Concept: A VR “Memory Palace”

Introduction – Blame it on the Black Star …

About a month back, early January, I was driving on the way to work, listening to CBC Radio Montreal, as I often do. On the radio, the announcers were talking about David Bowie – it was the one year anniversary of his death.  They talked a bit about his final album, Black Star, which came out just days before his death, and played a short clip from the lead-off title track Black Star. I had played this album a lot last January, and so, a few blocks up, at the next red light, I dug the CD out of the storage compartment between the seats and popped it into the CD player.  And the beautiful-haunting-sprawling-soaring music flowed out after probably a year since I had heard the song.

And as often happens to me when I hear a piece of music tied to an emotionally poignant moment in my life, like a key turning the tumblers of a lock, a door opened, and sweet memory gushed forth in brilliant flashes. I think everyone gets this from time to time, with songs, with smells – but for me it is an explosive experience. This album was tied to a particularly poignant period of my life last year. A week to the day after Bowie died, I got married – remarried after 5 years going it alone following a nasty divorce. This album was the soundtrack in effect of all my driving during that period, including my wedding and honeymoon. The sweetest flashbacks danced across my memory – my wife’s colorful wedding outfit and the look on her face, the sight of the flurries in Mississauga that day, the smell of the January air, the scent of jasmine blossoms in a garland around my neck, the drive along the Niagara Parkway to the grand old Victorian inn where we honeymooned, and the gorgeous room where we stayed, the quiet streets of Niagara-on-the-Lake in the snow. All of this flooded back – I could actually smell the jasmine! – all of this from a song.

The portals of memory open in such unexpected and wonderful ways. And how strange its ways sometimes. The knowledge in the pages of a textbook you were trying to learn from yesterday – information you struggled hard to understand and remember – can be lost in oblivion, despite all your efforts, but the rich details of a weekend a year before can come back to you effortlessly if the right trigger is applied.

On the nature of memory …

Memory is deeply tied to emotion and place. Places to which you have a strong emotional connection, places which have great significance to you, they stick in your memory very vividly. Everyone has places like this. Some beautiful natural spot from a wonderful vacation, the location of a meaningful life event. Probably right now you can think of such a place. If you close your eyes, you can probably see the place in your mind’s eye in rich detail. You could even close your eyes and walk around in the place. Places with high emotional significance are given preferential treatment in being placed into memory, and are easier to retrieve. Spatial information in general – information about the layout of places – in general is easier to remember than general pieces of information, but emotionally-laden spaces are particularly memorable. And it is effortless. Certain portions of our brain are evolutionarily tuned to efficiently store and retrieve spatial data, just as certain portions of our brain are tuned to store and retrieve emotionaly laden information. Compare this to random verbal or conceptual information we want to make ourselves memorize. How hard it is to push it into long term memory so that it sticks, and how hard it can be to retrieve it again.

An ancient technique – The “Memory Palace”

One millenias-old trick for remembering things takes advantage of this ease of remembering places. This trick involves building a “memory palace,” and putting the things you want to remember there. In this technique, a person goes through an exercise of visualizing some place – it could be a familiar place, and imaging placing information to be remembered in distinct places within this familiar space. Then, when it is time to remember, the person simply closes his eyes, and in his mind’s eye, traverses the space, going from place to place, retrieving the information stored in this spatial filing cabinet.

Modernizing the technique – Virtual Reality

Now, let’s modernize this technique a bit, bringing it into the 21st century with the help of technology.

Imagine building a  high fidelity 3D digital representation of one of these meaningful places that is viewable and navigable in an interactive VR (Virtual Reality) environment. A 3D scene with some interactivity laid over top. Imagine being able to place notes or representations of information or objects you want to remember within this virtual space that immerses you in a meaningful place already seared into your memory. These pieces of information then end up arranged spatially within this virtual space using an arrangement of your chosing.

This representation of place and spatially arranged information and items could be saved and accessed and modified later.

This VR tool would provide a nice visual scaffold to the classical memory palace technique, which traditionally depended entirely on the imagination.  I would hypothesize that actually being immersed in that space and “seeing” the spatial relationships would support the process significantly. In addition, I could see the physical action of virtually placing the items within the space with hand gestures could also serve to reinforce the memory.

Postscript

Interestingly, it looks like a company in England is already trying to do something similar to this, with a particular focus on using VR memory palaces to scaffold second language learning.

http://macunx.com/

 

Coursera trend – no pay, no grades

Introduction

Anyone who’s been taking courses on Coursera for awhile has probably noticed by now a recent development – that is, the move to charging students for submitting assessments for grading. Apparently, this new “feature” started in early 2016, but has become a disturbingly common trend lately. For courses that do this, you need to pay for the course to benefit from graded assignments. And, mind you, this is even for courses that use peer-graded and machine-graded assessments. A very informal sampling (probably at least 80% of the courses I’ve looked at recently) indicates that most new courses are adopting this.

This is problematic.

Coursera trend - No pay, no grade
No pay – no grade

The importance of assessment to learning

Seeing how the assessments work in a course is an important part of evaluating whether to take the course, and whether to pay for it. In many courses, the feedback from graded assessments is the best part of the course. For example, the completely free (funded out of an endowment project) Algorithms I and II courses from Princeton University use a machine grader that gives great feedback when you submit your assignments. This is where you do a lot of your learning.

Going through the assessments gives you a good sense of the quality of the course. When a course gives meaningfully rigorous assignments, you learn more, and being able to try out the assignments gives you more information about whether the course is worth paying for.

I understand that MOOC courses are expensive to develop, with estimates ranging between $30 000 and $100 000 per course. And I understand that Coursera is probably under pressure to gain more revenue to please investors. Given that, you can understand the desire of Coursera, and of  institutions to recoup their costs and investments. But they also need to reflect more carefully on what makes sense to a student.

Previously, as of 2015, the model was this: You could audit the course, including graded assignments, for free. If you just wanted to learn, you could watch videos, do readings, participate in the class discussion forums, do assignments, and get a peer-graded or machine graded mark. But, if you wanted a certificate of completion you had to pay for that. If you didn’t pay, you could still fully experience the learning, but wouldn’t have a sharable record of completion. This was the model since 2015, and it was a pretty fair model. It wasn’t quite the Utopian vision of free, accessible university level education for everyone,  but it was about the best you could reasonably expect from what is ultimately a for-profit platform.

For those who are newer to Coursera, in the beginning days it used to be even more open than this – you could get a completion certifcate or statement of accomplishment, even without paying. There was some minor uproar when free certificates were removed, but that at least I can accept as a fair compromise. “Freemium” is, after all, a pretty common model these days. It’s expensive to develop courses, not to mention host and serve them up to students. Someone has to pay the bills, and $70 – 100  to take a meaningful course  and have proof of completion you can share on LinkedIn is not a bad deal.

However, the key to freemium is that you need to offer a reasonably satisfying and enticing free product that stands on its own feet. You want a large number of people consuming the free product, because the larger that pool is, the more revenue you make from that percentage that opt in to extras like certifications. But getting an assignment graded is not an extra. That’s an essential part of the course. If you turn people away by hiding key parts of the educational experience behind a paywall, you threaten to shrink that pool of people.

I guess someone will probably make the analogy of auditing a university course. You can just pop in, sit in the back row and watch a lecture. But you’re not going to get assignments graded or sit for an exam. There’s maybe some merit to this analogy. But on the other hand, in the university course what you’re really paying for ultimately is the certified university credit (And the degree that a series of credit courses in a specialization ultimately add up to)

The problem with this move

Philosophical merits aside, there are two flaws I can see with the idea of charging to get assignments graded.

1. Any good course has a “long tail” – it’s expensive to get the course out there initially, but then it can be rolled out more easily month after month. A good course will be attractive to students over a long time, and this will give lots of opportunity to recoup costs. I see lots of good courses and programs like Stanford’s Machine Learning and Rice’s Principles of Programming in Python courses that continue to roll on with plenty of students years after they were first created.

2. Someone is going to be less likely to want to pay if he can’t get a proper representative feel for what the course is like first. Myself, I tend to take courses on Coursera in programming. Lectures are useful, but the hands-on programming assignments and the feedback I get from an auto-grader are the most useful part of any course. Applying the skills you learned from the lecture and getting feedback on what you’re doing right and wrong is a key part of the learning experience.

If a course prevents me from seeing what a course is like in this regard, I tend to drop it. Recently, I was looking into a Johns Hopkins specialization in Data Science. The first course looked interesting, but I felt compelled to drop it, because it wouldn’t let me submit even one assignment without paying for the course first. Why would I spend $80 for a course if I can’t get a proper sense of what it’s like?

And especially so when the course is the first in a specialization. A four week course is not really worth much as learning on its own. If I’d had a chance to fully try out the first course, maybe I would have gone for the specialization, and the organization might have gotten a few hundred dollars. Instead, because they were cheap, they got nothing.

I think Coursera probably believes, based on their own research, that enough people who would otherwise audit for free are going to find assessment valuable enough to pay for it if they can’t get it otherwise. And I agree that there are probably people out there for whom this will work. But I have the feeling there are more people like me who would otherwise have been inclined to pay after the fact, but now will not. I anticipate that many schools will see this play out over many courses, and will have to backtrack.

The good

This whole thing is unfortunate, because other than this, Coursera is making a lot of useful steps forward, and I want to give credit where credit is due. It’s come a long way in some respects in building both a great catalogue of training, as well as a nice, functional learning / learning management platform.

  • Many courses are being offered with great regularity, some even on a monthly basis. There is always a session going for many courses.
  • Course lengths seem to be settling into a nice average  – 4-6 weeks seems to be the norm, which is a nice size balance between having a meaningful chunk of content and being able to fit the course commitments into the life of someone with a full time job and family.
  • A lot more specializations are popping up – groupings of courses centered around a common theme or skillset grouping. This is immensely practical, and probably more meaningful as a show of learning than an isolated course. A single 5 week course on data science in Python? Ok, cool, but doesn’t add up to much beyond an introduction. But a 5 course specialization on Data Science in Python over 25 weeks that covers data cleaning, plotting and charting, machine learning, text mining, and social network analysis? There’s some value in that.
  • It is easy to switch sessions and transfer grading progress across sessions if you fall behind. I’ve found this very useful in a few courses where unexpected life business bogged me down and I got behind.
  • There is very good flexibility in terms of paying for courses. In courses where graded assignments are free without the certificate, you can complete the whole course and then later upgrade to a certificate by paying, even months after the fact. This is very nice.
  • Prices, while noticably rising over time, seem to be finding a sensible equilibrium. A few hundred dollars for 20 weeks of decent video instruction and meaningful assignments is a decent value compared to other training you see out on the market.

Larger question – the value proposition of Coursera Certifications

So really, I’m griping about one major misstep in what is, overall, a very good platform for learning. There are a lot of positive things happening.

The bigger question behind all of this is the value proposition for Coursera certificates. What value does a student get by paying for a course on Coursera? It’s still pretty fuzzy. I’ve taken a fair number of courses over the past few years on Coursera. Some on other platforms too, but Coursera has tended to be the one I have used the most. Some of those courses I have chosen to pay for, for various reasons. The first few, as a motivation to complete the course amidt work and family. Some others because I thought the courses were great and wanted to vote with my money – encouraging the people who made them and encouraging a model of continuing education I’m convinced is the future of career skills development. And finally, some others because I thought they might add something to my LinkedIn page or CV.

It’s this last one that has always been the most fuzzy and uncertain value proposition.

Do I get the sense that recruiters or hiring managers are going to see a LinkedIn or CV with Coursera courses on it and suddenly perk up? Not so much. I don’t think it’s realistically  there yet in terms of credibility.

Does a sequence of related courses add up to something interesting for the same people? Perhaps to the level of showing a certain discipline and commitment to continuing learning. Beyond that, hard to say. HR is often stubbornly conservative.

Given this reality, I think Coursera needs to do more work to justify the value of what students are paying for rather than focussing on paring back the free experience of their courses. Ultimately, that is the existential question upom which their ability to collect revenue and support the whole enterprise long term depends on. How can Coursera ensure that their certifications and specializations have value in the marketplace?