Wednesday, April 15, 2015

Veritas Forum Lessons, Part 1: Stop Optimizing Everything

As many of you know (from my Facebook event invitation or previous blog post), I coordinated the Veritas Forum at MIT for the second year in a row Monday night. Since everyone asks, it was recorded and will be on veritas.org/mit some time in the next 3-6 weeks. I'll be reflecting on the event over two or more blog posts, and this is the first.

What was it about? Well, you can read the super long title on the Facebook page to see how we advertised it. To get feedback on the forum, we ask all of the participants what their biggest takeaway from the forum was. Here are some random examples to give you a flavor of the discussion.
the motives of actions are important. In other words, it's not just actions, it's the truth behind the actions that matter. True belief matters, and belief is valuable because provable absolutes don't exist. Even "cogito ergosum" is doubtable
There are very smart people on both sides of this issue
Religion shapes how people act and is important in that way
Respect more important than convincing
MIT professors are not philosophers (mostly)
The limits of science and proving God's existence should be considered.
The presence of God cannot be proven nor disproven, though this fact should not deter belief.
I need to do my taxes in the next two days!
Okay, so that last one was fake, but it did come up in Professor Formaggio's presentation. If you weren't there, I guess you'll have to wait for the video to see how he tied that in to his agnostic beliefs, because you'll never guess.

Anyways, as the organizer, I was already familiar with the content that they presented, and my biggest takeaway was not generated from the discussion itself. Instead, it was through this event that I learned the value in letting go of my perfectionism.

Leading the Veritas Forum last year, I had many ideas for how to make the forum better. I had left frustrated the year before (2013) with so many of the tiny mistakes that the leadership had made, and resolved to solve them all myself. For instance:
  • We had communication issues with the Veritas national organization representative, because she wasn't on all of our e-mail chains. So I created dedicated MIT mailing lists for each of the teams, and subscribed her to all of them. Later, this would prove useful for finding e-mails related to the Veritas forum in my inbox.
  • At least one of my friends was turned away at the door and wasn't told where the overflow rooms were, so he just left. So I created and gathered a dedicated usher and signage team the next year that put up a carefully choreographed sequence of signs... that didn't get used because we didn't need the overflow rooms.
  • The MC memorably mispronounced Feynman's name, so I planned to just be the MC myself.
  • The forum felt like all the other science and faith discussions that get to the same small set of topics and end before discussing much of substance that the speakers disagreed on. So I decided to try a different tack the next year: Discuss what important decisions people make in their lives, focusing on questions of career, success and purpose.
As you might expect, the eventual result was that I basically spent an entire month solely on the Veritas forum. I added it all up at the end and realized that I had done the work of about six different volunteers. This was directly after passing quals, so it wasn't such a big deal for my academic work. But I knew it wouldn't be reasonable in future years.

So this year, I resolved to delegate. I would take on the coordinator's role (a double role, since many years have co-coordinators), but wouldn't let myself do anything else besides the occasional task here and there. And I found that people did step up to serve. There are too many people to thank in this blog post, but I'm particularly grateful for Maria Cassidy, who volunteered to take over both the Outreach leader and MC responsibilities.

The thing about delegating, I found, is that you have to be willing to accept what other people put together. I don't mean to say that Maria did a bad job in any way; she was definitely a better MC than me, for instance, adding some humor to her introductions of the professors. What I mean to say is that I had to give up my desire to do things in exactly the best way, from my perspective.

---

Let me spell out exactly what my problem with perfectionism is. In short, any time something isn't optimized, it bugs me.

For instance, I tend to drink a lot of water while I work, which often leads me to need to simultaneously refill my water bottles and drain my bladder. Due to plumbing considerations, most buildings put these resources directly adjacent. But in my current office building, the water fountains are attached to the back of the bathrooms, which makes my routing much less efficient:

And this frustrates me. You'd be surprised how often this just nags me to death.

Perhaps it's because I'm a mathematician. When I'm engaged in deep thought, it can be annoying to have to remember to go to the bathroom after filling up my water bottles when it isn't just next door, or remember that I had left my water bottles at that three-way intersection of the paths when I went to visit the bathroom. (There also isn't anywhere to leave them at eye level except a trash can near the bathroom which isn't perfectly flat on top...) And then when I sit down to think and realize I still need to go to the bathroom or I'm missing my water bottles, I didn't save any time at all by trying to do both things at once.

This is actually the feature I most dislike about this swing space we're in. As you can probably tell from the floor plan, there isn't any natural light in the grad student spaces, but I don't notice the difference. And it's apparently not as nice as our new offices (which will be completed in January), but I already have more space and cabinets than I need.

---

Of course, there's something a little odd about that behavior. In some sense, the problem here is just hard to optimize. It isn't clear which of my options are best at any given moment, but the differences are so small that it shouldn't matter much to me. Whatever I decide to do, I'm not missing out in any big way.

One of the topics that came up in Professor Van Voorhis's Veritas presentation was that as a fundamental researcher, he believes that finding the truth is important in and of itself. He used this to justify why we should be interested in questions of the existence of God, just for their intrinsic value, because we want to know.

I think my brain operates somewhat similarly with respect to optimization. I seem to think that there's something intrinsically valuable in knowing that I've done something to the best of my ability. Every time I do, I give myself a little pat on the back.

(By the way, the new optimum to lunch-on-the-go is now Soylent, even if my path takes me by a Chipotle. It reduces cost to $2.50 and prep/cleanup time to 3 minutes (amortized over four meals), so the tradeoff is 2 minutes to ~$5, or $150/hr. Lower order terms still slightly favor Chipotle, as Soylent wins more on health and flexibility but loses on taste.)

(Okay, maybe I'm not really learning my lesson.)

---

Let me pick the story back up. How did the Veritas Forum teach me to stop optimizing? Well, the week before the forum, I felt like it was falling apart in all directions. Posters still hadn't gone up, and when they went up, they were difficult to find. See if you can spot Waldo the Veritas poster in both of the pictures below:



Even if you know that you're looking for (an astronaut/Jesus ascending mashup), it can be tough. And yes, I have horrible photography skills, but maybe they mimic what you'd see at a glance walking by.

To keep my complaining short, since that isn't really the point, people weren't signing up to help, and my leaders kept going missing. Then in the week leading up to the forum, two of them got sick.

One of those days, I was walking, thinking to myself: "How did it come to this? What went wrong with my plans? Is this forum just going to fall on its face, an embarrassment to all of us planning for not even the first time?"

And that's when I realized I'd asked those same sorts of questions before.

---

This is my second year doing a lot of things at MIT. It's my third year at MIT, and, well, it didn't take long for me to take on leadership roles, from being a hall councilor in my dorm to captaining the Graduate Christian Fellowship intramural frisbee team. It was actually when I failed to turn in a deposit check by a mid-February deadline for that frisbee team that I realized: I'm doing almost all of the same things this year, but instead of building on my experience to improve, I'm doing a worse job across the board.

The starkest example is in the Graduate Christian Fellowship large groups. The story runs somewhat similar to that Veritas story.
  1. I joined the large group team during my first year, and saw some small issues that I wanted to fix (e.g. not enough communication with speakers and redundancy with topics covered in church).
  2. I set out a vision for the new year, titled "Burning Questions," and executed it decently well, collecting the best questions the fellowship had to offer and conducting a vote (approval voting, of course) before finding speakers for the top eleven topics as determined by the votes. People came and were mostly excited to hear about all of the controversial stuff they wanted to know about: hell, heresy, homosexuality, and more topics that didn't start with the letter h.
  3. This year, my second year in charge, I tried to democratize the task of deciding the topics we would cover, implicitly delegating it to the leaders in the fellowship who suggested ideas. In the end, some of these new ideas were successful and some of which weren't, but the end result was that across the board, there was lower attendance and enthusiasm about large groups in general.
When even at our most popular events were poorly attended, we realized something was wrong. GCF president Gerald Pho and I had multiple conversations about it, but I remember one in particular in Sebastian's pretty clearly.

In discussing how we had ended up there, I realized that a major motivation to my democratizing plan was due to a lack of a large group team to put on the events, and a recognition of my own limits to run it all myself. Instead of putting together the best large groups I could, I had been trying to put on large groups that were the best I could without committing a ton of time to them myself.

I had learned that keeping your own involvement constrained means that you can't optimize. With the large groups, it meant that I didn't have much latitude to pick and choose between ideas, because I wouldn't have the energy to come up with something better or maintain a constant theme throughout the large groups over the course of a semester or year.

In the end, we decided that this strategy was not sustainable for large groups, and set about the task of gathering a team of leaders to help me again. We held a town hall meeting asking for additional feedback, and shared the lessons we had learned from this experience.

---

Walking and thinking about how the Veritas forum had ended up in such disarray, I took comfort in remembering that I had explicitly constrained myself from taking over every little thing. I resisted the urge to just put the team on my back and spend my entire week postering, manning a proxy board, recruiting ushers, and writing questions for the moderator to ask the two speakers to make the forum happen, whether anyone else was going to help me or not.

Instead, I had to pick my battles. For marketing, I invested heavily in inviting people via Facebook, and sent e-mails encouraging everyone involved to do the same. I knew the content of the forum would be very important, and decided to think hard about challenging questions to ask both speakers. (Perhaps they were too challenging, given the "MIT professors are not philosophers" comment... I did have higher hopes for many of their responses, to be honest.)

And I made some mistakes along the way. For instance, I didn't bother editing and reprinting those response cards to accurately tailor to the followup events we would have, like I had the previous two years, and now we're not really sure who it's appropriate to e-mail or not to follow up.

So let me go ahead and advertise what I do know here: If you're at MIT (particularly if you're a grad student), GCF is hosting a similar panel with 3-4 Christian graduate students explaining what their evidence for belief is. It'll be in the MIT Chapel at 8pm on Friday. You should come!

In the end, the forum happened, and unlike with GCF, it actually didn't hurt to not optimize it. Over 400 people came, filling the room to close to its capacity, and we were able to find enough ushers with a last-minute e-mail I sent a couple hours before the forum. Possibly apart from the follow-up, the mistakes we had made weren't such a big deal after all.

---

Experienced optimizers/perfectionists like me will notice that it's perfectly possible to still optimize over a constrained time budget. You just add in the constraint of finishing by a certain time, or add a penalty for taking longer to do it.

The most obvious example of this came on the night itself, when the Veritas representative Diane and I were tasked with sorting through the questions people had submitted by text, realtime. Over the course of about an hour, we received over 80 questions, and had to sort through them to find six that the moderator would eventually ask. In fact, we picked nine, but we didn't end up having time for these three:
For Troy: How do you think of the miracle stories in other religions or culture? 
For Troy: Why are there suffering and unfairness if God is just and gracious?
For Joe: What evolutionary benefit is there to being predisposed to believe in a higher power?
So my apologies if you were the one asking one of these questions. I guess this is like getting an Honorable Mention. Anyways, I personally liked the first of these a lot, and I'm a little bummed we didn't get to hear Troy's answer.

At the same time, I'm sure there were many more questions that would have been better questions to ask than the six we got to. And you know, I'm okay with that. We were quite constrained: Fishing for well-worded answers that possibly showed up multiple times in various forms to suggest popularity of the question, all while trying to listen to the speakers and moderator ourselves to see if they'd already addressed these questions. Once we had most of our nine favorites, we stopped looking as closely.

In fact, I'm quite used to working under a constrained time budget from all of my time spent on math competitions growing up. Having a fixed amount of time helped me focus and eventually perform at the highest levels.

Of course, real life is much more open-ended. There are deadlines for some things (like taxes!) and then there are the things like getting reimbursed that you really should do sooner rather than later, but there's really no hard deadline.

As I've written about before, it's been difficult for me to transition from the first world of deadlines to the second. (That post is a bit out-of-date; look for a future post about my new organizational method.) Essentially my current solution is to artificially make the second world into the first by setting deadlines for myself, but allow myself to postpone some tasks to future days if I run out of time.

---

So what has changed as a result, if it's still possible to optimize with limited time? Here are the practical lessons I've learned that I'm hoping will stick:

1) Prioritize which things to optimize. I have a tendency to want to perfect, or at least improve, absolutely everything I come across. Normally, this is a good thing, as it helps me to leave behind lasting changes, like the various rides spreadsheets I've scattered about Caltech and MIT.

But sometimes it just leads me to optimize things that don't really matter. I notice this strongest every week when I hand back homework to the geometry students I'm teaching online. We strive to give detailed, helpful comments on their writing problems, usually 3-4 paragraphs describing what they did right and wrong and how they can further improve.

The way the process works is that graders (often undergrads) take the first look at the submissions and give feedback, and then I check it over before getting it back to the students. I can almost always find additional comments to make on the students' solutions, whether it's a style suggestion or the grader didn't quite understand what the student was trying to do and where their mistake really lay.

But taking a step back, this is a waste of time. Sure, I get paid by the hour, but I already refuse tutoring gigs because it's not a good use of either my time or their money. And giving the best possible individualized feedback is not so different from tutoring, except that the students might not read it.

The way I'm shifting my strategy on this is twofold: First, I'm changing my approach to releasing grader comments: If they say something useful and give an approximately good score, I'm just going to release it. Second, I've already begun compiling the most common errors and style comments to share each week in class. This way I can make sure that the effect of my work is multiplied to the whole class, not just one of my students.

2) Recognize when returns on effort are increasing or diminishing. If they are increasing, try to optimize. If they are decreasing, doing a poor job is not so bad.

I think this one explains the difference I noticed between GCF and Veritas above. With GCF large groups, as people pointed out in the town hall meeting, the community aspect made them more likely to come if they thought other people they wanted to see would be coming. Holding high quality events was key for not just the impression that it would be worth attending but the assumption that others would think the same.

On the other hand, impersonal forms of marketing have diminishing returns. Maybe people are going to see your poster, maybe not, but if you put up twice as many posters, less than twice as many people are going to see them, since some people will see multiple posters. Some people will even react negatively to seeing too much advertising for something.

3) Celebrate a job done well, but move on and recognize that not all jobs need to be done well. There have been several moments in my life when I've been really excited with the way something has turned out, where it came to a sort of closure, perfectly wrapping up all the loose ends. One of these was the Burning Questions series in GCF, which was exciting and by the end, we managed to cover exactly the eleven topics that were most popular with the fellowship. People look back now on that series as a good example of what large groups can be.

Another example that comes to mind is the Simple Person's Applied Math Seminar, which I restarted (with help) my first year here and ran for a solid three semesters before passing it on. At the end of the last talk of the Fall 2013 semester, I held an awards ceremony for the best talks of the semester, giving out themed awards related to each of the categories in the title. As a humorous afternote, this unwittingly hooked one of my classmates (William Yu) on dried fruit from his prize of dried apples for the best applications in his talk (get it?).

Apart from the questions asked of the speakers, I haven't had the same sense of satisfaction from the two Veritas forums I've organized, but it's okay. As Diane brought up at one point with regards to the title, sometimes done is better than perfect.

I also don't have a similar affect with respect to my two years as a Hall Councilor; I've made numerous mistakes along the way, like forgetting to order rice for a Chinese dinner at the last event we had. You'd think that two years into the job, I would be better, but you know, it's okay. We had our fun while we could, and I got to know some of my neighbors pretty well. I had dreams that we'd replicate some aspects of undergrad, but not all dreams need to come to fruition, and I'm learning to be okay with that.

6 comments:

  1. Images #2 and #3 link to gmail, and therefore don't show up for me, and probably anyone not logged into your gmail.

    ReplyDelete
  2. > Let me spell out exactly what my problem with perfectionism is. In short, any time something isn't optimized, it bugs me.

    You might like Gregory W. Dawes' Theism and Explanation. He argues, among other things, that for a theistic explanation to be a valid explanation, it requires having some model of God's mind, such that predicting that God would take the most efficient route gives one increased understanding of reality. One's theistic explanation has to rule out many logical possibilities in order to be a valid explanation.

    ReplyDelete
  3. I have a solution to your water bottle and bathroom problem. Fill up n water bottles at once, so for every n trips to the bathroom, you only need to make the water bottle trip once. For large n (limited by the number of water bottles you can carry), the cost of the additional trip is amortized and becomes negligible.

    ReplyDelete
    Replies
    1. Yeah, I've thought about that, but it's much more comfortable to carry two bottles at once than more than that (especially when full). I think four would be my upper limit if I didn't put them in a backpack or something, and if I'm bothering to do that, the time savings are mostly lost.

      Delete