Hey, that's me!

edX released some working papers on what they learned from MOOCs (and their students).

A graduate-level course in public health should attract different registrants, on average, than an introductory physics course, or a course on global poverty.

As someone who took all three courses, I want to ask why would you find this odd? Isn't curiosity normal? None of these courses assumed prior knowledge other than some high-school level math so they are accessible to a broad audience.

One interesting subpopulation is the group of people who enroll in and complete multiple courses. More than 4,000 registrants across MITx and HarvardX earned more than one certificate, including 1,912 who earned at least one certificate from each institution. Seventy-six registrants earned five or more certificates.

So our household accounts for 2 out of the 76. What about yours?

BTW, if anyone from edX is reading this, please note that you did recoup beaucoup bucks from us for these free courses. Bad Dad and I both attended Bricks and Mortar schools in the edX consortium. We both volunteered to work on our class reunion gift campaigns and give generously in part because we believe edX is a fantastic idea.

Go Bears! Go Beavers!

Bad Dad and I are now taking Jazz Appreciation and Probability together. He began playing jazz standards on our piano in spare moments. He does this even when it is not Valentine's day. Now, that's romantic. Actually, he did something even more romantic after Coursera's ModPo; that was so special, it deserves it's own post.

When I was volunteering as a tutor at a Title I school, I referred a couple of advanced students to edX and Coursera. They didn't complete every course they sampled or earn certificates. But, they did learn something and gained from the experiences. edX is great for motivated low-income kids who have access to broadband and adults who can help them in person. That's a lot of ifs. Instead of branding MOOCs a failure for disadvantaged students, we should try to help them access the tools to help improve their odds: broadband, a quiet place to study and adults who can help them in person.

CopyrightX 2014

CopyrightX is looking for a few curious and hardworking students like you.  If you create original content or derivative works based on the work of others, this will be very helpful to you.  Applications will be accepted December 13-23, 2013.  To apply you'll have to write essays explaining why you want to learn about copyright.

The toughest thing about the class for me was learning legal writing; it's vastly different from anything that I had done before.  However, learning how to decide cases based on arcane rules is a lot like abstract algebra.  You take a bunch of arbitrary rules, stir them up, and then see how they interact to produce a world that behaves in all sorts of unexpected ways.  Great fun but hard mental work!

From the assessment of the 2013 course:

We received over 4100 such applications during the three-week window in which applications were being accepted. When evaluating those applications, we looked for manifestations of intelligence, facility with English, and commitment to completing the course – but we did not privilege educational attainment or legal knowledge. Instead, we strove to select a class that would be diverse on many dimensions: gender, country of residence, age, occupation, and interests. We achieved at least the last-mentioned goal. The 500 admitted students included:

• 53% men; 47% women
• 29 lawyers; 43 persons with Ph.D.s; 177 persons with Master’s Degrees (not including those with Ph.D.s)
• a spectrum of ages, from 13 to 83
• 291 residents of the United States; 203 residents of other countries

My section of 25 included 3 working lawyers, 4 PhDs (2-physics, EE and music) and three archivists/librarians/historians.  How many times have you ever wanted to ask an archivist/librarian/historian how s/he decides what is worth preserving?  Or wanted to ask legal scholars  about ownership of intellectual property versus physical property?  This is your chance.

PS.  About half the admitted students stuck through the class till the end.  49% attempted the final (nearly everyone who stuck it out) and about 80% of them passed the final.  Interestingly, the PhDs struggled with the final more than the MA or JD students.  It could be the difference in expected writing style, but the sample size and difference is probably too small to give a definitive answer.

PPS.  I previously wrote in Notes from a RCT guinea pig that we were divided into four groups along two dimensions.  Half read a US-centric case law curriculum (much like the HLS students) and half read a broader global law curriculum.  People expected the global reading students to be at a disadvantage.  However, many students studied the readings for both groups anyway.

Half used Nota Bene, a collaborative pdf markup tool, in addition to the edX discussion forum.  Although I found Nota Bene extremely useful, they found no statistically significant difference in the exam scores between the four groups.  In other words, a sample size of 125/2 (initial sample times completion rate) lacked statistical power.  I would like to see the completion rate by RCT exposures.

Culture and the Internet

I used to wonder why the internet was so snarky (regretsy), until I took the CopyrightX class.  I learned that copyright exceptions are made for parody (making it clearly legal), while other forms of quotation are legally more complex.

Thus, if you want to appropriate someone else's work in order to expand upon it, you need to secure permission or go through some other legal hoops to gain a "fair-use" copyright exception.  However, if you just want to mock someone, then go ahead.  Copyright law gives you carte blanche!  

This supports Professor Terry Fisher's argument that the law makes culture.  Does that imply that, if we reform the laws, we can reform our culture?

In a fast-paced digital environment, it is much easier and quicker to build content around parody than by relying on other forms of fair-use and needing to stop to secure permission or pass legal review.  (Or creating your own content from scratch.)

I've seen how technology can also create culture.  But, I was so struck by the observations of Zeynep Tufekci, a self-described scholar of social movements and of surveillance, that I was moved to post.

First, read Zeynep Tufekci on protest movements and capacity problems on Ethan Zuckerman's blog.  (His posts are long, but so thoughtful and intelligent, they are worth your time.)

I feel like I don't quite understand social media yet.  At first, I thought there was something wrong with me because there are so many much younger people who purport to understand social media marketing themselves as consultants.  Why am I so stupid?  Why can't I grasp it?

Then I read Tufekci's comment.

She explains that it’s hard to do conceptual work in this space because events are changing every few months, making it very hard to extrapolate from years of experience.

Then I realized, perhaps I've just been around--collecting data and observations--long enough to realize what I don't know.  Moreover, I like to collect more data before I make up my mind.

This observation about why social media has been so influential in the Arab Spring and Gezi Park protests but hasn't brought lasting change is brilliant (I added the emphasis):

To understand these protests, Zeynep turns to Amartya Sen and capacity building, looking at those capacities, not traditional outputs, as the benefits of development. The internet gives us some new capacities, but that may undermine other capacities: we end up at base camp very easily, but we don’t know how to negotiate Hillary’s step. We can carry out the spectacular street protest, but we can’t build a larger movement to topple or challenge a government.

This is an analogy for internet-enabled activism. In talking about internet and collective action, we tend to talk about ease of coordination and community. Zeynep worries that we’re getting to base camp without developing altitude awareness – in other words, some of the internet’s benefits have significant handicaps as side effects. The result: we see more movements, but they may not have impact or staying power because they come to public attention much earlier in their lives.

Social media, as it is currently built, is geared towards quick communication and gathering followers. But it's not a platform for holding deep and nuanced debates to come to a consensus on difficult subjects with people who don't generally agree with us.  That platform hasn't been built yet.

Is technology holding us back?  Or are our collective lack of imagination holding back technology?  Is governance even possible when people can't agree on axioms?  Or when many (most?) people cannot follow logic and decide rationally?

* Photo of Taksim Gezi Park protest courtesy of Wikimedia Commons.

Blocked

Sorry, I didn't have a chance to write about Cinderelmo today.

Instead, I went to the gym, tried to catch up on MOOCs after a week away, took a tour of a local power plant and attended a public comment session regarding the power plant.

I'm trying to catch up with Coursera's Metadata class. But, I can't do this problem

because the link gives this message.

Library of Congress is closed, both in bricks and mortar and on the internet.

Sigh.

This will take more than eyeballs

I hear and read much about the competition for eyeballs to view advertising.  Does anyone else feel icky about being reduced to body parts?

I'd like to suggest a couple of rewarding and ad-free ways to spend time on the internet.  However, these will take some cognitive work.

Firstly, don't let the name of the course, Writing in the Sciences, put you off.  This course will help you right more clearly and quickly about any subject.  You need to take this class.  Seriously.  The instructor, Kristin Sainani, is justly famous for her teaching and editing prowess.  Even if you don't have time to do the writing and peer editing exercises, enroll and listen to her lectures.



I felt it was time to brush up on calculus so I signed up for Calculus One.  It's structured very similarly to "self-paced calculus", which was offered at University of Colorado at Boulder (CU) when I was a graduate student there in the 1990s, and is just what I was looking for.  But, it may not work for someone who is learning calculus for the first time and completely on their own.



--------------- Let me digress a minute.-----------------

College department funding is often tied to the number of students majoring in their subject. Some departments also need to teach "service" classes to students in other majors. E.g. physics majors need to take math classes, chemistry students need to take math and physics classes, and biology students need to take math, physics and chemistry classes. Who pays for the service courses?

There is much interdepartmental fighting over resources, which are always too scarce for the number of students that need to be taught. That's how I ended up teaching three .different. classes one semester for $800/month and completely burned out on teaching.

Service classes are chronically underfunded and CU came up with a creative and cheap response for calculus. Self-paced students used the same textbook as the traditional calculus students and were expected to master the same material. However, they didn't attend traditional lectures because there weren't enough seats in the lecture hall.

Instead, they went to a math lab staffed with at least one teaching assistant (open days, evenings and weekends) and equipped with computers that offered up practice questions (and answers). Students could do as many or as few practice problems as they liked on their own (in the math lab or elsewhere), but they needed to come into the math lab to take unit quizzes. After a set number of units, they took a mid-term. If they passed that, they earned one unit of calculus credit.

One of my chemistry students told me that, since she was a returning student and had taken calculus decades ago, self-paced was perfect. She breezed through the class in the first 5 weeks of the semester, before her other classes got serious. Then, she could focus on 3 classes instead of 4 for the rest of the semester. Another student recovering from a head injury (car accident), took 1-2 units of calculus per semester until she finally finished the required 8 (or was it 10?) units.

Yet another student said that he worked intensively on self-paced calculus when the skiing was bad and then ignored it when the snow was good.  So he got ahead in the fall, fell behind in the winter, and caught up again in the spring.  All three styles/strategies were successful in that they learned and retained enough math to pass quantum mechanics (which I taught at the time).

Students are charged only for the number of units that they passed, taking the financial pressure off. In a traditional class, students are charged for all classes (after the final drop date), regardless of whether they pass or even complete the classes.

The TAs in the math lab proctored exams and tutored students face-to-face (F2F). I think that the F2F contact is vital for keeping students engaged when the going gets rough.  I haven't figured out a better way to discern which concepts students are having difficulty grasping besides working F2F one-on-one or in small groups.

In a MOOC, students can replay a lecture segment ad nauseum, but a F2F tutor can change tack and come from a different direction when they see one approach isn't working.  Or they can diagnose what foundational understanding is missing and send the student back to learn the prerequisite(s).  F2F time is expensive and necessarily needs to be rationed in a public university.  Self-paced math was a good way to do it.

--------------- End of digression and back to MOOCs.-----------------

I wish Calculus One was available when I was learning calculus.  The lecture segments are so good--clear, yet also (mathematically) rigorous.  They aim to demystify calculus for new initiates, yet lay the foundation for more formal math analysis later on.  It's a seriously great use of MOOCs.

My only hesitation is that one really needs a F2F teacher to explain the whys and wherefores when students get stuck.  If you have a friend or a tutor who can help you get unstuck, Calculus One is perfect. If you teach math, you may want to sign up and witness some seriously great teaching.

I previously wrote about calculus for the Atlantic Monthly's website:
Lessons in Freeway Calculus

Required reading about MOOCs

Commentator Douglas Kretzman left two excellent links.   The MOOC Moment and the End of Reform By Aaron Bady is the best discussion of what MOOCs means for education in particular and society in general I have ever read. It is truly not to be missed.

Then read this tongue in cheek but deadly serious CUCFA President Meister's Open Letter to Coursera Founder Daphne Koller.

The California Master Plan is in tatters.  I could go on and on about how the promise is illusory and about the craven cynicism of the people who pretend that it still exists other than on paper.  I want to be part of the group who push back against privatization of education.

Thank-you so much for the links DK!

The CopyrightX model of online education is not scalable.  It depended on TAs that (as far as I could tell) volunteered their time in order to gain experience and teaching experience for their resumes.  It also piggy-backed upon a class happening in real time on campus at HLS and the technological assistance and platform provided by edX.

Mostly, the experience depended on a small group of students capable of surpassing hurdles such as admissions essays.  At the end of the class, we posted our LinkedIn profiles and added each other to our networks.  It turns out that nearly all of my North American counterparts had already graduated from elite universities for either undergrad or grad school (or both).  One cannot responsibly extrapolate the data from this online learning experiment to educating under-served communities.

Notes from a RCT guinea pig

I hope you all had a wonderful mothers' day.  I opened mothers' day while madly writing responses to  questions for a CopyrightX final exam.  Bad Dad went to Trader Joe's for a food run and brought me some flowers to liven up the computer desk and chocolates for caffeine and energy.

Much has been written about MOOCs, massive open online classes, but mostly by nontechnical writers.  I think that MOOCs can be wonderful for computer programming courses.  But, they are lacking for humanities and other open-ended disciplines.  If reporters only sample the nontechnical MOOCs, then they will come away with the impression that MOOCs are just an educational fad with no real substance behind it.

Nathan Heller's long article for the New Yorker,  Laptop U, is representative of the coverage Bad Dad and I have read.  He opens with The Ancient Greek Hero, a class that Bad Dad is taking very seriously and I have all but given up on, and closes with CopyrightX, a very rigorous class I just completed.

CopyrightX is not a MOOC in the sense that it is not truly open.  Potential students apply and are selected.  I'll refer to it as a MOC, though, at 500 students, it's neither open nor massive.

We also pledge that, if selected, we will put in at least 8 hours a week of study and discussion time.  To further weed out dilettantes, the online application included several essay questions about our backgrounds and motivations for studying copyright.  I wrote about my past work in data collection and curation and my sewing and blogging hobbies.  My answers were compelling enough to land me one of the 500 internet student slots.

Heller wrote: "In picking students, [professor Terry Fisher] looked partly for a range of ages and professions: his goal is to seed knowledge of digital-age copyright law among people who will apply it creatively in their own circles and work."

 Of the 25 people in my section, three held PhDs, one two were practicing IP attorneys and a handful worked as librarians and archivists for a range of federal and state agencies or nonprofits. This was not your average (wo)man on the street cohort.  Students came from Europe, North and South America and China, but the North American and European students were most active in discussion boards and at the "live" weekly recitation sections.  I'm curious to learn how many of us took the final (which is a proxy for sticking it out in the class).

In addition to the weekly lectures by the professor, we read review articles and case law, "met" with our section for weekly discussion with our TA as a moderator and watched biweekly guest lectures.  Our TA also offered weekly office hours and sent out additional study materials throughout the weeks. If we asked a question that our TA couldn't answer, she sent it out to her fellow TAs and the professor and replies rolled in reasonably quickly.

Functionally, CopyrightX was a small college seminar, taught by a graduate teaching assistant and overseen by a rock-star professor.  The TA was super generous with her time.  The professor, though not approachable directly by students, spent considerable time preparing the class and listening to the TAs and the students through the TAs.

The internet 500 (CopyrightX group) took the class at the same time as students enrolled in the bricks and mortars Harvard Law School (HLS), studying the same units at the same pace.  In discussion section, HLS and CopyrightX students discussed the same "hypotheticals", or hypothetical cases.  During the special guest seminars, both HLS and CopyrightX students could submit questions.  Other students and TAs would upvote questions on the board, and the guests answered the top-voted questions from both groups.

The CopyrightX welcome message mentioned that we were assigned to one of four groups.  I instantly wrote my teaching assistant to ask if we were taking part in a randomized clinical trial (RCT).  (I learned about RCTs in PH207x Health in Numbers: Quantitative Methods in Clinical & Public Health Research.)

My TA asked the edX staff and a response rolled in many weeks later as the question got kicked around until it found someone who knew the methodology.  I know that you are on the edge of your seat, waiting for the methodology.  Here goes.

Each of the 500 admitted and enrolled students checked off one or more of the five available time slots.  The computer assigned 100 to each time slot, based on our availability.  The 20 TAs were assigned, four to a time slot, in a similar fashion.  Then we were randomly assigned, within each time slot, to one of four groups.

There were two "exposures" and four possible combinations.  One half read US case law only (controls) while the other read fewer US case law but more international law.  One half used the edX discussion boards only (controls) while the other half also used Nota Bene PDF annotation software to collaboratively mark up the weekly readings.  I was assigned to the "double exposure" group, which studied international copyright law and used Nota Bene.  (If you haven't tried Nota Bene, give it a whirl.  I really like it.)

Come to think of it, there is a third exposure--online or bricks and mortar.  In one section, the TA let it slip that we came to consensus on the relevant section of law and precedents faster than the HLS section.  That's when I realized that they could compare us to not just other CopyrightX students, but also possibly to the HLS students.

In early June, we will be notified of our final exam results.  Undoubtedly, the performance data between groups will provide researchers with rich material.  I hope that our participation as guinea pigs will make online education better for all.

I learned as much from my fellow students as from the TA. The accreditation doesn't matter to someone in my situation (though the librarians and archivists were taking the class as part of their jobs).  Being in a very serious cohort with breadth and depth of real world experience and with true interactivity were valuable to me.   I would pay money for such courses once they move out of clinical trial--if the RCTs prove MOC efficacy.  ;-)

That said, I initially put a blanket copyright notice at the bottom of this blog because I didn't want to give away any rights before I understood them.  Now that I am more aware, I think I will change the copyright to Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported as a baseline and then put separate notices for content that I want to protect differently.

I'll also post later about what the licenses mean and how I will apply them to my blogging, sewing, knitting, software and data activities.  Stay tuned.  Watch this footer.  ;-)

Leave a comment if you think this is interesting or I should just stick to the sewing and knitting stuff.

Addendum:

Commentator Douglas Kretzman left two excellent links.   The MOOC Moment and the End of Reform By Aaron Bady is the best discussion of what MOOCs means for education in particular and society in general I have ever read. It is truly not to be missed.

Then read this tongue in cheek but deadly serious CUCFA President Meister's Open Letter to Coursera Founder Daphne Koller.

Team Antigone

Bad Dad, Little Hunting Creek and I are all enrolled in CB22x The Ancient Greek Hero. We'll take a quick tour of the entire Iliad in 2.5 weeks with close readings of selected passages.  Then we will move on to the Odyssey and beyond.

Students are divided up into sections of ~1000 participants. I'm in Antigone. Bad Dad is in Iphigenia.

Which section are you in?

Bad Dad mentions that things don't end well for either character.

Done and done!

2 final exams in one week. Add volunteer work at my daughter's school: math tutoring and math team coaching 3 times each, nutrition docent for two classes and staffing the CSA booth at the school wellness fair.  I was a stress monster this past week.

Oh, I also ordered supplies for the sets for the school play and made half of my daughter's costume.

How did I do?  Grade wise, I did ok.  Keeping my cool and not snapping at people?  Room for improvement.

CS 600x, Intro to Computer Science and Programming (with Python), allows students to resubmit homework code many times until the software passes the suite of software tests.  Thus, 100% on homework is not difficult to attain. The exams, however, are not so easy and offer more limited checks.  I passed the course even before the final so there was less stress.

PH207x, Biostatistics and Epidemiology, gave much more limited checks (only one try at vaguely-worded multiple choice problems--even for homework), so I wasn't sure I would pass.  I needed 78% on the final to attain the 85% overall average required to pass.  Fortunately, the final was not any harder than a typical homework from the latter half of the class.  Whew.

Biostats is definitely a class that starts out easy and then builds to some very deep and difficult stuff.  Some favorite things I learned:

• The visual explanation of what's going on beneath "turn the crank" linear regression.  I've never seen it presented that way before.  Why didn't someone explain that to me 25 years ago?
• What are the odds?  I didn't know and I never before cared because I preferred my statistics stated in probabilities.  But, I had never before pondered the beautiful symmetry of odds tables. Now, I will keep both odds and probabilities in my toolkit and use the one appropriate for the situation.
• The application of Bayes' theorem to experimental design.  Suppose you are interested in cancer research, but it's unethical to give people cancer just so you can study them.  With Bayes'  theorem, you can just flip the conditional probabilities around to design an experiment that you can perform without giving people cancer (or some other disease).
• Soft skills are more highly valued in the biological sciences than in the physical sciences.  One professor began .each. lecture segment with, "Welcome back everybody.  Thank-you for coming."  I can't recall any physics professor saying that*.

* The nicest thing any physics professor said was when I asked him why he always picked on me in class.  He looked kind of taken back and then replied, "Because you are my benchmark student. You are the kind of student that should be able to follow the lecture, if I am doing my job.  So, if you are confused, then I know I have to backtrack."

He further went on that he can see the confusion written plainly on my face, and the (mostly much taller) male students leave a front row seat open for me, which makes for a handy benchmark student.

Self-learner project: Friendship Paradox

Your friends have more friends than you.  It't not you, it's the friendship paradox.  Your friends are not random.  They are more likely than you to be social butterflies or they wouldn't have met you (or maintain ties with you).

One of the professors teaching MITx: 6.00x Introduction to Computer Science and Programming stated that his goal was for us (the students in the class) to learn enough to break down problems and write python programs to solve them.

In the introductory chapter of my PhD thesis, I wrote that one of the reasons for modeling is to help explain the results of physical experiments.  Sometimes, modern experiments (not to mention, nature!) are so complex, it can be hard to interpret the data.  Through modeling, we can break problems down into simpler components and then add effects in until we match the observed data.  Then we can explain what actually likely happened.

Anyway, I wrote a python program to clarify the friendship paradox because it also explains why I, with a BMI of 22.5, am often the fattest person in Pilates or spin class.  Steven Strogatz explained:

For example, imagine going to the gym. When you look around, does it seem that just about everybody there is in better shape than you are? Well, you’re probably right. But that’s inevitable and nothing to feel ashamed of. If you’re an average gym member, that’s exactly what you should expect to see, because the people sweating and grunting around you are not average. They’re the types who spend time at the gym, which is why you’re seeing them there in the first place. The couch potatoes are snoozing at home where you can’t count them. In other words, your sample of the gym’s membership is not representative. It’s biased toward gym rats.

So back to the friends problem.  I started by thinking about the shape of the distribution of the number of friends one might have.

• The distribution should be self-similar; the shape of the distribution of the friends of friends should resemble the shape of distributions of friends of friends of friends.  
• The distribution must be positive; you can't have less than 0 friends.
• The distribution should have a long right tail

A gamma distribution of shape = 2-3 looks about right.

My intuition was not bad.  In 2003, Bronius Grigelionis wrote a paper about the self-similarity of the Euler Gamma Function.  Gamma distributions are useful in modeling computer networks--why not people networks?

I sorted people by number of friends, from hermits on the left to social butterflies on the right. The number of friends for each person is in blue.  The average number of friends of friends for each person is in red.  Note that, for the first 400 or so of the 500 people, their friends have more friends than they do.

I had a heck of a time assigning friends to the social butterflies in a computationally efficient manner.  If you sort people by numFriends (taken from the Gamma distribution) from hermits to butterflies, than go up the sequential list, assigning them more friends from those remaining until they reach numFriends and having those people friend them back, you will eventually reach a problem.

There are simply not enough people leftover to fill up the friend lists for those social butterflies.  In that case, I had them friend everyone to their right and then randomly select friends from those on the left that had not initially selected them.  I assumed that those on the left friended them back out of politeness, which added slightly to the total friend count.

The distribution of number of friends for a sample population of 500 people:

The initial number of friends taken from the Gamma distribution is shown in green.  The distribution of adjusted friend counts is shown in blue.

The commands to run the program:
numf = pickNumFriendsRecur(500, shape=3, scale=5 stretch=8)
flist = assignFriends(numf)
fof = howMany(flist)

The output:
max(numf) > nsubj:  501 500
trying reducing scale to:  4.9
avg numf after iteration:  115.732
Target average friends:  115.732
Adjusted average assigned friends:  121.724
Average friends of friends:  151.410319998

I encourage you (and your kids!) to view and download the sample code from pastebin and play around with it.

So, what are you taking?

Currently Enrolled and taking seriously:

• MITx: 6.00x Introduction to Computer Science and Programming
• PH207x: Health in Numbers: Quantitative Methods in Clinical & Public Health Research
• Writing in the Sciences
• Greek and Roman Mythology

Took seriously and recently completed:

• Statistics One
• Computing for Data Analysis

Took seriously and completed in the past:

• Internet History, Technology, and Security
• Fantasy and Science Fiction: The Human Mind, Our Modern World
• Introduction to Artificial Intelligence
• Learning from Data (I dropped out about 2/3 of the way through, but that was more work than any other online course.  Plus, with fewer than 100 students completing the homework assignments, it wasn't really a Massive Open Online Course.)

Enrolled while course shopping but not completed:

Too many to list. Perhaps this explains why Massive Open Online Course (MOOC) completion rates are so low, ~10%. If they offer a "sample the course" option, they can collect more accurate course completion statistics.


I'm loving 6.00x from MIT. The content, interface and projects are excellent. If you have 8-10 hours a week to spare, I highly recommend this class.  This is the best use of the MOOC platform I've ever seen for the sciences.  In the first four weeks, we've implemented bisection search, Newton-Raphson, Hangman and Words with Friends (but really against a greedy search computer algorithm).  I can't wait to see what else we'll do in the next eight weeks.

 I've always wanted to learn more about biostatistics and epidemiology*. So far Harvard's Public Health 207 is underwhelming.  I can complete the homework before watching the often slow and confusing lectures.  Fortunately, EdX let's you speed the lectures up by increments of 0.25 between 0.75 to 1.50.  (Coursera allows [0.75-2.00].)  The CC transcripts also let you zip through review materials.

 I'm also taking Writing in the Sciences and Greek and Roman Mythology through Coursera.  I recommend WitS with Kristin Sainani for anyone that wants to write better; you need not be a scientist. Sainani has degrees in Philosophy, Statistics and Epidemiology and teaches BioStatistics and Writing at Stanford. Someone posted one of her talks in the discussion board for PH207 after a headscratcher of an explanation of life tables in the class.  I wasn't the only one confused by it.  Fortunately, a classmate posted a ppt that she found helpful and I laughed out loud when I saw Sainani's name on the opening slide.  She's an excellent teacher and writing coach.

Bad Dad and I have recently completed Statistics 1 and Data Analysis through Coursera and agree that both are very good and worthwhile.  We became comfortable with R after using it to complete homework and programming assignments for the two courses.   Stats 1 is geared toward the social sciences and Data Analysis is more useful for sciences that don't include science in their title.

However, Stats 1 is not easy.  I give kudos to Princeton's Andrew Conway for explaining the subtleties of experimental design in the social sciences.  There's less math, but plenty of mental heavy lifting.  As a result of Stats 1, this physical scientist has learned to respect well-designed social science experiments.  Conway is planning to repeat Stats 1, using the lessons he learned from the inaugural class.  I highly recommend signing up for that when it repeats.

*I had never heard of epidemiology until my senior year in college. One of my friends/classmates mentioned that she was applying to grad school in epidemiology at UC Berkeley's school of Public Health. Then my academic advisor in the Chemistry department told me that roughly one third of their graduates went on to med school, one third to PhD programs in Chemistry and the remaining third were evenly split between graduate programs in related fields like epidemiology and industry. Two mentions of epidemiology in one week!  I never forgot and always meant to go back and learn more about it later.

CS 101 has started

Coursera's version of CS 101 went live today. Iris is viewing/practicing lecture 1 materials right now. We signed up for it months ago after watching this engaging video. But, there were many delays.

From what I've seen, the extra time was well spent. The browser-based API to perform the programming exercises works well. Iris was up and running (tweaking the sample programs provided) in minutes. In six weeks, the class will cover:

Syllabus
CS101 topics are covered with a mixture of video lecture and active lab work, all in the browser:

• The nature of computers and code, what they can and cannot do
• How computer hardware works: chips, cpu, memory, disk
• Necessary jargon: bits, bytes, megabytes, gigabytes
• How software works: what is a program, what is "running"
• How digital images work
• Computer code: loops and logic
• Big ideas: abstraction, logic, bugs
• How structured data works
• How the internet works: ip address, routing, ethernet, wi-fi
• Computer security: viruses, trojans, and passwords, oh my!
• Analog vs. digital
• Digital media, images, sounds, video, compression

The class is free. It's a great diversionary activity for kids who are suffering through standardized exams this season. In addition to having fun, they might just learn something useful.

Kids 13 and up can sign up with parental permission. The class is certainly appropriate for advanced kids younger than 13, but they can only do it with an adult. (Sorry, that's the law governing internet correspondence with minors, not Coursera's policy.) I signed up using my email address, log Iris in, and then take a hands off approach while she plays around.

Thank-you Coursera, Nick Parlante and Google for giving him time to work on this project.

What am I doing? I'm learning from data. This Caltech class is so much more challenging and rigorous than the AI class I took last Fall. Limits, functional analysis, proofs by mathematical induction, perceptron learning algorithms, error bounds and a heck of a lot of homework problems that can only be solved by writing your own code.

It's a time commitment, but I am enjoying the deep dive into rigor. I also like how you can view the lectures live and then ask questions, just like in a real class. Students type their questions in and the head TA reads them aloud to the professor during the Q&A session following each lecture.

Here's another computer/human hybrid teacher story:
At Virginia Tech, computers help solve a math class problem

Update:
She just finished the exercises for lesson 1 and proundly announced that she got an A+. Thank-you so much, Nick, for giving her something more useful to do on the computer than dress and groom ponies or Barbie!

AI Finis!

Remember when Iris and I signed up for the online AI (Artificial Intelligence) class as a team?

But she quit in frustration during the Bayesian algebra unit?

I carried on alone.

But she wandered by while I was working with the (logic) Truth Tables for a final exam question and asked me to explain the notation.

Then she proceeded to rattle off the correct answers by inspection.

It't not cheating because I would have gotten the correct answers in the end. Truly.

Besides, we signed up as a team.

And, is it fair that a guy who competed in all of the DARPA Grand Challenges (and placed quite highly on some of them) and already has a PhD in robotics is also taking the class?

Yes, the percentage of people in the on-line version of the class that has gotten perfect marks on all the homework assignments and the midterm is double the percentage among the students actually taking the class at Stanford, but what if we disqualify all of the ringers like the one above? From the discussion forums, it appears that quite a few professors are taking the class to learn about e-learning rather than robotics.

Or is it sour grapes because I can do complicated Bayesian statistics calculations but can't subtract 14 from 100 correctly? (I was a math major. We don't use actual numbers after the first year.)

This class was a blast. It took the full 10 hours a week that the teachers estimated we needed to put into it. But, looking back over the homework and exams, I can see how much makes sense to me now. Before the class, I had no clue even how to read the gobbleygook notation.

Serendipitously, I embarked on a new project at work where I can apply some of what I learned. What fun! And I get paid for it.

A big thank-you to instructors Peter Norvig and Sebastian Thrun and all the IT staff that kept the servers (mostly) running during this record-breaking huge class.

Related Posts:

• Trip report from the DARPA Urban Challenge Semi-Finals
• Trip report from the Carnegie Mellon Robotics Lab
• Use the search bar for our experiences with Lego and Robotics
• Particularly Robot Sumo
• What Do Automobiles and Spacecraft Have in Common? guest post for the Atlantic Monthly website

Mommy and me AI

I am not sure if this is a violation of the rules. But Iris and I signed up to take an online AI (Artificial Intelligence) class together, but using my email account only.

We are going to listen to the lectures and discuss the problem sets together. I will be responsible for writing up and emailing our homework sets. I don't know in advance whether she has a long enough attention span for this. Or whether either of us really have the time to do this in between her school and my work commitments.

We signed up for the advanced class, which requires us to do all the homework and take all the exams. If we fall behind in our homework, we can always change to the basic (audit) version of the class.

We are really excited about this. I have had a long-time interest in Markov models. Are there any other people doing this with their kids? Is this considered cheating or collaborative learning? After all, she's 10 years old--too young to sign up on her own.