the time. We don’t need to change Education: that’s too simplistic and would, inevitably, just replace one set of mistakes with another. We need to change educations.

Address of the bookmark: http://davecormier.com/edblog/2016/10/24/planning-for-educational-change-what-is-education-for/

A Devil’s Dictionary of Educational Technology – Medium

Delightful compendium from Bryan Alexander. I particularly like:

Analytics, n. pl. “The use of numbers to confirm existing prejudices, and the design of complex systems to generate these numbers.”

Big data. n. pl. 1.When ordinary surveillance just isn’t enough.

Failure, n. 1. A temporary practice educators encourage in students, which schools then ruthlessly, publicly, and permanently punish.

Forum, n. 1. Social Darwinism using 1980s technology.

 World Wide Web, n. A strange new technology, the reality of which can be fended off or ignored through the LMS, proprietary databases, non-linking mobile apps, and judicious use of login requirements.

 

 

Address of the bookmark: https://medium.com/@bryanalexander/a-devils-dictionary-of-educational-technology-1c3ea9a0b932#.aqn3aqsho

Scraping Google Scholar to write your PhD literature chapter

Gephi diagram of data created by BibnetThis looks really excellent – it scrapes Google Scholar, starting with a search that reveals work you already know about and that you think is significant. From those search results it generates an exportable Gephi map of authors, subject/disciplinary areas and links between them. Basically, it automatically (well – a little effort and a bit of Google Scholar/Gephi competence needed) maps out connected research areas and authors, mined from Google Scholar, including their relative significance and centrality, shaped to fit your research interests. Doing this manually, as most researchers do, takes a really long time, and it is incredibly easy to miss significant authors and connections. This looks like a fantastic way to help build a literature review, and great scaffolding to help with exploring a research area. I see endless possibilities and uses. Of course, it is only as good as the original query, and only as good as Google Scholar’s citation trail, but that’s an extremely good start, and it could be iterated many times to refine the results further. The code for the tool, Bibnet, is available through Github.

Address of the bookmark: https://mystudentvoices.com/scraping-google-scholar-to-write-your-phd-literature-chapter-2ea35f8f4fa1#.y43s1qg4l

Multiclick

This is great fun and quite fascinating – do try it out. You get to click on a rectangle, then see where other people have clicked – many thousands of them.

This system is incredibly similar to part of an experiment on collective social navigation behaviour that I performed over ten years ago, albeit mine was at a much smaller scale and graphically a little coarser, and I deliberately asked people to click where they thought most other people would click. What’s interesting is that, though I only had a couple of hundred participants overall, and only just over a hundred got this view, the heat map of this new system is almost exactly the same shape as mine, though the nuances are more defined here thanks to the large numbers involved.

In my experiment (the paper was called ‘On the stupidity of mobs’) this was the control case: the other subjects got to see where others in their group had previously clicked. They did not see the clicks of the control group and did not know how later subjects might behave, so finding the most popular point was not as trivial as it sounds. I was expecting stupid behaviour in those that could see where others had clicked but it was not quite so simple. It appeared that people reacted in three distinctly different ways to seeing the clicks of others. About a third followed the herd (as anticipated) and about a third deliberately avoided the herd (not expected). About a third continued to make reasoned decisions, apparently uninfluenced by others, much as those without such cues. Again, I had not expected this. I should have expected it, of course. Similar issues were well known in the context of weighted lists such as Google Search results or reviews on Amazon, where some users deliberately seek less highly rated items or ignore list order in an attempt to counter perceived bias, and I had seen –  but not well understood – similar effects in earlier case studies with other more practically oriented social navigation systems. People are pretty diverse! I wonder whether the researchers here are aiming for something similar? It does offer the opportunity to try again later (not immediately) so they could in theory analyze the results of the influence of others in a similar way. I’d love to see those results.

Address of the bookmark: http://boltkey.cz/multiclick/