MIT 6.002x Circuits & Electronics: Tries Hard, Could Do Better.

I hope this goes some way to explaining where I have been for the last few months.

MIT 6.002x Circuits and Electronics

6.002X is the online version of the respected Massachusetts Institute of Technology course 6.002 “Circuits and Electronics”. It has just completed its first run-through, and I have been one of the first "guinea pig" students taking it.

Full Term Report.

When I was at school, like most children I used to get regular school reports sent home to my parents. The first page consisted of “advice to both parents and teachers” informing them that vague and unhelpful clichés such as “could do better” or “tries hard” should be not be used and that parents should be given more useful comments. After working my socks off and (I thought) doing really rather well in my class, this eleven year-old’s very first term’s report said:

                                "Tries hard, could do better."

As you will see, it seems appropriate for this course. However, in an attempt to be less vague and unhelpful, I decided that a blog was probably the best way to get my opinions and ideas across. 

Kindergarten

I think I need to offer some context here. My first Degree is not purely in electronics, but is shared with software engineering and system development. Despite finding the systems and software aspects both interesting and useful in my career, as time has gone on I have regretted not having a deeper understanding of “pure” electronics, but there is only so much you can fit into 4 years study. So, when the esteemed M.I.T. announced that it was making one of their electronics courses available as “Open Courseware” then it seemed the perfect match for my needs. A good time to brush up on forgotten facts, and learn some new ones too.

I had already watched most of the recorded 6.002 “live”(i.e. recorded in a lecture theatre) lectures on YouTube and they seemed okay. No better than that though, and certainly not up to the quality of WalterLewin’s  MITphysics lectures. Given the prestigious name of the institution and their reputation for ground-breaking methods developed in the Media Lab, my expectation was that, as part of the transition to on-line delivery, they would have taken time to prepare a clear and thorough presentation of the subject, taking academic lectures to a whole new level.

Open Courseware is Nothing New!

Here in the UK, we have an institution called The Open University. It was founded by the government in 1969 as "the university of the air" with the aim of making high quality higher education available to everyone. One of the unique aspects of the O.U. was the use of broadcast television in order to distribute lectures – remember that in the 1960s and 70s virtually nobody had a video recorder.

Sadly, as time went on, the O.U. realised it could distribute lectures on VHS cassette, then DVD, with the result that by 2000 virtually no academic programmes were being broadcast on free-to-air, just the “fluffy” populist programmes with little academic rigour. I would really like to see the O.U.'s back-catalogue of lectures available free on YouTube or similar service. The technical quality may not be great (heck, many of them are in black and white) but the content is exceptional.

I may be an engineer by profession, but I grew up in the 70s, 80s (and 90s) watching low budget but well designed programmes, not just ones on physics and mathematics, but history, chemistry, art, languages, literature… You get the idea. Lots of things. I didn’t understand all of it by any means, but I feel it made me a better person, and it instilled in me a hunger to learn. Free, high quality education changes lives for the better. Even, or perhaps especially, the less advantaged parts of the world.

If I seem critical of 6.002x in this report, then it is from a desire to see it work, and work well, and to foster the development of much more open courseware. I know what effect the O.U. had, and the thought of what MIT (together with Harvard, Stanford and the like) can achieve with MITx is one of the most exciting things I have seen in a long time. It really does need more work though. So, here we go. The criticism.

Presentation

Professor Agarwal is to be lauded for his work in making this course available to the entire world, and I think his enthusiasm makes him the ideal person to spearhead the open courseware movement. I have to be honest though, and admit that his presentation style is not ideal in this particular context. Note that I am not talking about his accent, although I can see how people whose first language is not English may well be confused.

When I was at university (which to my horror I have realised is now nearly 20 years ago and thus several generations in technology terms) the lecturers would often use pre-prepared overhead projector (OHP) slides with spaces to work through problems. This produced a nice linear work-flow which made it easy to write clear and effective notes. The slides themselves were also available later as photocopies for students who had been unable to attend the lecture. I found it worked very well indeed both for learning and revision. Think of it, if you will, as pre-written blackboards. The lecturer doesn't need to spend time writing things down, so you can get on with the business of actually learning.

The existing real-world 6.002 lectures take place in a multi-hundred seat lecture theatre, with a conventional blackboard (chalk-board to all the non-native English speakers). This retains some linearity but can get messy, and the lecturer spends a considerable amount of time simply s-l-o-w-l-y writing things down. A phrase which takes two seconds to say and understand can take ten seconds to write.

However, rather than repeat the existing video lectures, MIT have taken the decision to go with virtual slides; basically Dr Agarwal writing on a screen via a graphics tablet. As you can imagine, each slide can get rather cluttered and confused by the end of each lecture video. The flow, rather than being in a linear sequence, is all over the place and useless forrevision.

    

The move to a web-delivered format should, in theory, enable a more carefully crafted and controlled learning experience with a good flow and clear graphics and animations. Instead, we have the lecturer writing on a graphics tablet, over a tiny area. This is exactly the same content seen on the black/whiteboard in the live lectures. The tiny area makes things get cluttered very quickly indeed, and the thick “pen” used is rather unclear. The resultant image is crammed into 674x379 pixels, as you can see above.

I appreciate that using a small screen size allows lectures to be viewed on small hand-held devices, but it compromises the delivery of those lectures, and at this stage we should not be looking at the lowest common denominator but, instead, how current technologies might be used to improve the presentation.

I’m afraid the lectures have the feeling of “oh bugger, I forgot to prepare anything, I will do it as I go along”. You can get away with this in real lectures, but for video presentation then I’m afraid it isn’t good enough.

The very first presentation I gave in front of an audience at university was hacked together in a couple of hours the night before, and I hand-wrote many of the slides as I went along. I got a C grade for doing enough, but the other students showed me that I should have prepared far better than I had. These lectures have a similar feel to mine.

With all the resources of MIT, it must surely be possible for someone to digest these lectures and present them in the form of mostly pre-made slides? Some of the slides we see are partly pre-drawn, but get cluttered very quickly (see above) and Prof. Agarwal has a habit of adding unnecessary “fluff” and comments.

The original 6.002 lectures are recorded in a live lecture theatre, the sound is generally crisp and consistent. I wish I could say the same for the 6.002x version. Not only is the sound frequently muffled and masked with rustling, but the audio levels vary massively, even within one 5-minute video. It is clear that they weren't being monitored during recording, or even reviewed afterwards.

(Not) Recorded In Front of A Live Audience

I have just been watching an interview with the instructors ( https://6002x.mitx.mit.edu/section/instructor_interview/ ) and they raise some relevant points which are largely unanswered among the “didn’t we do well” mutual back-slapping. The main one is the important role that feedback plays in lectures. Students point out mistakes, and a puzzled expression on several faces can tell the instructor that their explanation is unclear, and thus should perhaps be trying an alternative approach. This is never going to happen when recording in a booth with a monitor and graphics tablet for company.

I'm afraid that Prof Agarwal gets into the habit of repeating himself quite a lot. By the final lectures he was saying the same thing as many as four times. This is annoying, wastes time, and breaks up the train of thought. It could (and should) have been edited out. Recaps at the start of every section just aren't needed when the student can click a button and see the previous sub-lecture. When you are watching a sequence of lectures, it gets really irritating to have a recap every 5 to 10 minutes.

Rate of Delivery

In the official introduction, we are told that we should be spending around ten hours a week on 6.002x, including lectures, tutorials, lab work and homework. To do justice to the subject, ten hours is probably somewhat on the low side, especially if you are returning to formal education after years away and need to do some catch-up work on the maths or whatever (remember: you can't pause the course to revise another subject). At over 1000 pages, the course text book would take well over 10 hours a week to complete, even if you were only doing a few of the exercises.

That doesn’t seem too bad does it, ten hours in a whole week. Let me put it another way though. That is 25% of a full time job, or 1.5 hours every day.  Let us assume that we are unable to study one day. This means that we need to find three hours study time the next day in order to catch up. Sadly, real world events happen and these hours mount up, and it is horribly easy to get left behind. The chain nature of the subject means that it is difficult to cut your losses and start a new subject. The result is that it becomes impossible to catch up. Unfortunately, this is what happened to me when a family member became ill. This is always going to happen in the "real" world, more so than among a group of full-time students without young (or elderly) families to look after.

Rather than having two lectures per week and running for 13 weeks, I would run it at half that speed. Better still, why not have it on-demand? If the need for regular testing is removed (see later in this blog post) then I see no reason why a student could not progress at their own rate, and do full justice to the subject, instead of having to gloss over parts because of time restrictions. Mid- and end-of-term testing could still happen at fixed quarterly intervals, with the student undertaking them only when they felt confident to do so.

Testing, Testing, 1... 2...

Currently, testing is in two formats. Firstly, each week has a set homework which must be completed within 2 weeks of being set. Labs are effectively a second homework, with the same issue and completion dates.  So far as I can see, there little policing of homework answers on the official forum, and other web sites also contain discussion about homework, with extensive worked answers for most of the questions.

It is perfectly possible for a student with no electrical knowledge (but a whizz on Google) to be able to find the homework answers and copy them over. In fact, most discussion seems to be along the lines of “I need the answer to Hw7 Question 3”. The on-line chatter about 6.002x definitely has the emphasis on passing tests rather than understanding the subject. I am aware that there are groups of students out there who share sets of answers. The result of this is going to be a set of students who know nothing about engineering but who have an A grade in an MIT course. This utterly devalues the brand, and is something MIT need to be aware of.

I think running the course asynchronously would help reduce this reliance on copying, but I am afraid that it is going to be endemic in any system like this, which relies on automated pattern matching rather than a human being interpreting the submission.

I guess I was spoiled as a student. The emphasis was very much on understanding how to solve problems. So, if you got the numerical result incorrect, or made a mistake in one step along the way, you would still get the majority of the marks. An engineer is someone who can tell you why something works, and this testing system just doesn’t support that ethos. All it wants is a number, or a line of preformatted mathematical text.

Talking of which, the method of submitting expressions is horrible. The question setters do their best to allow for some flexibility, but the software cannot take individual style into account when checking an answer, whereas a human would not see it as an issue at all. I think it was week 5 or 6, which involved extracting MOSFET parameters via calculus. This resulted in a horrendously complex expression entered as a line of text, which I must have tried to enter a couple of dozen times before giving up (my calculation was actually correct). The LaTeX mark-up language is supported in the labs, I don't see why the homework system shouldn't support it too. Ideally the student could write the answer (and intermediate steps) in longhand on paper, but this would obviously not be machine-readable.

I noticed an excellent article on the BBC Education site concerning this issue. MITx (and EDx) seem to be looking at a form of peer-evaluation. This is an excellent idea if correctly carried out. Quite how they are going to trust a set of random (and inherently uneducated and inexperienced) individuals to score a paper, I am not quite sure. It definitely has potential though.

Because of commitments, I was unable to take either examination. I have heard several comments that people were surprised how easy it was in comparison to previously published examples.

Labs

All labs are based around a simplified web-embedded PSPICE client. This is good so far as it goes, but I was disappointed that an engineering course lab didn’t have a physical aspect to it. I understand that this would incur a potential cost on the part of the students, but I believe this could be done inexpensively with a cheap breadboard and a small collection of components. It need not be compulsory, but I am sure many students would welcome the opportunity to understand electronics at a more physical level.

Either way, I would love to see more labs, whether virtual or physical. Getting a real feel for how components work is vital for any engineer. Just knowing that an Amp or a Volt is “so much” and “that much makes a wire melt”. Too many paper engineers don’t have this sense of what is real.

Tutorials

Each weeks' lectures are accompanied by a small number of recorded video tutorials. These may cover content from the lectures, or related skills such as soldering. (Memo to Americans recording electronics videos: You know how you laugh when we call a cigarette a fag? Well we do the same when you pronounce soldering as soddering)

More tutorials please! These are the strongest part of the course at the moment. If you want to see how they might be improved, go and watch some of Dave Jones' EEVBlog tutorial videos on YouTube.

Course Text

The course set text is “Foundations of Analog and Digital Electronic Circuits” by the lecturer (Anant Agarwal) and Jeffrey Lang . At 1008 pages it is certainly no lightweight book. It currently retails for £66 here in the UK, excluding postage (approximately £3). That is US$107. Yeah.

To be fair, the entire book has been scanned in and is available online on the 6.002x web site. You may be one of the fortunate souls who can read books online with no problem, sadly I am not one of those people, and printing 1008 pages doesn’t really appeal to me.

If the book weren't the official course text then I wouldn't look at it twice. There are many better books out there, which present the subject in a clearer and more methodical manner. I would go as far as to say that basing the course on this book is holding it back.

The Emperor’s New Clothes and Education by Omission

I have seen various comments on web forums concerning 6.002x. Inherently, many are from students who were, for one reason or another, unable to attend a “real” university. There is a certain amount of comment that the lectures only cover a small proportion of the subject content. Quite often, the response to their concerns it “it is MIT, one of the best engineering schools in the world, of course it is difficult” or “the lectures give you the basic tools, you are expected to do the rest”.

There is a certain amount of validity in this, where the student is physically within the learning environment. In this case, however, this approach simply isn't sufficient.Why not just buy the book and work through the tutorials? 6.002x is no longer a physical university course where students have multiple sources of support, and the existing lectures do not provide sufficient information.

A few weeks ago, I was watching the excellent “Ask AnEngineer” live Internet video programme, where (MIT alumnus) Limor Fried was talking to (fellow MIT alumnus) Amanda Wozniak. Someone asked what they thought of 6.002x. They both expressed surprise that MIT had chosen this particular module as the first to be trialled. Amanda said she had nearly abandoned the module half way through, as the maths was becoming overwhelming, and she was really struggling. I have to say that I have a lot of respect for Amanda as an engineer, she really knows her beans, and if she was struggling then heaven help the rest of us.The maths rather overwhelmed me at the same point.

You can call stuff “a bunch of fun” as much as you like (and Prof Agarwal never shies away from doing so) but hard work is hard work and should be treated as such. If a student thinks they should be enjoying it and are simply finding it impossible, then they may well quit. Engineering is hard. Sure, it can be presented well, and explained from various perspectives, but there is a lot to learn and no amount of "hey wow guys ain't this cool" is going to make it quick and easy.

So... Tries hard. Could do better. But How?

I am not egotistical enough to assume that MIT give a rat’s arse about my opinion, but I do think they will pay attention to the general consensus. I also think that if nobody says anything then they will assume the course as it stands is perfectly fine and will continue with it in the same format. It is a good start, for sure, but it isn't fine. 4/10 at most.

Overall structure.

• Remove the examinations (for reasons of mass-cheating mentioned above). Without having a real physically moderated examination, where the student is required to attend a controlled environment, I think they are ineffective and serve only to devalue the course, and MIT’s qualifications as a whole.

• This can be balanced by having more coursework. The only “qualification” on offer should be a certificate stating that the student has attended all the lectures and passed the homeworks.

• Crucially: Deliver the course flexibly and on-demand, with no time limits. Allow the student to start when they want and to progress at their own speed.

• More of the excellent tutorials.

Technical quality must improve vastly.

Let me state at the start; Budget is no excuse for poor quality.

The audio must be clearer and more consistent in volume. If MIT’s broadcast facility won’t allow you to use their equipment then inexpensive semi-pro equipment such as the $100 Samson C01U studio microphone is excellent quality. Software such as Audacity allows the levels to be normalised within seconds at the click of a mouse. This is supposed to be professional. If the quality isn’t good enough then re-record it! Choose a good bit-rate too, there is some awful compression distortion on some parts.

I won't even call the existing quality amateurish. If anyone at MIT wants to see how good lectures and teaching sessions can be, then please look around YouTube.

The method of using a graphics tablet to write on the screen is horrible. Worse in terms of readability than the live lectures, and very wasteful of time. Pre-prepare a full set of clear slides at a good resolution. Having written a book on this subject, a set of slides should not be difficult. Yes, this takes time and effort. If you want to give a quality educational course and don't want to do the work, then let someone else do it instead. In the spirit of Open-ness I am almost inclined to make a set of animated slides myself, just to show what can be done even with my mediocre skill in presentations.

If need be – script it! This is basically a broadcast TV show now. Ums, errs, aaaahs and repetitions are utterly unprofessional. Absolutely fine in lectures, but not in this context.

Avoid clichés like the plague (yes I really did say that). Avoid calling things “a bunch of fun” or “exciting”. Interesting maybe, but not fun. Remember you are talking (partly) to teenagers who have recently discovered alcohol, sex and really loud music. Possibly even all three at the same time. Is power rail bounce or signal reflection really fun? I'm not getting all Puritan here, it just really annoys me when someone tries to hype something up which is clearly rather dull.

UK people will know what I mean when I say it gets a bit happy-clappy vicar with a guitar and tambourine. the more you try to "get down with the kids" the further away you push them.

The book. It needs to be available in print form for under £30. There are on-demand print services (not to mention dubious sources in China and India) who will print it for a fraction of that, if the will is there from MIT and the authors.

There is content missing or lacking in detail, and content which could more logically be located in a subsequent course. (For instance: What happened to bipolar transistors?) This must be the only course I know of which teaches active components before passives. I don't buy the "because it is MIT then it must be right". My alma-mater has Nobel prizes too, and they don't teach electronics in this rather muddled order.

Edit and edit again! Edit the script. Edit the videos and audio.The final few lectures on power rail bounce and signal reflection could have been edited down to 75% of the time or less, with no loss of content. At one point, Prof Agarwal said the same thing four times. Yes four times. He said the same thing four times. Isn't it bloody irritating and a waste of time when someone says the same thing four times?

Conclusion

All good experiment write-ups should have a conclusion, so here we are.

• Was it worth my time? Yes, definitely.
• Would I do it again? Not unless the video lecture content is massively improved.
• Would I do another similar online academic course? Unreservedly yes.
• Would I recommend it to someone looking to study electronics? Maybe. I suspect it would put a lot of people off serious engineering.

I really enjoyed getting my brain working again, remembering long forgotten facts and learning a lot of new stuff. I also learned that I have forgotten a huge amount of maths. It did soak up a huge amount of my spare time though, and only part of that was watching the videos and working on the assignments.

With the introduction of edX, which will effectively be covering all the MITx content now, and the promotion of Prof Agarwal, I suspect 6.002x may well change before the next offering. Hopefully this bringing-together of skills and services will raise the quality of content and delivery.

With such big names getting serious about Open Courseware, this entire field is going to get very big indeed.On a human scale, it is going to have a huge impact. Possibly even bigger than people yet realise.