Everyone on campus kept asking me:
- "You mean you are not enrolled but still physically go to lectures with all the other Harvard students?"
- "Yes, I just walk in and sit."
- "Is that even possible? Don't they somehow stop you?"
- "No, most universities welcome listeners. And in a class there is usually around 200 students, so you wouldn't be noticed anyway. Why not just go there?"
This is the kind of conversation I was having almost every day when I moved to Harvard for a month: I had just signed a full-time offer as a Software Engineer at Facebook, but I asked to get a couple of months for "holidays" before starting. The holiday plan was to book an Airbnb for a month in Cambridge Massachussets and hang out everyday in the Harvard campus.
Turns out (to my surprise) that not a lot of people do this, so I decided to write down a sort of Practical Guide to attend Harvard classes without being enrolled to try to convice folks out there that it is indeed a good holiday plan.
The schedule I was following, blue background was for Harvard campus and green background for MIT campus (they are 20 mins walk distance from each-other)
I will organize this post in four main observations on my Harvard holiday,the first three will be of general interest whereas the fourth will be more technical, suited for reader interested in CS related courses in Harvard.
>>> Observation #1 : don't think you can get away just with listening, you will be engaged and asked to participate
I heard one professor saying something on the line "if we [teachers] as educators don't succeed in making you involved in the class, then you might as well go and take a online course and it would not make a difference". Hence, the first fun fact of being a listener is that you will be engaged and asked to actively participate in the class like all other enrolled students, and this could take you a bit off guard. Let me tell you about what happened in my first class.
My first day in Harvard, I made sure to sit in the back of the class in order not to be noticed. I didn't know what was waiting for me
When I got in I was a bit intimidated, I wasn't completely sure whether I could sneak into classes and just sit. The class was about ethics in Machine Learning. The professor was Finale Doshi-Velez , a fairly known woman in the ML field that I rememberd from some of her TED talks. She was explaining how searching on Google for people with different skin colors will yield different results, discussing whether Google search engine is racist or not. She suddently says something on the line "now discuss with the person next to you", and I of course retreated in the back of the room avoiding to speak with anyone, I was still pretty afraid of being kicked-out at any moment. So I sit in the back of the room, and the whole class is starting an engaged discussion in small groups, when someone it next to me and asks me "What do you think about the topic?". I turn around and guess what, is the professor herself that saw me not speaking with anyone and just walked all the way to the back to the room to engage me in the discussion!
This woman is a renowed Harvard professor and TED talk speaker, and when she saw I was not participating into class she just came all the way and sit next to me to get me more involved in the conversation.
To my surprise she didn't ask me "Who are you and what do you do here" but she actually initiated a pleasant and stimulating five minutes conversation only between us two about the racist-bias of Machine Learning algorithms, and how to eliminate that bias forcing a prior into the model.
As I went to more and more classes I actually started having more and more fun with this engagement thing, and I started actively asking questions, participating in labs and group activities and even stopped to speak with professors after class, tell them observation I was having during lecture! In all of this process I never met any sort of resistance by either students or teaching staff on my illegitimacy of being there.
>>> Observation #2: lots of Harvard students have mental health problems, but when you break the social alienation barrier they are actually super nice
The question I got most from the outside has been Are these Harvard kids smart as they say?. Well this is a non-trival question, given the ill-definied concept of smart across our society. What I can claim tough, is that I felt from the first day an unusually weird vibe all throughout the campus. This feeling got actually confirmed by other Harvard people I got into contact with, and most of them articulated it as a strong social isolation, rooted in a painful blend of hypostor-syndrome and early-stage clinical depression.
If you just quickly tour Harvard Campus you will see an extremely wide-spread campaign "Not everyone feels included, let's find out why", it's Pulse, a university wide inclusion survey related to mental health and depression.
I don't like unjustified sensational claims, so I will report here some facts from a study conducted in Harvard Economics Department in 2018: Graduate Student Mental Health: Lessons from American Economics Departments
- About 18% of students are experiencing moderate to severe symptoms of depression and anxiety. The comparable national rate for depression is 5.6% and 3.4-3.6% for those aged 25-34 (Kocalevent et al. (2013)). A study of the German population found the comparable national rate for anxiety to be 5% (Lowe et al. (2008))
- 11% of students (56 people) reported having suicidal thoughts on at least several days within the last two weeks.
- The prevalence of depression and anxiety symptoms among Economics PhD students is comparable to the prevalence found in incarcerated populations.
- Loneliness and isolation are major issues. The average Economics PhD student feels considerably lonelier and more isolated than a retired American.
My own experience was that trying to socialize with Harvard and MIT people is sensitively harder then people off-campus, and took me a long series of failed attempt before I couldn't actually go beyond the cold 'What is this class about'. Observing the social dynamic of my interactions, I estimate a population of around 30-40 strangers approached on-campus during my month in Harvard, and most of them (70%) just didn't wanted to engage in any conversation at all after I asked them technical questions about the class. (Warning for the reader: this could also be justified by the fact that I am a weirdo!)
From left to right: me playing piano, me playing with chairs in Smith Building, me excited because of food and social aggregation, me pretending to be serious looking at Jupyter Notebooks
At the end my persistence won my weirdness and I actually penetrated the social isolation layer and I met a super nice group of people from CS109B (Eunice, XiHan and all the others <3). Having a group of friend on-campus sensibly improved my living-standard: I started going into labs, study rooms, Harvard canteens and other places of social aggregation commonly reserved exclusively for Harvard students. If you are doing the same holiday plan I highly recommend you to invest time in creating social bonds with enrolled students on-campus.
As a side note, I asked my new friends their opinion about social isolation at Harvard, and they agreed with my perspective. They also introduced me to the concept of click, that is the local slang to define how social groups are highly exclusive in Harvard and people tend not to aggregate with people not from their social group.
>>> Observation #3: the Harvard learning experience is surprisingly stimulating and fruitful, an order of magnitude more than online courses
While discussing with some of my Romanian friends whether to embark in this holiday, the main counter-argument on coming to Harvard was: "I assume you could take all of the courses online and it would not make a big difference in term of relative knowledge gain." Well, back then I agreed with this statement (being a Coursera aficionado), but I have been happy to change my mind about this.
The Smith Building, one of the 24/7 study areas open to Harvard student, definitely a fancy place that would encourage you to spend there your nights.
What I gave me personally a huge boost in my learning experience was the human-to-human collaboration aspect that was highly stressed in all the classes I took: most of my study time was spent on completing group assignments with other students enrolled. Every class has a quiz to be completed at the end of the lecture, and week by week you have a graded homework to be completed in groups of 2 or 3 people. There occasionally a final project, but rarely a final exam. The final grade of the class will be a weighted average of your quizzes and homework.
[left] STAT123 professor take a break and show us an early interview with Pink Floyd that was remarkably conducted by his British uncle, [right] CS109B professor take out the guitar and start singing the 'Bayes Song', a song written by him inspired by the Bayes Theorem
The collaboration aspect is on one hand enhanced by the fancy facilities, but is mainly enhanced by the attitude of the teaching staff that often become surprisingly creative (see example above) and efficient in their work: one example might be a late Wednesday night I was with a couple of my classmate from CS109B in the Smith Building (picture) above, right before the deadline of our (their) Reinforcement Learning homework. The homework was extremely well prepared (using OpenAI gym) and with an increasing difficulty level, working on it for a couple of days taught me definitely more than similar online courses I had taken before.
>>> Observation #4: The most valuable knowledge I got was no on the study methods of the (relatively advanced) topics I followed
I concentrated all my classes in Harvard around (Grad Level) Machine Learning, and I wasn't of course expecting to become an ML expert after one month of classes but I still learnt a surprisingly large amount of knowledge on the topic. More importantly, I saw how the different classes I took (CS181 Advanced Data Science, CS109 Harvard Machine Learning, 6.038 MIT Machine Learning) had different approaches to the subject and what kind of learning resources they were using. If you are interested in ML topics you might find the following resources delicious!
- Starting with Books, the main booked used were
- Bishop - Pattern Recognition and Machine Learning, really math heavy but still at undergrad level used in CS181
- Murphy - Machine Learning, math heavy undegrad level used in MIT 6.036
- Hastle, Tibshirani - The Elements of Statistical Learrning, extremely math heavy, graduate level, used in MIT
- James, Witten - An introduction to Statistical Learning, less math heavy, more on the practical side, used in CS109B
- Goodfellow, Benjo - Deep Learning, also pretty practical, used in CS109B
- Besides books, most of the courses material is online, here are the links to the courses I was following
- CS181 Machine Learning - website, not much on the website but you can find everything on the github repo
- CS109B Advanced Data Science, the website is pretty rich, they even had calendar with lecture topics and homeworks
- 6.S897 Machine Learning for HealthCare, MIT courses were harder to find online, infact the main course I was following 6.036 didn't even have a public web-page and was everything on Piazza. At the Same time I was following 6.S897 that was extremely interesting, worth having a look at the material online.
Actionable Item: this was a short overview of my Harvard holiday, I didn't write dozen of things I still have in mind so if you have any comments or question I highly encourage you to reach out to me. Also if you are even remotely considering to do something similar I might be able to give you some useful practical tips, I did the same thing in Harvard, MIT, Berkely and China, so I became pretty familiar with becoming a serial listener! You can drop me an email at firstname.lastname@example.org or DM me on Twitter.