Category: Online Learning
MOOC Datamining and Plurality Video #EDCMOOC
I finally got around to rolling through the Plurality video, a short clip which intermixes a fair bit of Jason Bourne themes with a very low crime and highly secured, surveillance society. Its the proverbial sacrificing liberty for security thing taken to a much wider audience than body scanning folks at airports.
On the outset, I was thinking… ok, cool on the Jason Bourne themes, but not cool at all the DNA surveillance thing. Not cool at all, a society which feels a five sigma approach to justice and everyone is sampled is a good thing. Uber cool on the future doing away with the dog slow polymerase chain reaction being replaced with something way faster as a means of amplifying DNA strands.
I then went into the forums and did some poking about. It turns out the allure of lower crime is pretty high on folks radars, so much so more than a few seemed to be ok with giving up liberty to get there. I guess this isn’t anything unexpected in light of the TSA, Patriot Act, NDAA, stand your ground laws. etc… but its still disturbing. Ultimately though, EDCMOOC is about education, not politics so I had to ponder how the video and education might be connected.
Credentialing is a big deal, consider coursera’s signature track and as of the last weeks announcements that a few courses when combined with a proctored exam may qualify for transferable college credit. The thing is, any security system can be gamed, provided there is enough time and/or finances to do so. Signature track wont be perfect, just as the F2F world isn’t either where one student takes another’s tests, or even a entire course of study for a few bucks.
One possible security scenario is not biological DNA as presented in the above video, but ones digital footprints. Consider the potential of datamining every post for every student within the MOOC universe. Assume that textual, interactive, spatial, and time domain patterns will emerge, that if the tech proves out, might serve as a means of ID and/or even the credentialing students.
Such would provide for a greater sense of security that MOOC land credentials were authentic, perhaps even more so than a lot of weak F2F credentials.
Yet, freedom dissipates. Imagine the chilling effect on risk taking for fear of compromising ones identify, and/or possibly credentials for an entire course of study.
The thing is… how many would sacrifice educational freedom for a little bit of credentialing, especially in a world filled with credential lust?
Constructivism and DI its not an either/or #FOEMOOC
Constructivism is a big deal in education, and if done well, it can work wonders for a much wider sector of a student population than direct instruction. Alas, just like DI can bore students to tears, reach only a limited population, or fail to achieve depth of learning, constructivism done poorly sucks as well, just in a different matter.
Kirschner, Sweller, (Clark 2006) comes down pretty hard on PBL, minimally guided constructivism and related as being drastically inferior to DI. Having personally experienced far too many PBL failures as a student, I tend to gravitate toward his thinking. In poking around MOOC land this week, I see working memory overload, and even a bit of negative learning playing out in a huge way across a vast number of cMOOCs. I see folks getting snowed under left and right… at least with a stage on a sage with a firehose, you at least know where its coming from, and can often modulate the negative effects.
A big problem I’ve found with PBL, is that you often don’t know the negatives, and may not know them until its too late. Getting the foundation knocked out from under you in failed PBL is as impossible to recover from as being 6 weeks behind in a DI course… unless the course is truly 100% async, and you go back to nearly day 1 and restart, provided you still have enough motivation to do so.
On the other hand, I’m taking a Python course via Edx, which is mostly DI, but there is a constructivist component which rolls in and out of the mix from time to time. The homework nasties which drive one to other students to pound things to death is where some amazing learning goes on. I doubt I’ll ever forget that:
when varA=’32’ that (varA>85) evaluates to true
Had it not been for the constructivist aspect of the course, I likely would not have picked up on that, likely even not have caught it after being burned a time or two.
At this point, I tend to think the big deal issue is not that 1 is good, and the other is bad, but more so how one integrates and applies PBL and/or DI to course design is really the key. Its far too easy to pick one and completely ignore the other to one and ones students peril. There is also an issue of how to go forward as concerns the course objectives, and perhaps even more importantly what the individual students objectives are in relationship to them. A mismatch here, especially if its not blatantly obvious, can have some pretty ill effects.
Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: an analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based learning. Educational Psychologist, 41, 75-86.
Whats all this Credentialing Hoopla
I realize that credentialing is a big deal. Employers use it as a filtering tool. The public uses it as a crutch to determine the authority or lack there of for a given source of information. Students use it as a means of competition for jobs, for schools, even just to poke at one another. The thing is, credentialling is often gamed, whether covertly (cram for the exam and forget) or overtly (pay someone to take an exam or write papers for you).
In general, credentialing its a feel good thing that society seems to thrive on, but its far from the robust system far too many folks believe it is. For the most part, credentialling is not about mastery or competence, but a matter of hoop jumping, monetary transfer, and a whole lot of wishful thinking.
Trade credentialing is likely the most useless. I wrote on this last year. http://www.ronamundson.com/blog/2012/02/11/credentialling-is-bogus/
A scarier case is that a friend of mine is authorized by his state to practice as a professional engineer. Ie., he can sign off on projects where thousands, or even 100’s of thousands folks lives could be at risk if he is wrong. He was authorized based upon Carnegie units at an accredited institution, a given number of employment hours in a related field, and passing some difficult exams based upon his knowledge of undergrad coursework, most of which had little to do with real world engineering practice… and a chunk of change to a test firm and another chunk to his state. Even he himself believes he has little competence, or mastery of engineering in regards to safety. For obvious reasons, he does not practice as a professional engineer.
On the other hand, from an engineering perspective, albeit not within the confines of public safety, consider the following individual. http://en.wikipedia.org/wiki/Jeri_Ellsworth
While credentialing via a high profile digital footprint worked incredibly well for her, the high profile part will only work for a few. Its quite likely that digital foot prints, even without the high profile bit, will work for a significant number of people. On the other hand, what about the competent individual that finds digital foot prints in any form problematic? What about those who just can’t go down such a path for reasons of personal safety?
If MOOC’s and other self directed learning paths end up decoupling credentialing from mastery, the door should be open to a better path than the current norms… but society is fickle, and the how part is riddled with potential trouble areas.
#FOEMOOC Projections equate to Significant Diversity in Course Perception
My Expectations for #foemooc
My expectations for #foemooc were to learn best practices for implementing online courses, not the mechanics of how to take a set of face to face direct instruction course materials, tweak them a tiny bit, and transfer them to an online LMS. Dr Wirth’s book aligns with my expectations, as did many of her recommended readings and supplemental materials. A class from her should be a goldmine, it wasn’t for me… A few other folks professed differing opinions, many focused on the course logistics, rather than the underlying educational philosophy.
Sure, there were organizational, and technical issues, anytime you try something new, they are bound to happen. Even more so, when you can’t really test scaling, and you are out of your sphere of expertise tech wise, when things go wrong, they really go wrong. She took on a huge risk, and it blew up… this happens to every risk taker sooner or later… been there done that myself and I will do it again, just in different ways. The thing is, with each crash and burn, you pick up more tools, and better ways to handle things when something goes south. From my point of view, had Dr Wirth cranked out even a crude, but enthusiastic Flip video on day 1 to reassure the class, given a few pointers on getting rolling and restating her end goal, folks would have grumbled, but it might have served to take the focus off of the tech issues.
The big disconnect and the courses down fall for me, was the “do as I say, not as I do” element. The dramatic differences between her writings, advocacy and practice drove me to distraction. Even her welcome to class video had it followed the precepts of her own book and week II student assignments, it could have been huge. Within her book you get a sense of her philosophy and its this majorly grand thing, its way cool, its contageous. Or at least, this is how I, a reader in isolation percieve it.
What I wonder about that, how much I am projecting my own philosphy onto her text. In thinking about this for a few days… and trying to square more of a DI model into the mix, for some strange reaon, it seems possible.
As concerns my own philosphy, consider the following:
Most students dont intially like constructivism, its messy, often hard, but in ways other than what they expect, and it usually requires a lot of critical thinking.
Its a bit disheartening to see a fair bit of commentary on the #foemooc site where students seemingly want to throw constructivism out the window and embrace direct instruction. I present some objections and my responses below.
We dont need nor want to discuss the material, the lectures are fine as is. (I don’t see how you can build much or even any knowledge off of them, but ok )
Reopen the course and lets get rolling, it was working ok. (I did all the vids, the reading, the supplemental readings, and the quiz in a few hours on day 1 to be prepared for things to roll, they never did in my group, perhaps you were very fortunate? )
We dont need groups, We can do it on our own (Ok, you can, but what about those who can’t, I know I couldnt, at least not to get the real benefits of the course )
We need instructor feedback a great deal as this is a foundational course. How is an professor going to grade 40,000 quizzes in a week? The quizzes are stupid, you can put anything in the answer box and get 5/5. (While I agree that the quizzes were stupid, I have no idea what they were really for, other than as a means of measuring group dynamics and a bit of participation, sort of like attendence sheets / Carnegie units.)
Bottom line, a lot of folks see a lot of different points. To make this as explicit as possible, I’ve thrown out a couple of extreme examples as concerns how a specific topic might be taught within the confines of 2 vastly different approaches.
The Overall Lesson Topic and Background:
The chosen topic is employee compensation, the field chosen is automotive repair, as nearly everyone has to get their car, or a friends car repaired at one time or another. Dont worry, we wont talk about nuts and bolts, we merely want to look at how employees are paid, the resulting motorist and demotivators of a couple common methods, and how said methods have distinct advantages and disadvantages for the parties involved.
In the US, flat rate (s subset of which is bookrate) work is the norm when it comes to automobile repairs, but a few shops pay hourly, and some shops have a mixture, where warranty work is flat rate, and out of warranty is paid by the hour.
For example, a mechanic could be paid $15/hour to perform a 50,000 mile service, and if it takes him 4 hours, his pay is $60. A mechanic might also be paid $50 for each 50,000 mile service they do, whether it takes them 20 minutes, or an entire days.
Direct Instruction Approach
One could create a lecture, power point deck, and some handouts explaining both types of compensation in detail, how they were are arrived at, what the pros and cons are for the mechanic, for the shop owner, for the customer and perhaps even for the vehicle brand. Such a lecture is likely to take a full class period, assuming a little bit of time for q-a, perhaps some worksheets to fill out etc. This is direct instruction in a classroom. One could morph this to an online environment by adding some cool bells and whistles, such as the tools demonstrated in this video.
One could design a class period around the following 4 minute video (bare in mind, this is a real body shop guy working in a real shop). Its not intended to be an educational video, its not highly polished, and the language is likely to offend some folks. On the other hand, it works, the body shop guy really knows how to tell a story and has built up quite a following.
In and of itself, the video is entertainment, does have a some educational value, and is good for laughts). It might provoke a few insights, but it might not.
Building a class period under a constructivist model
However, if one were to build a class period around the video, by selectively seeding a few bits for discussion, and monitoring said discussion / leaning it one way or another as needed, some great things are likely to come out of it. In fact, after 45 minutes, there is a much greater likelihood that your students will really understand the material than had the same 45 minutes been devoted to a lecture and a few student questions. Its also quite possible the students will hit on something that you never considered in a DI lesson plan. (How might one applying that employee compensation model in other areas? Does this have a positive or negative community effect? What about minimum wage laws? How stereotypical is this? etc)
There is no doubt the work involved to package a class around such a video is likely to be quite a bit greater than the work to simply prepare a lecture / slide deck, and add some cool bells and whistles for an online class. Also, one must be very much aware, that if constructive packaging is done poorly, the outcomes are not likely to be all that positive. Initially students are likely to view this as merely a funny youtube vid. It may take a bit of leading / swaying to really get them thinking. Also, if the students have not been exposed to this format before, there is likely to be some fussing about… or being pragmatic rather than idealistic, the probability of some serious complaining is really high.
Diversity of Projections
What I projected onto the #foemooc, were the tools such that we could take videos like the body shop one above, either of our own, or of others via licensing etc and build a package around it in an online fashion in order to facilitate a constructivist model. I could not fathom merely making a powerpoint, videos, and some cool bells and whistles to try and teach the same material as effectively.
Other folks may have walked in thinking… ok, what do I do to morph my non-virtual world content into an online course, and what unique tweaks must I apply to make it work online. How do I build videos, collaboration tools, how do I handle the course logistics.
The expectaions and projections of the 2 approaches are like night and day.
Converting Short Coursera Videos into Audio ( mp3 ) Files
Coursera is a great platform for online learning. The videos are intentionally short for purposes of optimizing the learning experience. However, from a review perspective, its would be really nice to have them available to listen to casually, say when one is commuting to work, exercising, or working around the house.
Sadly, downloading mp3 files of the lectures from most Coursera courses is not an option, so a little bit of work is needed. In addition, it may be useful to have one large continuous file, rather than 5-10 individual files, depending upon the capabilities of your audio player.
I use an an audio editor called Goldwave* to do mp3 conversions and consolidations of mp3 files. Unfortunately, unless one has other uses for such a program, the cost may not be justifiable. That being said, it appears one might be able to accomplish the same thing with the open source audio editor program Audacity. Alas, I have never used Audacity so I dont know for sure, but it looks possible.
The following is the procedure I use with goldwave to convert a Coursera video to mp3 for use on my audio player.
Download the lecture from Coursera.
Load the mp4 lecture file into goldwave and then using the LAME encoder, convert it to an mp3 file. Be sure to use some form of descriptive title so you can keep them in order.
Set the album field to the course name via editing the mp3 file tag. If your mp3 player supports playlists, you could also set the playlist tag to keep the files in order. Goldwaves tag editor is text only… and for a lot of folks thats all they need. I happen to use the Sony Sound Organizer software as its what my mp3 player is happiest with. There are a number of mp3 tag editors that could do this as well, or likely even better.
Upload the mp3 file to my player
The following is the procedure I use with goldwave to make one large mp3 file of the entire weeks lectures.
Download the entire weeks lectures from Coursera.
Load the mp4 lecture files into goldwave and then using the LAME encoder, save them as mp3 files. Be sure to use some form of descriptive title so you can keep them in order. Unless one has a powerful computer with a lot of ram, its best to do the mp3 conversions a single lecture at a time. **
At this point, you will have a weeks worth of lectures in mp3 format.
Next, I load each mp3 file into a goldwave starting with the 1st lecture in the series, and then cut and paste them in sequence into one large continuous track in the first lecture file. Again, unless one has a powerful computer with a lot of ram, its often best to load a sequential file, and cut and paste, and unload said file rather than trying to load all the lecture files at once.
At this point I now have one continuous file that comprises the entire weeks worth of lectures.
I then save this track with an appropriate descriptive name as one large continuous mp3 file.
Finally I set the album field to the course name via editing the mp3 tag. There are a number of mp3 tag editors that could do this as well, or likely even better. I then upload it to my mp3 player.
If one has an vehicle like my fifteen year old truck, you likely know you can’t play mp3 files via legacy CD players. All is not lost, its pretty easy to use Nero-Express to take your mp3 lectures, and burn them to an audio CD. Granted, you can’t put anywhere near as many lectures on an audio CD as you could with an MP3 CD, but it is an option for folks with older technology.
Lastly, one should keep in mind, that while doing any of the above should be ok for personal use, sharing said audio files, or worse selling them would be in violation of copyright law.
*Goldwave may look scary as it can do some amazing things. It does have a fairly steep learning curve to do the really cool things… but mp3 conversions are only a tiny fraction of its capabilities. After you convert the first few lecture videos to mp3 files, it gets a lot easier.
** I realize that quality can be lost by converting all the lectures to mp3 first, rather than saving them as raw audio files to connect them, only to resave the the large continuous track back to mp3. The reason I do this is to keep the load on my ancient laptop reasonable. Personally I dont even notice the difference, but folks with audiophile ears might.
On MOOCs, Scaling, and Teaching Models #FOEMOOC #EDCMOOC
Direct instruction (DI) scales with ease, the only real limitations are imposed by the means of assessment. Ie, if one chooses essay style exams, you need a huge amount of resources to assess them, where as if you use multiple choice exams, scaling is generally pretty simple. This is one of the reasons that Khan, EdX, Coursera, have such a following amongst self directed, highly dedicated, or in general top students. Its also fairly easy on the instructors part, as its a modest investment to build and manage a course, resource requirements after the initial course development are generally low, and scaling is a no brainer for the most part.
Direct Instruction Limitations
On the other hand, direct instruction does a poor job of reaching the less than motivated student, and/or those who are motivated, but for one reason or another run into lots of trouble. I’ve often heard a figure that around 30% of a student population will do well under a DI model with minimal hand holding. I dont have stats to back this up, but it seems reasonable based upon my experience. There is also the issue of depth vs breadth. DI can cover a lot of ground quickly, but rarely reaches deep into the brains of all but the most dedicated students.
The Constructivist Answer
Thus, the focus on constructivist models of learning. If such an approach is done well, its reasonable that it can reach not just 30% of a class, but somewhere nearing triple that number. In addition, rather than just a few students reaching the higher levels of mastery and application, its likely a majority of students will be able to apply what they learned. Even more impressively, the number of folks who can apply the material well after a course of study has been completed increases multifold.
Constructivism as a whole doesnt scale
The big problem is that constructivist models of education dont scale anywhere near as well as direct instruction. A number of folks have suggested there should be an upper limit as to the number of participants, often times in the vicinity of the Dunbar number. If such is the case, then a MOOC operating under a very constructivist model is an impossibility by definition.
I dont believe there is an upper limit to constructivism, but there are parts of constructivism that are group size limited. In other words, in order for a limited form of constructivism to work, some positive attributes and/or expectations will need to be put aside. The sad part about this, is often the attributes which dont scale are often the areas which often provide the greatest benefit to the students having the greatest difficulties. If taken too far… a highly stripped down constructivist model on a large scale will have the same problems as a DI course, but with significantly less content.
Three MOOC examples
Consider DSDMOOC as a full bore direct instruction model of learning
- New technical concepts and vocabulary can overwhelm students for whom this is their first exposure.
- There is little wide scale collaboration. Rather the discussions are dominated by well versed highly technical individuals coming from very diverse backgrounds.
- Experts often throw out half baked ideas as seeds for discussion, and things get frenzy. Its great as a gearhead myself, but I could see it being a bit intimidating for the novice.
- Smaller less experienced fish occasionally throw out ideas, and they are welcomed and discussed in depth… the tendency is to go a bit too far in depth which may not be all that helpful.
- The professor is truly a guru, he explains things quite well, and its very easy to be motivated to do more… but I understand the background well enough to get excited.
- The focus of the discussions is very technical and gearhead oriented. Topics like the emotional responses to sound get overshadowed by technical analysis, and hw/sw issues. About half the lecture content of week 1 was on emotional response / philosophy, but you’d never see it from the discussions.
Consider EDCMOOC as using a somewhat stripped down constructivist model of learning.
Active participation rates are quite low in light of the massive number of folks enrolled.
Information overload can overwhelm students who lack experience in dealing with it.
The massive nature of the course results in a lot of duplicated revelations rather than the uniqueness and diversity of views and discussion of such that occurs in a much smaller environment
Unless you have quite a bit of confidence, or foolishness to begin with 🙂 , few want to throw out half baked ideas for further development.
There is no time or even tools for robust team building and/or emotional connectedness amongst peers thrown into a very busy and churning ocean.
Previously big fish in small ponds may feel intimidated in a vast ocean, but what happens to the smaller fish from small ponds.
The smaller fish encountering difficulty are often left behind to flounder.
Consider FOEMOOC as a attempt to implement a full bore constructivist model on a large scale. Rather than its failure, (it was completely pulled on day 6) consider the following utopian ideals.
Imagine if directed team building had worked… that 500-1000 teams of folks willing and ready to participate formed in a day or two. (I’m not idealistic enough to think that all 40.000 registered students were willing and ready to participate… real life happens.
Imagine if peer to peer learning was going full bore. The larger fish captured great depth as they assisted the smaller fish, and the smaller fish were not left to flounder.
Imagine if somehow emotional connectedness was fostered in just a few days, such that all felt they were in this together.
Imagine the cross pollination as the online curricula / presentations were developed.
Imagine developing your own online curricula that worked not only for the top students, but for the vast majority of students, even for some who were not motivated at all.
On flight training and educational theories (its not just for pilots)
Over my many years of teaching folks to fly… the ed psych stuff and models of learning dont really help a whole lot as concerns the top students. One could take almost any type of curricula / instructional model, and the best of the best will do well with it. I think of the premier aviation universities, where most of the less talented students wash out long before they engage in heavy duty flight training, and the resulting impact on the instructors that come out of those programs. They’ve never really had massive challenges in teaching like they will see should they get involved in other than another university or military operations. This might be one of the reasons instructor churn rates are so high… beyond time building that is.
Where the big challenges truly lie are with motivation in general and students who hit roadblock after road block. This is where the educational theories, laws, and models prove to be of the greatest value.
Learn by doing??? Transformers #FOEMOOC
A common theme running through my edu focused mooc classes was greater depth of learning based upon being actively involved with the task at hand. Ie if you want to learn about something jump right in and do it, your instructor will give you some nominal direction, but the greatest depth of learning will occur between you and your peers.
A Newbie engineer needs a lot of hand holding
It makes sense, sort of… a newbie EE is often times great at mathematical modeling, but other than that narrow aspect, often requires quite a bit of hand holding and mentoring. In 5-10 years, they will likely be a decent engineer, but it is a rare engineer outside of those who become specialty modeling gurus that still possess the math skills they left the university with.
The sage on the stage models
The sage on the stage math and often science models of education work pretty well at creating a basic set of knowledge, but unless one is exposed to actual usage and repetition over time, generally doesn’t sink in very far. One of my undergrad profs used to say… “In general, you really wont understand your undergrad work until you finished your masters. You are also unlikely to really understand your masters work until you are done with your postdoc… and this same scenario will play out through out your lifetime.” Over the years, I’ve found this to be true time and time again.. albeit their are a few rare exceptions.
As a young guy, I had a job offer at a transformer company… the idea of dealing with a single simple component with simple math, and relatively simple construction led me to turn it down. Years later, the outfit I worked for built their own transformers from raw materials. Injection molded the bobbins, stamped the steel, wound the bobbins, applied the insulation, assembled and tested them.
It turns out if you want to make a few tons of transformers a year, they end up not being quite so simple anymore. In fact, a lot of the knowledge of high performance transformers is the sort of thing that is tightly held IP, and is only alluded to in EE course work. Its the sort of deal that when you read an IEEE or AES paper from academia, you often think, what were those guys smoking?
However… if one has been making power transformers for years, and then is asked to make an audio transformer, and doesn’t have the theoretical background, there might be trouble. Simply reading up on the task at hand, scanning IEEE and AES papers, and reverse engineering competitors units only goes so far. In all likelihood, once a typical manufacturing guy runs up against 100 differential equations in a AES paper, he will dismiss it. Rather, he will go to the lab with the knowledge he has gathered the best he could and start making prototypes. He might get lucky, or he might build a number of prototypes to get a design up and running. Things are good… but now that 100,000 units have been built, the last batches dont meet performance specs… and its back to the lab again. Some of the time, those IEEE and AES papers present some insights buried under boatloads of academic language. In other cases, academic research papers end up not being even close… and its really back to square one. This is the point where a theoretical EE background, in combination with research, in combination with experience often combine to present a solution.
Its not either or, its both, and it takes a long time.
A sage on a stage cant teach this, nor can a peer group with minimal guided instruction jumping in and just doing it. Its not an issue of one functionality to the exclusion of the other, it is simply a need for both, and there in lies three big problems. The first is intrinsic motivation to build the theoretical skills on the part of the experienced guru. The second is the intrinsic motivation to build the experience skills on the part of the theoretical academic guy. The third, is that either path takes quite a bit of time… and more so, most EE’s dont work with just one component, they work with hundreds if not thousands.
Over the years, there has been a push for greater learn by doing instruction in EE… it doesn’t cover anywhere near the breadth of a DI sage on the stage, but it is massively huge in depth. There is no doubt such has the potential to make it easier to bring in an entry level engineer, but will such an approach really deliver over the long haul. It will be interesting