8. PhD Survival Guide: Top Lessons I Wish I Knew Earlier
If you are still working towards your PhD degree, this episode is for you!
In this week's episode, we dive deep into the challenges and opportunities facing early PhD students. From navigating idea generation to building key connections, Henrique and Rohan share their personal stories and advice for surviving and thriving as a beginner in academia. We discussed the messy behind-the-scenes struggles most students face but rarely discuss out loud. We talked about the importance of invest in yourself and be patient. How to deal with peer pressure and sometime self-pressure.
I wished I could hear discussions like this when I started my PhD journey and I hope you find it helpful too.
- 0:00 - Intro
- 1:13 - Rohan: The challenge of early PhD years - Being patient with yourself
- 6:25 - Henrique: How to deal peer pressure and/or self-imposed pressure
- 21:30 - When to build vs when to buy
- 27:40 - Celebrating small wins and taking all the help you can early on
- 34:10 - Ding's two stories on asking questions and reaching out to people
- 40:01 - How to keep oneself motivated.
- 44:49 - Pair up with peers in your field
- 49:03 - The Ira Glass quote on creativity and taste
- 51:00 - Project timelines across subfields
Full transcript text
[00:00:00.000 --> 00:00:03.360] If you are still working towards your PhD degree, this episode is for you.
[00:00:03.360 --> 00:00:08.240] This week, we dive deep into the challenges and opportunities facing early PhD students.
[00:00:08.240 --> 00:00:11.600] Henrique and Rohan share their personal stories and advice
[00:00:11.600 --> 00:00:14.320] for surviving and thriving as a beginner in academia.
[00:00:14.320 --> 00:00:18.960] We discuss the messy behind-the-scenes struggles most students face but rarely discuss.
[00:00:18.960 --> 00:00:23.120] We talk about the importance of investing in yourself and be patient.
[00:00:23.120 --> 00:00:27.120] We also touch on how to deal with peer pressure and sometimes self-imposed pressure.
[00:00:27.120 --> 00:00:31.200] I wish I could hear this kind of discussion when I started my PhD journey,
[00:00:31.200 --> 00:00:33.360] and I hope you find this episode helpful too.
[00:00:33.360 --> 00:00:41.680] Hello everyone, in this episode we have Rohan and Henrique to discuss some of our thinking around
[00:00:41.680 --> 00:00:44.400] challenges and opportunities for early PhD students.
[00:00:44.400 --> 00:00:48.640] Some of us are more recent graduate like Rohan and Henrique.
[00:00:48.640 --> 00:00:52.240] I think they just wrapped up their PhD less than a year ago,
[00:00:52.240 --> 00:00:55.120] and I got my PhD around five years ago.
[00:00:55.120 --> 00:00:59.440] So we can share about some of our thoughts on challenges and opportunities.
[00:00:59.440 --> 00:01:06.160] Rohan, we came to this topic because you brought it up in our earlier private conversation.
[00:01:06.160 --> 00:01:09.520] Do you want to just talk about how you started thinking about this problem?
[00:01:09.520 --> 00:01:16.240] Yeah, just sort of something I've been thinking about as I've been on the latter stages of the
[00:01:16.240 --> 00:01:18.800] PhD, getting out of the door, essentially.
[00:01:21.840 --> 00:01:25.440] It's interesting how you sort of come a long way, you know, over six years,
[00:01:25.440 --> 00:01:31.760] where you sort of come in not knowing anything, you know, doing completely random things,
[00:01:31.760 --> 00:01:35.360] things don't panning out, things not really panning out.
[00:01:35.360 --> 00:01:41.360] And then it's an interesting time sort of earlier on in your PhD where you don't know anything.
[00:01:41.360 --> 00:01:47.600] And it takes a few years to really settle in and find a groove, essentially.
[00:01:48.560 --> 00:01:52.720] It's also interesting because that's normally the people who talk to you when you're in your
[00:01:52.720 --> 00:01:56.400] latter years of your PhD, like towards the end, it's people just starting out,
[00:01:56.400 --> 00:02:00.800] or people like looking up to you thinking that you have figured it out somehow.
[00:02:00.800 --> 00:02:04.800] And really, you're still kind of like, oh, I'm finally starting to hit that groove.
[00:02:04.800 --> 00:02:08.080] The thing that I had planned to do this whole time, like now it's starting to come together.
[00:02:08.080 --> 00:02:11.120] And so it's sort of a random walk.
[00:02:11.120 --> 00:02:12.960] I think we all have slightly different experiences.
[00:02:12.960 --> 00:02:16.720] And so getting there is an achievement on its own.
[00:02:16.720 --> 00:02:22.080] And then keeping in mind how far you've come is also something you have to keep in mind.
[00:02:22.080 --> 00:02:24.560] Yeah, still very far from figuring anything out.
[00:02:24.560 --> 00:02:29.360] So one like a quick baby title for this episode could be
[00:02:29.360 --> 00:02:32.880] things you tell yourself five years ago when you started your PhD.
[00:02:32.880 --> 00:02:34.880] Would have been super useful to know.
[00:02:34.880 --> 00:02:37.760] Okay. So Rohan, what's the first thing you want to tell your past self?
[00:02:37.760 --> 00:02:43.520] That was a weird noise.
[00:02:43.520 --> 00:02:49.200] Well, what I would tell myself is, I think I still tell myself this is sort of, you know,
[00:02:49.200 --> 00:02:55.840] you have to really be patient with yourself because, you know, skills don't develop in a day.
[00:02:55.840 --> 00:02:59.120] They don't develop even in three months or six months.
[00:02:59.120 --> 00:03:01.360] It sometimes takes a year, takes two years.
[00:03:01.360 --> 00:03:07.200] And especially when you're doing a PhD in a technical topic, right?
[00:03:07.200 --> 00:03:12.400] You really sort of have to become a master on a topic.
[00:03:12.400 --> 00:03:16.480] Maybe not even a master, but you have to develop expertise and that takes time.
[00:03:16.480 --> 00:03:22.560] You're reading lots of papers, you're reading textbooks, and this is dense material.
[00:03:22.560 --> 00:03:26.320] It's not easy to absorb in one day and, you know, next day,
[00:03:26.320 --> 00:03:28.320] code it out and have a paper the next day.
[00:03:28.320 --> 00:03:34.400] So, you know, just sort of being patient with yourself in the first few years
[00:03:34.400 --> 00:03:37.760] and investing in yourself and that can happen in multiple ways.
[00:03:37.760 --> 00:03:42.960] And we can talk about that is, I think, sort of the primary thing,
[00:03:42.960 --> 00:03:45.280] I would say, is sort of the most important.
[00:03:45.280 --> 00:03:47.840] I agree 100%.
[00:03:47.840 --> 00:03:51.440] It's like, it's also, you have to keep in mind you're coming from,
[00:03:51.440 --> 00:03:56.560] you just either graduated undergrad or did some sort of masters and you're coming out of like
[00:03:56.560 --> 00:04:01.280] being at the top of your pool, wherever you were, and you worked so hard to get here to sort of
[00:04:01.280 --> 00:04:04.720] impress or you're coming in from industry or wherever.
[00:04:04.720 --> 00:04:09.040] And so you're in this completely other mindset and then you join this pool of people who are at
[00:04:09.040 --> 00:04:10.480] different stages in their PhD.
[00:04:10.480 --> 00:04:13.600] They're a couple years in, they're six years in, they're wherever,
[00:04:13.600 --> 00:04:15.600] and everyone's doing such impressive stuff.
[00:04:15.600 --> 00:04:18.160] And the reason you're there is because you want to be doing impressive stuff.
[00:04:18.160 --> 00:04:20.400] Like you work really hard to get there also.
[00:04:20.400 --> 00:04:24.240] But then it's coming to realize all those things you don't know,
[00:04:24.240 --> 00:04:28.560] even if it's like as simple as like how to like format a paper,
[00:04:28.560 --> 00:04:30.960] what's the procedure, like what are the deadlines,
[00:04:30.960 --> 00:04:34.080] like how you have to start code from scratch, how you have to do all these things.
[00:04:34.880 --> 00:04:38.720] Whereas before you normally had like assignments that you were really good at completing.
[00:04:38.720 --> 00:04:42.400] Now you're sort of like taking care of yourself the whole time.
[00:04:42.400 --> 00:04:46.320] And then you're getting advice from an advisor and you're getting advice from your peers,
[00:04:46.320 --> 00:04:50.160] but you really, it's a lot more independence than you've had before.
[00:04:50.160 --> 00:04:56.560] While surrounded by really talented people and reading really dense works and papers and
[00:04:56.560 --> 00:04:57.040] projects.
[00:04:57.040 --> 00:05:00.400] And so trying to sort of stay afloat while that's happening.
[00:05:01.040 --> 00:05:04.560] Yeah, so to sort of piggyback on this and maybe give more context,
[00:05:04.560 --> 00:05:08.720] at least this is how I felt during the early years of my PhD, which was,
[00:05:08.720 --> 00:05:14.560] okay, one aspect is sort of the ability to execute, to write a paper,
[00:05:14.560 --> 00:05:18.720] to actually do some research, to come up with new ideas.
[00:05:18.720 --> 00:05:22.480] But there's the whole other aspect of, well, what do I work on?
[00:05:22.480 --> 00:05:26.880] Sometimes your advisor can tell you what to work on.
[00:05:26.880 --> 00:05:29.680] Sometimes that's not always the case.
[00:05:29.680 --> 00:05:34.080] In fact, sort of after my first project, my advisor sort of actively said,
[00:05:34.080 --> 00:05:39.040] well, now you're on a random walk and go figure out what you want to work on.
[00:05:39.040 --> 00:05:45.280] And so then it took a year or two years to sort of land on something.
[00:05:45.280 --> 00:05:48.880] And even then it wasn't sort of fully fleshed out.
[00:05:48.880 --> 00:05:54.800] So it's sort of different from sort of before your PhD, you might sort of develop skills
[00:05:54.800 --> 00:05:58.560] in undergrad or master's or your work where you become good at executing.
[00:05:59.120 --> 00:06:03.760] But now it's sort of a different game in terms of, well, what do I work on?
[00:06:03.760 --> 00:06:10.240] And how do I make sure that the thing that I'm working on, you know, is not everyone else is
[00:06:10.240 --> 00:06:12.640] working on it, right? Otherwise it would be sort of done.
[00:06:12.640 --> 00:06:16.720] So it's an interesting place to be and to go through that experience.
[00:06:16.720 --> 00:06:21.120] Yeah. I like you to use the term invest in yourself.
[00:06:21.120 --> 00:06:25.760] I think investments, as we all know, it's kind of irrational.
[00:06:25.760 --> 00:06:29.120] People don't follow financial investment advice.
[00:06:29.120 --> 00:06:34.320] They usually expect short-term return, but usually we know we should invest in
[00:06:34.320 --> 00:06:39.280] Dex Fund and then wait for long, long, long decades before we see the return.
[00:06:39.280 --> 00:06:41.600] And same mentality applies here.
[00:06:41.600 --> 00:06:50.560] If we, as early PhD students, chose this invest in yourself mindset in one year, two years,
[00:06:50.560 --> 00:06:55.680] maybe three years, we won't see big returns because we're spending all this year, like you said,
[00:06:55.680 --> 00:06:58.000] random walk or just reading textbooks.
[00:06:58.000 --> 00:07:05.440] I see that's a challenge because you will see your peers, they are cranking up either papers
[00:07:05.440 --> 00:07:11.840] that are not well thought out, maybe papers that do not result from random walk.
[00:07:11.840 --> 00:07:15.360] Instead, their advice might be giving them tons of great ideas to work on.
[00:07:15.360 --> 00:07:18.880] And they crank up papers while you are investing yourself.
[00:07:18.880 --> 00:07:23.280] And do you feel like there is kind of peer pressure or kind of, I don't know,
[00:07:23.280 --> 00:07:25.520] like pressure in general, self-pressure?
[00:07:25.520 --> 00:07:30.800] I think this is such a good analogy because you really are investing yourself and you have to
[00:07:30.800 --> 00:07:34.320] think of it as you're investing yourself with no money in the bank.
[00:07:34.320 --> 00:07:35.760] Like you have not earned it yet.
[00:07:35.760 --> 00:07:39.360] You don't have the funds to throw around and take big risks.
[00:07:39.360 --> 00:07:44.560] And so a lot of the times you want to start these really cool, really impactful, large projects.
[00:07:44.560 --> 00:07:50.560] But people like your advisors and the people around you don't necessarily know your background.
[00:07:50.560 --> 00:07:51.920] They don't know what you were able to do.
[00:07:52.880 --> 00:07:55.040] And you want to take all these big risky gambles.
[00:07:55.040 --> 00:07:56.800] And a lot of it just takes time.
[00:07:56.800 --> 00:08:00.640] And so you really, what you have to do is plan for the long run.
[00:08:00.640 --> 00:08:06.240] And what I would say is just like an investment to sort of start small and let your investments
[00:08:06.240 --> 00:08:07.600] snowball over time.
[00:08:07.600 --> 00:08:12.560] Like you want to be able to pick a project that's really small that you know will succeed,
[00:08:12.560 --> 00:08:13.840] that will work.
[00:08:13.840 --> 00:08:18.320] And that will give you the sort of confidence and the tools to then take on a bigger bite
[00:08:18.320 --> 00:08:22.800] or make riskier gambles and assessments and investments or choose projects that are
[00:08:22.800 --> 00:08:25.280] slightly harder but will have more payoff.
[00:08:25.280 --> 00:08:26.720] But in the beginning, you don't want to do that.
[00:08:26.720 --> 00:08:28.160] You want to sort of hit the ground running.
[00:08:28.160 --> 00:08:29.280] You want to do something amazing.
[00:08:29.280 --> 00:08:31.120] You want to make the really big risky call.
[00:08:31.120 --> 00:08:37.280] But I think looking back five years ago, that's what I would have told myself is sort of start
[00:08:37.280 --> 00:08:38.400] with a really small project.
[00:08:38.400 --> 00:08:43.440] Let that build your confidence and sort of work forward from that because other ideas
[00:08:43.440 --> 00:08:44.080] will follow.
[00:08:44.080 --> 00:08:45.040] People will read your paper.
[00:08:45.040 --> 00:08:46.320] They'll give you advice.
[00:08:46.320 --> 00:08:50.800] And all that stuff is much better than getting stuck and trying to overcome some really big
[00:08:50.800 --> 00:08:53.120] problem.
[00:08:53.120 --> 00:08:58.720] Yeah, I mean, a couple of things if I sort of look back, right, I think something that
[00:08:58.720 --> 00:09:04.560] sort of helps early on in the PhD especially is if your advisor can put you on a project
[00:09:04.560 --> 00:09:10.560] where, you know, ideas partially fleshed out or it's being guided by a senior student,
[00:09:10.560 --> 00:09:18.320] you know, it's sort of like so first of all, yes, you know, it's natural of us to both
[00:09:18.320 --> 00:09:23.920] be pressured by what other people are doing and, you know, impose this weird self pressure
[00:09:23.920 --> 00:09:24.320] on us.
[00:09:24.320 --> 00:09:31.280] And it's, I think it sort of makes the landing easier if sort of your first project is a
[00:09:31.280 --> 00:09:36.480] little bit more guided if you have that opportunity, you know, sort of actively asking your advisor
[00:09:36.480 --> 00:09:40.560] what even specifically saying, can I be put on a project where there are already some
[00:09:40.560 --> 00:09:41.280] results, right?
[00:09:41.280 --> 00:09:43.120] So that I sort of learn the ropes.
[00:09:43.120 --> 00:09:44.640] I don't have to be the first author.
[00:09:44.640 --> 00:09:46.640] I can be the third author, fourth author.
[00:09:46.640 --> 00:09:47.680] It's all fine.
[00:09:47.680 --> 00:09:50.720] But it's sort of all about gathering experiences, right?
[00:09:50.720 --> 00:09:57.920] And the more experiences you gather, the more sort of your brain churns and the more, you
[00:09:57.920 --> 00:10:01.840] know, ideas you can produce that are not vague, right?
[00:10:01.840 --> 00:10:05.360] That's not becoming more concrete and you get a sense of the field.
[00:10:05.360 --> 00:10:09.200] So there's a lot of pressure.
[00:10:09.200 --> 00:10:09.600] I agree.
[00:10:09.600 --> 00:10:11.680] There's a lot of pressure that you put on yourself.
[00:10:11.680 --> 00:10:12.880] At least I used to put on myself.
[00:10:13.680 --> 00:10:21.280] But at the same time, you have to, you know, don't like earlier on, I feel like you don't
[00:10:21.280 --> 00:10:24.560] have to be the lead in a project, for instance.
[00:10:24.560 --> 00:10:28.960] And then you have to, in the meantime, in those early years, sort of invest in yourself.
[00:10:28.960 --> 00:10:33.040] So things that people have already been saying, you know, read textbooks.
[00:10:33.040 --> 00:10:36.560] I think reading textbooks is awesome, especially early on.
[00:10:36.560 --> 00:10:38.560] You know, read papers, people read papers.
[00:10:38.560 --> 00:10:44.880] But like, you know, reading a textbook from start to finish, you sort of gain an appreciation
[00:10:44.880 --> 00:10:48.320] of a topic in the whole, which is just awesome.
[00:10:48.320 --> 00:10:50.160] So you sort of understand the context.
[00:10:50.160 --> 00:10:55.200] And for me, at least, it's like understanding a context of, okay, let's say if I'm doing
[00:10:55.200 --> 00:10:58.640] machine learning, why are things done in a certain way?
[00:10:58.640 --> 00:11:03.520] If I'm doing generated modeling, why are GANs the way to go?
[00:11:03.520 --> 00:11:05.520] Or why are diffusion models the way to go?
[00:11:05.520 --> 00:11:08.000] Or why are normalization flows the way to go?
[00:11:08.000 --> 00:11:10.080] There's so many different options, right?
[00:11:10.080 --> 00:11:13.840] But these are all sort of different sides of the same coin with different trade-offs.
[00:11:13.840 --> 00:11:18.800] And, you know, okay, so a textbook won't give you the latest and greatest on these topics,
[00:11:18.800 --> 00:11:24.800] but it sort of sets the footing and the foundations to sort of tackle the reason people are coming
[00:11:24.800 --> 00:11:26.240] up with these advanced topics.
[00:11:26.240 --> 00:11:29.680] So I don't know, reading a textbook, I think is a great investment early on.
[00:11:29.680 --> 00:11:33.360] I think there's so many great investments you can make.
[00:11:33.360 --> 00:11:37.600] And I think we'll all sort of agree that ideation is sort of the hardest part of a
[00:11:37.600 --> 00:11:38.080] PhD.
[00:11:38.080 --> 00:11:43.200] Like, we all can churn out code, we all can read papers, we all can like attend and make
[00:11:43.200 --> 00:11:45.920] slides and attend conferences and do all this stuff.
[00:11:45.920 --> 00:11:48.080] Some of these things we enjoy more than others.
[00:11:48.080 --> 00:11:51.440] But like coming up with an idea from scratch is just like so hard.
[00:11:51.440 --> 00:11:55.040] And you have to really be paged into what's happening in the state of the art.
[00:11:55.040 --> 00:11:59.680] All have the foundations of like having essentially read the textbook and understanding the language
[00:11:59.680 --> 00:12:02.160] that everyone's using to talk about these things.
[00:12:02.160 --> 00:12:05.200] And so doing all those things simultaneously is really hard.
[00:12:05.200 --> 00:12:09.360] And then you have to sort of invent something new, like be creative rather than just executing
[00:12:09.360 --> 00:12:10.400] on top of that.
[00:12:10.400 --> 00:12:16.000] And so if your first project, you can piggyback off of someone by being the third author or
[00:12:16.000 --> 00:12:20.880] your advisor gives you a project or does something like that, then you're saving yourself from
[00:12:20.880 --> 00:12:22.720] having to do this hardest part.
[00:12:22.720 --> 00:12:24.480] You're learning all the other skills.
[00:12:24.480 --> 00:12:29.920] And then you're managing to sort of like make your name, start making collaborations and
[00:12:29.920 --> 00:12:33.600] understanding how the process works, while considering what ideas you actually want to
[00:12:33.600 --> 00:12:34.000] work on.
[00:12:34.640 --> 00:12:38.640] The other nice part of like working on someone else's project is you don't have to necessarily
[00:12:38.640 --> 00:12:42.080] pick something from scratch and then have it be your whole career.
[00:12:42.080 --> 00:12:46.800] Like I found myself like if you worked on one project once, suddenly you're the guy who
[00:12:46.800 --> 00:12:47.360] does that.
[00:12:47.360 --> 00:12:51.520] And especially if it's something new or something you proposed, there's a sort of expectation
[00:12:51.520 --> 00:12:55.120] that you keep pushing that direction and that people will keep coming to you to continue
[00:12:55.120 --> 00:12:55.680] that topic.
[00:12:55.680 --> 00:13:01.440] And so you have to sort of be careful in a way with what topics you start off with, unless
[00:13:01.440 --> 00:13:03.040] you really know that that's what you want to do.
[00:13:03.600 --> 00:13:08.080] Otherwise, you'll just sort of dig yourself deeper and deeper into this general area,
[00:13:08.080 --> 00:13:11.840] which you may have randomly found yourself doing just because you thought it would be
[00:13:11.840 --> 00:13:13.440] whatever, easy or quick or something.
[00:13:13.440 --> 00:13:18.960] You both touched on the point of joining somebody else's project to begin with.
[00:13:18.960 --> 00:13:22.960] It's a great way to learn and just grow as a early PhD student.
[00:13:22.960 --> 00:13:28.800] You also mentioned at some point, the idea that you have will be more concrete.
[00:13:28.800 --> 00:13:31.120] And I think that's an organic process.
[00:13:31.120 --> 00:13:32.800] You will know when you get there.
[00:13:32.800 --> 00:13:38.720] Is there more measurable signal that you know you're getting there or you just have to know?
[00:13:38.720 --> 00:13:46.880] My personal experience is that in the lifetime of a project, there is sort of like a tipping
[00:13:46.880 --> 00:13:49.360] point before which it doesn't work.
[00:13:49.360 --> 00:13:54.160] After that tipping point, things work and you're confident it can carry through.
[00:13:54.160 --> 00:14:01.600] I haven't found a science to making that tipping point happen sooner.
[00:14:01.600 --> 00:14:07.280] I wish I can find that magic or the science of making that happen more often.
[00:14:07.280 --> 00:14:09.760] I don't know.
[00:14:09.760 --> 00:14:11.360] That was often my experience.
[00:14:11.360 --> 00:14:15.760] But it's again, sort of one of those things.
[00:14:15.760 --> 00:14:20.160] I think there's a bit of randomness involved in any research project.
[00:14:20.160 --> 00:14:26.720] It's like you don't achieve anything for six months and then suddenly sort of on a particular
[00:14:26.720 --> 00:14:31.760] day, you or a co-author has some insight which sort of works in your favor.
[00:14:31.760 --> 00:14:42.800] So sort of nonlinear amounts of work can happen in short, very short periods of time.
[00:14:42.800 --> 00:14:48.320] It's then during the rest of the period, sort of the rest of the time where things aren't
[00:14:48.320 --> 00:14:52.560] happening where you have to sort of balance the pressures and the emotions and not put
[00:14:52.560 --> 00:14:58.560] too much of that on yourself while continuing to invest, I guess, as we were discussing earlier.
[00:14:58.560 --> 00:15:03.040] Yeah, I think my question is more along the line of for PhD student when they're in the
[00:15:03.040 --> 00:15:06.720] early stage, they don't know how to come up with good ideas.
[00:15:06.720 --> 00:15:11.600] As they become a third author of a paper and then maybe in the second project, they are
[00:15:11.600 --> 00:15:14.480] the second author and they get skills, right?
[00:15:14.480 --> 00:15:20.000] They get better at coming up with ideas, they're getting good at implementing them, writing
[00:15:20.000 --> 00:15:21.040] papers and so on.
[00:15:21.760 --> 00:15:27.840] How do they know they are independent enough to start their own project?
[00:15:27.840 --> 00:15:33.520] Because I've seen PhD students who even till they graduate don't really have first author
[00:15:33.520 --> 00:15:35.600] papers and somehow they can still graduate.
[00:15:35.600 --> 00:15:38.080] I think they show independence some other ways.
[00:15:38.080 --> 00:15:43.200] But I just hope that I just wonder if you two have any thoughts on what are the signals
[00:15:43.200 --> 00:15:46.400] that you can observe and, hey, I'm ready.
[00:15:46.400 --> 00:15:49.600] I should start my own project instead of just tagging along other people's projects.
[00:15:50.800 --> 00:15:53.120] I would agree that there's no exact science.
[00:15:53.120 --> 00:15:57.200] So anecdotally, I have my couple of different signals that I saw.
[00:15:57.200 --> 00:16:02.720] And so like one of them is you read these papers and you're constantly hearing what
[00:16:02.720 --> 00:16:06.000] other people are doing and you're always kind of excited and happy for them or just sort
[00:16:06.000 --> 00:16:08.640] of like a little bit jealous of like, oh, this project seems so cool.
[00:16:08.640 --> 00:16:09.600] Why didn't I think of that?
[00:16:09.600 --> 00:16:10.240] Right.
[00:16:10.240 --> 00:16:13.280] And so once you stop thinking, why didn't I think of that?
[00:16:13.280 --> 00:16:14.320] That's a good sign.
[00:16:14.320 --> 00:16:17.120] You're starting to move the needle towards that direction.
[00:16:17.120 --> 00:16:22.320] And then it's when papers come out and you're you can sort of understand and immediately
[00:16:22.320 --> 00:16:26.400] appreciate the impact without having to read through the whole paper or try and implement
[00:16:26.400 --> 00:16:27.200] yourself.
[00:16:27.200 --> 00:16:30.880] And you're like, OK, those two ideas I very solidly understand.
[00:16:30.880 --> 00:16:34.160] And now they combine them in this way and that unlocks these four doors.
[00:16:34.160 --> 00:16:34.560] Cool.
[00:16:34.560 --> 00:16:34.960] Got it.
[00:16:34.960 --> 00:16:36.720] Let's see how they did in the results.
[00:16:36.720 --> 00:16:40.560] That's when you start to understand, OK, now you have a good finger on the pulse of
[00:16:40.560 --> 00:16:41.200] what's happening.
[00:16:41.200 --> 00:16:45.760] And then in my case, a lot of it was just like brute force.
[00:16:45.760 --> 00:16:50.080] Like I would suggest an idea and I would do a little bit of research and say, I think
[00:16:50.080 --> 00:16:50.800] we can do this.
[00:16:50.800 --> 00:16:52.080] And they'd go, cool, great.
[00:16:52.080 --> 00:16:54.080] Someone did this five years ago.
[00:16:54.080 --> 00:16:54.880] I'd be like, all right.
[00:16:54.880 --> 00:16:56.320] And then I would suggest another idea.
[00:16:56.320 --> 00:16:58.640] And they'd be like, OK, someone did this two years ago.
[00:16:58.640 --> 00:17:02.640] And then you would keep trying and keep exploring until you suggest an idea that someone is
[00:17:02.640 --> 00:17:04.240] like actively working on right now.
[00:17:04.240 --> 00:17:07.840] The next idea you have likely will be something new.
[00:17:07.840 --> 00:17:12.320] But at that point, you can't just be sort of throwing out random ideas and trying to
[00:17:12.320 --> 00:17:13.200] get something new.
[00:17:13.200 --> 00:17:15.840] It's because you understand, OK, this is what's currently happening.
[00:17:15.840 --> 00:17:19.520] Now I know I'm in a good position to make something new here.
[00:17:19.520 --> 00:17:25.200] And that just takes a lot of pitching ideas, a lot of reading other people's papers, and
[00:17:25.200 --> 00:17:28.640] then finally starting to understand them, which is not the piece.
[00:17:28.640 --> 00:17:30.240] You're not just reading there the whole time.
[00:17:30.240 --> 00:17:33.840] So as you're exploring other ideas and working on someone else's project, you have to sort
[00:17:33.840 --> 00:17:36.480] of actively be searching for other stuff.
[00:17:36.480 --> 00:17:40.640] And the ideas will come in the shower, on the commute, somewhere.
[00:17:40.640 --> 00:17:42.800] Not necessarily when you're trying to force the idea to come out.
[00:17:43.360 --> 00:17:48.160] I think, yeah, sort of to Henrique's point, what I would say is early on, read a lot.
[00:17:48.160 --> 00:17:53.680] And so as you read a lot, one thing to keep in mind is no idea is too small.
[00:17:53.680 --> 00:17:58.000] So you're reading a paper, and I really liked Henrique's point about unlocking things.
[00:17:58.000 --> 00:18:02.480] So OK, you're reading something, and you're like, oh, what if I change just this?
[00:18:02.480 --> 00:18:04.560] Would that unlock a new capability?
[00:18:04.560 --> 00:18:09.760] So sort of always having a mindset of, oh, can this change unlock that?
[00:18:09.760 --> 00:18:11.440] Or that change unlock this?
[00:18:12.080 --> 00:18:16.000] I think it's sort of just an awesome perspective to digest content with.
[00:18:16.000 --> 00:18:19.920] Not just reading for the sake of reading, but what else can I do?
[00:18:19.920 --> 00:18:22.160] How can I change the question a little bit?
[00:18:22.160 --> 00:18:23.520] This kind of stuff early on.
[00:18:23.520 --> 00:18:29.920] And then sort of what naturally starts happening later on is if you have a few papers, one,
[00:18:29.920 --> 00:18:34.640] two, or three, then those already serve as sort of feeding grounds for new ideas.
[00:18:34.640 --> 00:18:37.040] So at that point, you're coming up with your own ideas.
[00:18:37.040 --> 00:18:41.120] But sort of early on, I think you have to work off others' ideas.
[00:18:41.120 --> 00:18:47.520] And over there, sort of keep that mentality of what can I unlock if I use this in a different
[00:18:47.520 --> 00:18:50.640] context or if I change this here, that kind of thing.
[00:18:50.640 --> 00:18:56.640] The unlocking thing is fun because when you get to your later, like middle to later part
[00:18:56.640 --> 00:18:59.040] of your PhD, you're now a reviewer also, right?
[00:18:59.040 --> 00:19:02.320] Like you're being asked to review because you've been on a paper.
[00:19:02.320 --> 00:19:06.400] And then that's the fun part of what can you unlock because you read someone's work,
[00:19:06.400 --> 00:19:09.600] and every time they make any statement, like why did you do that?
[00:19:09.600 --> 00:19:10.800] Why didn't you do this instead?
[00:19:11.040 --> 00:19:11.440] Why?
[00:19:11.440 --> 00:19:11.680] Why?
[00:19:11.680 --> 00:19:14.080] Like every single time you're sort of questioning and pushing back.
[00:19:14.080 --> 00:19:18.640] And then a lot of times it's just because of like facility or ease or whatever, or it
[00:19:18.640 --> 00:19:19.280] makes sense.
[00:19:19.280 --> 00:19:23.680] But as you're beginning to understand why they made these choices and what choices didn't
[00:19:23.680 --> 00:19:28.720] they make, now you're asking the right questions that if you get to a point where you ask why
[00:19:28.720 --> 00:19:33.520] did they do this and the thing they didn't do makes sense to you or potentially has some
[00:19:33.520 --> 00:19:37.920] like really cool ramifications that they didn't consider, now you have an idea.
[00:19:37.920 --> 00:19:40.160] Now you can sort of push and open in that direction.
[00:19:41.120 --> 00:19:43.200] I'd also say one other thing.
[00:19:43.200 --> 00:19:49.520] This doesn't apply to everyone, but specifically for sort of folks who might be coding a fair
[00:19:49.520 --> 00:19:55.280] bit in what they're doing, I think it's super awesome in addition to reading textbooks and
[00:19:55.280 --> 00:19:58.720] papers and whatnot to build libraries from scratch.
[00:19:58.720 --> 00:20:05.360] I think that what I found other people in my lab as well, they develop libraries that
[00:20:05.360 --> 00:20:07.040] they're using till date.
[00:20:07.040 --> 00:20:09.440] And in fact, open source a lot of code.
[00:20:10.240 --> 00:20:11.360] People use this stuff.
[00:20:11.360 --> 00:20:20.480] So yeah, I know people who have won best software awards for libraries they wrote four or five
[00:20:20.480 --> 00:20:23.440] years ago, which is just amazing.
[00:20:23.440 --> 00:20:28.720] And it's like the libraries that you're writing aren't necessarily new research.
[00:20:28.720 --> 00:20:34.800] But I think there is, for instance, if you do rendering, I think there is so much to
[00:20:34.800 --> 00:20:39.440] gain from writing your own, let's say, ray tracer from scratch.
[00:20:39.440 --> 00:20:43.760] Develop architecting it on your own.
[00:20:43.760 --> 00:20:47.120] Nowadays, differentiable ray tracing is a hot topic.
[00:20:47.120 --> 00:20:49.360] So write a differentiable ray tracer from scratch.
[00:20:49.360 --> 00:20:58.560] It's good to use other people's framework, but to sort of spend six weeks or three months
[00:20:58.560 --> 00:21:03.920] on just sort of hardcore engineering, I derive a lot of sort of weird pleasure from that in
[00:21:03.920 --> 00:21:07.280] that you sort of get a sense of the pinpoints.
[00:21:07.280 --> 00:21:15.760] So that's sort of another way I think of unlocking ideas where implementing algorithms that are
[00:21:15.760 --> 00:21:24.000] not your own or implementing standard technology, it's totally worth it to sort of get a sense
[00:21:24.000 --> 00:21:25.840] of where does this suck?
[00:21:25.840 --> 00:21:30.080] Because that's when you can be like, OK, I can maybe improve that.
[00:21:32.320 --> 00:21:38.960] I think this is a topic because I agree with the part that I think we need to get our hands
[00:21:38.960 --> 00:21:39.360] dirty.
[00:21:39.360 --> 00:21:43.680] We need to try code, build our own stuff so that we become better engineers in general.
[00:21:43.680 --> 00:21:46.240] And that in turn might give us better ideas.
[00:21:46.240 --> 00:21:53.040] But I think I'm worried that we're also getting to the topic of when to buy versus when to
[00:21:53.040 --> 00:21:53.520] build.
[00:21:53.520 --> 00:21:55.520] That's also a common topic in companies.
[00:21:55.520 --> 00:21:58.400] Do we want to build GitHub from scratch?
[00:21:58.400 --> 00:21:59.280] We don't like GitHub.
[00:21:59.280 --> 00:22:00.800] We want to know all the things about Git.
[00:22:02.000 --> 00:22:04.560] But of course, our company, our research is not about Git.
[00:22:04.560 --> 00:22:05.360] We should just use it.
[00:22:05.360 --> 00:22:07.840] And this is an example, of course.
[00:22:07.840 --> 00:22:10.800] I think ray tracing is a more related one.
[00:22:10.800 --> 00:22:14.640] Should we build a ray tracer or should we lease, like buy a ray tracer?
[00:22:14.640 --> 00:22:21.120] It's interesting because if you're in the point where you're still sort of reading textbooks,
[00:22:21.120 --> 00:22:22.960] you learn a lot from coding yourself.
[00:22:22.960 --> 00:22:26.080] And then it's a really big investment long term.
[00:22:26.080 --> 00:22:30.160] And so using other people's code, it's essentially an API.
[00:22:30.160 --> 00:22:34.240] And if you're doing a systems paper, you don't really care about how it works under the hood.
[00:22:34.240 --> 00:22:37.760] You just need the pieces to connect so that you arrive at something new.
[00:22:37.760 --> 00:22:41.280] But then you don't understand the pain points, like Rohan was saying.
[00:22:41.280 --> 00:22:45.920] And so you end up missing out on, I could do this a little bit faster, or I could squeeze
[00:22:45.920 --> 00:22:49.680] a little bit more optimization out of this, which again is not the point necessarily of
[00:22:49.680 --> 00:22:50.080] the paper.
[00:22:50.080 --> 00:22:55.440] But that's where the ideas start rolling in, where next time you go to do something,
[00:22:55.440 --> 00:22:58.560] you don't not do it because the API doesn't allow you to.
[00:22:58.560 --> 00:23:01.680] It's because you really understand what's happening under the hood in terms of that.
[00:23:01.680 --> 00:23:03.760] But I sort of agree.
[00:23:03.760 --> 00:23:07.440] There's like a when do you buy, when do you build question.
[00:23:07.440 --> 00:23:12.720] And so if it's something simple where you think you can like code it up in a week or
[00:23:12.720 --> 00:23:16.000] so, and obviously everything takes way longer than you think it will.
[00:23:16.000 --> 00:23:18.400] So you think it's a week and it turns out to be a month.
[00:23:18.400 --> 00:23:24.400] But it's like, like Keenan and Alec are two great examples of this, where they, a lot
[00:23:24.400 --> 00:23:29.760] of the math, a lot of the like half edge matrix multiplication, random things, like things
[00:23:29.760 --> 00:23:33.680] that they do a lot with meshes, they coded it up themselves rather than just using an
[00:23:33.680 --> 00:23:34.080] API.
[00:23:34.080 --> 00:23:38.240] And those guys now they like their brains work differently, where they manipulate these
[00:23:38.240 --> 00:23:42.080] things on like a different level because they understand exactly what's underneath the
[00:23:42.080 --> 00:23:43.280] hood and they don't worry about it.
[00:23:43.280 --> 00:23:44.480] They're not intimidated by it.
[00:23:44.480 --> 00:23:46.560] They just they know what they can do and they can't do.
[00:23:46.560 --> 00:23:50.640] Whereas a lot of other people are like, well, if this doesn't work, I'm going to have to
[00:23:50.640 --> 00:23:52.880] convert it to some other form and then convert it back.
[00:23:52.880 --> 00:23:56.560] And it's because you're sort of unsure of what's going on underneath the hood.
[00:23:56.560 --> 00:23:56.800] Yeah.
[00:23:56.800 --> 00:24:01.280] I mean, just to give another example, right, you can say, OK, you know, mesh data structures
[00:24:01.280 --> 00:24:02.960] are a topic that's so old.
[00:24:02.960 --> 00:24:04.640] It's a 30, 40 year old topic.
[00:24:04.640 --> 00:24:05.840] You know, this is this is done.
[00:24:05.840 --> 00:24:10.880] But, you know, there's so many ways to sort of encode a mesh.
[00:24:10.880 --> 00:24:11.200] Right.
[00:24:11.200 --> 00:24:19.200] And it's sort of, you know, learning how to manipulate a mesh for different data structures
[00:24:19.200 --> 00:24:21.680] is there's so many options.
[00:24:21.680 --> 00:24:22.160] Right.
[00:24:22.160 --> 00:24:27.280] And and then you sort of start realizing, well, there are, you know, from a data structure
[00:24:27.280 --> 00:24:30.480] point of view, actually interesting even topological questions, right?
[00:24:30.480 --> 00:24:35.120] The geometric questions about handling a mesh, but topological questions that could maybe
[00:24:35.120 --> 00:24:40.320] sort of, you know, reduce storage or what if I don't store vertex positions?
[00:24:40.320 --> 00:24:42.400] What if I only store edge lengths in my mesh?
[00:24:42.400 --> 00:24:43.280] Right.
[00:24:43.280 --> 00:24:47.440] What does that enable or what does that not allow me to do?
[00:24:48.560 --> 00:24:53.120] So these kind of things, it might be an old topic, but I think, you know, totally worth
[00:24:53.120 --> 00:24:56.480] coding up your own mesh data structure, for instance, at least once.
[00:24:56.480 --> 00:25:03.280] Makes you think about what you can do with that data structure and how you can manipulate
[00:25:03.280 --> 00:25:03.920] it.
[00:25:03.920 --> 00:25:08.480] I think also if you're an early on PC student, you know that there's lots of different skill
[00:25:08.480 --> 00:25:08.960] sets you need.
[00:25:08.960 --> 00:25:12.480] Like you need to be able to read and understand papers and to collaborate with people.
[00:25:12.480 --> 00:25:13.280] You need to code.
[00:25:13.280 --> 00:25:15.040] You need to do all these things.
[00:25:15.040 --> 00:25:21.280] And so I think bounce around, like if you want to build up some confidence, build up
[00:25:21.280 --> 00:25:25.200] your own half-edge data structure, whatever, but then just take some library when you're
[00:25:25.200 --> 00:25:27.520] going to go print it out or render it or whatever.
[00:25:27.520 --> 00:25:32.720] Like use a little bit of both so that you're constantly pushing the levels on like your
[00:25:32.720 --> 00:25:35.200] reading or ideation, your collaboration, whatever.
[00:25:35.200 --> 00:25:39.440] And you don't have to sort of go all in on one, finish it before you move on to the next.
[00:25:39.440 --> 00:25:43.520] Just be aware that you can always be doing a little bit more coding and building up that
[00:25:43.520 --> 00:25:45.280] confidence versus the math.
[00:25:45.280 --> 00:25:47.680] There's always more math to learn, right?
[00:25:47.680 --> 00:25:52.560] Like each individual thing, you sort of pick your battles so that you're not sort of falling
[00:25:52.560 --> 00:25:56.080] behind or losing confidence in your ability to do one or the other.
[00:25:56.080 --> 00:26:00.880] And coding is a very easy win because we all have to do that beforehand or can pick it
[00:26:00.880 --> 00:26:04.720] up and do it at different scales, whether we're just taking advantage of other people's
[00:26:04.720 --> 00:26:06.320] code or writing from scratch.
[00:26:06.320 --> 00:26:11.440] And so especially when you're lacking confidence and you're sort of intimidated by everyone
[00:26:11.440 --> 00:26:14.560] else around you, like that's an easy way to win a few points.
[00:26:14.560 --> 00:26:20.720] I think, yeah, that really is a good point, especially early on because, you know, it's
[00:26:20.720 --> 00:26:27.280] like frustration or pressure always comes from not feeling productive, right?
[00:26:27.280 --> 00:26:34.480] So if you have four things going on, you know, a site coding project where nobody's evaluating
[00:26:34.480 --> 00:26:37.040] on that, evaluating you on that, right?
[00:26:38.000 --> 00:26:43.440] Putting some slides together for a paper, your main research project and something else
[00:26:43.440 --> 00:26:44.720] you're thinking about.
[00:26:44.720 --> 00:26:49.680] Like if you're sort of going on four fronts and you're, you know, if you're stuck on one,
[00:26:49.680 --> 00:26:53.840] you sort of switch focus to the other, make a little bit of progress, you're stuck on
[00:26:53.840 --> 00:26:55.200] that, switch to the other one.
[00:26:55.200 --> 00:27:01.600] You know, if you're making progress on even one or two fronts out of, let's say, the four
[00:27:01.600 --> 00:27:07.680] fronts that you're working on, you feel productive and as a result, sort of less frustrated.
[00:27:08.560 --> 00:27:16.240] And so it's sort of all right, you know, to find ways to feel less frustrated is to feel
[00:27:16.240 --> 00:27:19.760] productive in small wins, if that makes sense.
[00:27:19.760 --> 00:27:21.120] Totally.
[00:27:21.120 --> 00:27:25.120] And even if that small win is just sort of coding up a mesh library, for instance.
[00:27:25.120 --> 00:27:30.240] Yeah, I think it's an industry thing, right?
[00:27:30.240 --> 00:27:34.560] Like you feel like sometimes you're a little bit stuck and you want to keep pushing things
[00:27:34.560 --> 00:27:38.720] forward, but then there's other little wins you can take where you can squash a couple
[00:27:38.720 --> 00:27:42.000] of smaller bugs and then you can at least feel like the day wasn't lost because you
[00:27:42.000 --> 00:27:43.760] were just staring trying to figure out what was going on.
[00:27:43.760 --> 00:27:46.260] Yeah.
[00:27:46.260 --> 00:27:49.680] Yeah, I think that's a great discussion.
[00:27:49.680 --> 00:27:53.760] And do you have any other tips or advice you want to share?
[00:27:53.760 --> 00:27:58.240] For new PhD students?
[00:27:58.240 --> 00:28:01.040] Yeah, for like beginning or early PhD students.
[00:28:01.040 --> 00:28:01.540] Yeah.
[00:28:03.040 --> 00:28:06.880] We can do another episode on like final year PhD students, what's your advice?
[00:28:06.880 --> 00:28:09.840] But let's focus on early PhD students in this episode.
[00:28:09.840 --> 00:28:16.240] There's so much pride in coming out and trying to impress your advisor because they sort of
[00:28:16.240 --> 00:28:21.120] took a gamble on you, they're investing in you and you chose them, hopefully from a couple
[00:28:21.120 --> 00:28:24.640] different people you were deciding between or they chose you from different people.
[00:28:24.640 --> 00:28:27.280] But there's all this pressure and all this pride.
[00:28:27.280 --> 00:28:31.120] And I think it's like Rohan saying, if you can be productive as you're going, you'll
[00:28:31.120 --> 00:28:34.720] feel better and they'll feel better, but you're at least making progress in some direction.
[00:28:34.720 --> 00:28:40.560] And so if someone offers you an idea, take it, like don't feel like you have to go and
[00:28:40.560 --> 00:28:42.160] do and invent your own idea.
[00:28:42.160 --> 00:28:46.000] If someone offers you some help coding or some slides as a template, you don't have
[00:28:46.000 --> 00:28:47.600] to rewrite it from scratch.
[00:28:47.600 --> 00:28:51.760] Like learn what you can from what they're giving you so that next time when you go to
[00:28:51.760 --> 00:28:56.000] do it from scratch, you sort of already have a head start and you know what you're doing.
[00:28:56.000 --> 00:29:00.480] And if you somehow manage to like get on the paper, second, third, even first author early
[00:29:00.480 --> 00:29:03.760] on, now there's all this pressure is taken off your shoulders.
[00:29:03.760 --> 00:29:05.600] You're like, okay, they already have one paper.
[00:29:05.600 --> 00:29:07.200] They know that they can go through the process.
[00:29:07.200 --> 00:29:09.120] You can get to all this other parts.
[00:29:09.120 --> 00:29:14.400] And so whatever the quickest path to that is in the beginning, I'd say take it because
[00:29:14.400 --> 00:29:19.520] that will sort of set you up to do the longer, harder road of like coming out with your own
[00:29:19.520 --> 00:29:22.320] projects and doing things that are much more impactful.
[00:29:22.320 --> 00:29:26.240] Because without a doubt, all of our papers sort of got more complex as we went on.
[00:29:26.240 --> 00:29:30.400] And it's just because you understand more and you're able to sort of juggle more things.
[00:29:30.400 --> 00:29:34.880] And so like look forward to that, but don't feel like you need to rush there immediately.
[00:29:34.880 --> 00:29:36.880] Like take the small steps and let it snowball.
[00:29:36.880 --> 00:29:38.880] Yeah.
[00:29:38.880 --> 00:29:43.840] If I was also to summarize, I'd say, you know, one, number one, learn to be patient with
[00:29:43.840 --> 00:29:52.000] yourself, which sort of carries on, you know, with you through life and find ways to be
[00:29:52.000 --> 00:29:56.640] productive outside of your research project and how it's going.
[00:29:56.640 --> 00:29:58.080] You know, it might be going super well.
[00:29:58.080 --> 00:30:03.360] And when it's not going super well, then learn fine ways to be productive, reading textbooks,
[00:30:03.360 --> 00:30:07.440] coding, coding, coding, you know, those kinds of things to keep yourself busy and not get
[00:30:07.440 --> 00:30:08.080] frustrated.
[00:30:08.080 --> 00:30:12.240] I think that that's sort of a good sort of mentality to have.
[00:30:12.240 --> 00:30:13.920] Yeah.
[00:30:13.920 --> 00:30:18.560] And then, and Naiki, you said when somebody gave you an idea, for example, your advisor
[00:30:18.560 --> 00:30:22.800] or somebody else, just take it without like, don't feel like you have to invent your idea.
[00:30:22.800 --> 00:30:27.440] What do you think it's blocking some people from picking those ideas?
[00:30:28.400 --> 00:30:34.640] I think a lot of it is you feel guilty and sort of just taking an idea or you feel like
[00:30:34.640 --> 00:30:35.680] you don't understand it.
[00:30:35.680 --> 00:30:41.680] So I'd say I either rejected a lot of ideas early on because I thought they would be like
[00:30:41.680 --> 00:30:42.720] boring or no impact.
[00:30:42.720 --> 00:30:45.760] When there was, it was like a very clear path to a paper.
[00:30:45.760 --> 00:30:48.560] But maybe I was like, I'm not interested in this area.
[00:30:48.560 --> 00:30:52.880] I sort of brushed aside everything else I would have learned along the way.
[00:30:52.880 --> 00:30:56.320] And also just the ability to add another paper is always great.
[00:30:56.320 --> 00:30:59.040] We would all love to just have one more paper, like no problem.
[00:30:59.040 --> 00:31:05.440] Or I was intimidated because I looked at it and I didn't understand the math and I didn't
[00:31:05.440 --> 00:31:06.480] even start to try it.
[00:31:06.480 --> 00:31:06.720] Right.
[00:31:06.720 --> 00:31:10.480] I was just like, OK, there's other ideas sort of on the horizon that I feel like are more
[00:31:10.480 --> 00:31:10.960] manageable.
[00:31:10.960 --> 00:31:12.400] So I'm going to push for those instead.
[00:31:12.400 --> 00:31:17.600] Or it was a pride thing where I felt like I needed to be the one coming up with the idea.
[00:31:17.600 --> 00:31:21.840] All those are terrible excuses for not taking a paper.
[00:31:21.840 --> 00:31:25.600] Like if there's actually stuff you don't understand, maybe there's another piece these
[00:31:25.600 --> 00:31:29.040] students can help you or your advisor is suggesting they're willing to sort of put in the effort
[00:31:29.040 --> 00:31:29.920] and meet you halfway.
[00:31:29.920 --> 00:31:32.080] And so talk to them about that.
[00:31:32.080 --> 00:31:35.120] Tell them that you're insecure about maybe the math or this.
[00:31:35.120 --> 00:31:39.440] And they'll either tell you, OK, then maybe we'll pick another idea or like, don't worry
[00:31:39.440 --> 00:31:39.760] about it.
[00:31:39.760 --> 00:31:41.360] It's actually much simpler than it looks.
[00:31:41.360 --> 00:31:44.800] Or like read these three papers and come back and decide if you like it or not.
[00:31:44.800 --> 00:31:48.560] There's always ways to get more information and more help.
[00:31:48.560 --> 00:31:52.720] But I think a lot of the times because we're trying to impress our advisors or at least
[00:31:52.720 --> 00:31:57.840] I was in my case, you end up sort of not asking for help and sort of holding back.
[00:31:57.840 --> 00:32:02.160] And that's when you're like you're starting to work with one hand tied behind your back.
[00:32:02.160 --> 00:32:02.320] Right.
[00:32:02.320 --> 00:32:04.400] Like you really should just take all the help you can.
[00:32:04.400 --> 00:32:10.400] When someone else in another university or in an industry reads the paper, they don't
[00:32:10.400 --> 00:32:12.000] know all the help that you got.
[00:32:12.000 --> 00:32:16.880] For better or for worse, like they assume that you had a good equal part in it and they
[00:32:16.880 --> 00:32:18.720] give you the respect for doing it.
[00:32:18.720 --> 00:32:21.520] And so early on, take all the help you can.
[00:32:21.520 --> 00:32:24.720] And then afterwards, feel like, OK, maybe I need to contribute more and more.
[00:32:24.720 --> 00:32:27.280] And they'll equally respect you for it.
[00:32:27.280 --> 00:32:31.200] And they also understand when they did their papers, they got help from like places they
[00:32:31.200 --> 00:32:32.320] didn't imagine.
[00:32:32.320 --> 00:32:32.720] Right.
[00:32:32.720 --> 00:32:34.240] So it's just part of the process.
[00:32:34.240 --> 00:32:35.600] Yeah.
[00:32:35.600 --> 00:32:41.440] Yeah, I think for me, when I started out, I felt like I'm a creative person.
[00:32:41.440 --> 00:32:42.640] I'm a researcher.
[00:32:42.640 --> 00:32:45.200] If the idea is not from me, it's not my idea.
[00:32:45.200 --> 00:32:51.280] I feel like maybe ashamed to take it and then put it as my my especially when I'm the lead
[00:32:51.280 --> 00:32:55.280] author on the paper and I did not come with the idea, I felt the struggle at the time.
[00:32:55.280 --> 00:32:59.120] And as time goes, I realized that's just a natural process.
[00:32:59.120 --> 00:33:00.960] Yeah.
[00:33:00.960 --> 00:33:01.360] And then.
[00:33:01.360 --> 00:33:08.880] I was just going to say, I'd add to that, that even if you've not come up with the idea,
[00:33:08.880 --> 00:33:16.240] you can sort of compensate during the lifetime of the project in a multitude of ways.
[00:33:16.240 --> 00:33:23.440] Right, in terms of actually making it happen, writing a paper, releasing really good code,
[00:33:23.440 --> 00:33:27.120] you know, that there's multiple ways to compensate.
[00:33:27.120 --> 00:33:29.920] So it's not ever just about the idea.
[00:33:29.920 --> 00:33:34.720] The whole is finding the right idea is definitely a very tricky thing to do.
[00:33:34.720 --> 00:33:35.920] Yeah.
[00:33:35.920 --> 00:33:41.440] Sometimes the idea only gets you so far, like the execution, the implementation, the use
[00:33:41.440 --> 00:33:46.800] cases, all those things become as important as the initial just like moment.
[00:33:46.800 --> 00:33:47.040] Yeah.
[00:33:47.040 --> 00:33:51.440] Yeah, I feel like that's almost more for like mid and PC series.
[00:33:51.440 --> 00:33:55.040] Like you sort of get stuck on an idea for so long because you came up with it and you're
[00:33:55.040 --> 00:33:57.760] like, well, I know for I was the one who suggested it.
[00:33:57.760 --> 00:33:58.880] It's original, it's mine.
[00:33:58.880 --> 00:34:00.640] And so I'm going to make it work.
[00:34:00.640 --> 00:34:04.320] Even though you get to a point, you hit a wall and you're like, maybe this doesn't work.
[00:34:04.320 --> 00:34:07.040] You still try and push through in the beginning.
[00:34:07.040 --> 00:34:10.320] Don't there's no need to do any of this stuff and make it so hard on yourself.
[00:34:10.320 --> 00:34:11.760] Yeah.
[00:34:11.760 --> 00:34:12.400] Yeah.
[00:34:12.400 --> 00:34:12.640] Yeah.
[00:34:12.640 --> 00:34:19.360] Yeah, I think I hear you touched on another point that I wanted to share as my like my
[00:34:19.360 --> 00:34:25.120] experience is always just ask people your questions, what you want to speak, please.
[00:34:25.120 --> 00:34:32.000] And as a junior PhD student, I was really afraid of reaching out to senior either senior
[00:34:32.000 --> 00:34:35.120] researcher in the industry or professors from other universities.
[00:34:35.120 --> 00:34:37.440] I just don't feel like comfortable emailing them.
[00:34:37.440 --> 00:34:41.520] I feel like I'm just a first year PhD student, I had zero SIGGRA paper at the time.
[00:34:41.520 --> 00:34:44.800] How dare I reaching out to them.
[00:34:44.800 --> 00:34:48.880] And as time goes by, I realize it's people, people read their emails.
[00:34:48.880 --> 00:34:50.400] People don't really care who you are.
[00:34:50.400 --> 00:34:53.920] People just care that you're interested in their work and they have a genuine question
[00:34:53.920 --> 00:34:55.680] and you're being respectful.
[00:34:55.680 --> 00:35:00.320] At least that's how I feel when I get email that's about my research.
[00:35:00.320 --> 00:35:02.400] Or they're simply procrastinating.
[00:35:02.400 --> 00:35:03.520] Yeah.
[00:35:03.520 --> 00:35:07.200] If they don't respond, it's probably just that they're just procrastinating or something.
[00:35:07.200 --> 00:35:07.700] Yeah.
[00:35:07.700 --> 00:35:12.640] Almost every time I've ever reached out or someone's reached out to me, it's been almost
[00:35:12.640 --> 00:35:14.240] like, oh, we should do a collaboration.
[00:35:14.240 --> 00:35:18.880] Like we're always looking for more ideas and more work just because we want to be involved
[00:35:18.880 --> 00:35:19.680] in more things.
[00:35:19.680 --> 00:35:22.640] And so usually it's an opportunity to be like, oh, they're working on their instant.
[00:35:22.640 --> 00:35:27.120] And it's also maybe we'll do something or we'll do a follow up once they touch up and
[00:35:27.120 --> 00:35:28.480] like do their project.
[00:35:28.480 --> 00:35:30.080] So usually it's like a good thing.
[00:35:30.080 --> 00:35:34.720] And like you said, no one sort of rejects the email and goes like, I'm not going to
[00:35:34.720 --> 00:35:36.240] share my ideas with you or whatever.
[00:35:36.240 --> 00:35:37.440] Yeah.
[00:35:37.440 --> 00:35:37.760] Yeah.
[00:35:37.760 --> 00:35:44.160] I can share two anecdotal stories that I had that helped me change my mindset around co-emailing
[00:35:44.160 --> 00:35:46.160] people and just asking for what I want.
[00:35:46.160 --> 00:35:54.000] One is in the process of applying for Green Card, I need to get all like a number of expert
[00:35:54.000 --> 00:35:58.400] papers from top university professors that in my field.
[00:35:58.400 --> 00:36:01.840] And my lawyer just suggested to me, you should reach out to the top, top, top people.
[00:36:01.840 --> 00:36:04.560] And I said, I met those people on paper.
[00:36:05.120 --> 00:36:10.000] I read the papers, but if I asked them to write a recommendation for me, well, they
[00:36:10.000 --> 00:36:11.760] say, yes, they don't even know me.
[00:36:11.760 --> 00:36:13.440] And they said, just go ahead.
[00:36:13.440 --> 00:36:16.320] Like in our experience, they usually will respond.
[00:36:16.320 --> 00:36:17.040] And I did.
[00:36:17.040 --> 00:36:19.280] And I got 100% response rates.
[00:36:19.280 --> 00:36:20.560] So that's great.
[00:36:20.560 --> 00:36:24.480] Those are professors from the top schools like Stanford, like Harvard, you name it.
[00:36:24.480 --> 00:36:31.520] And that's the one moment I realized, oh, if I just be clear of what I want, what I demand
[00:36:31.520 --> 00:36:35.920] of their time, and they usually would do the favor for me.
[00:36:35.920 --> 00:36:38.800] And the other thing is even more reasons.
[00:36:38.800 --> 00:36:44.800] In this summer, I'm working with a few interns on kind of podcast project because I'm doing
[00:36:44.800 --> 00:36:45.840] research on podcast.
[00:36:45.840 --> 00:36:52.240] And we want to do formative interviews to learn what existing podcast do.
[00:36:52.240 --> 00:36:57.040] And one of my favorite podcast company is Freakonomics.
[00:36:57.040 --> 00:36:58.560] I don't know if you know Freakonomics.
[00:36:58.560 --> 00:37:02.240] They have a franchise of podcasts and books and blogs.
[00:37:02.240 --> 00:37:07.120] And I just co-emailed them, say, hey, I'm working on research projects.
[00:37:07.120 --> 00:37:08.640] Do you want to answer our questions?
[00:37:08.640 --> 00:37:11.680] And maybe we'll build a tool for you to help you.
[00:37:11.680 --> 00:37:13.920] And the next day they replied.
[00:37:13.920 --> 00:37:19.520] And I got in touch with their-- I forget their title, but it's senior, senior in their company.
[00:37:19.520 --> 00:37:21.680] And they would like to talk and share their use case.
[00:37:21.680 --> 00:37:24.400] And so that really changed my mind.
[00:37:24.400 --> 00:37:29.440] Like I was telling my interns just the response rate might be low, but we just need the few.
[00:37:29.440 --> 00:37:33.600] We don't need like 100% respond rate from all these famous podcasters.
[00:37:33.600 --> 00:37:41.280] And so to sum it up, my advice to junior PhD student is just reach out to people that you
[00:37:41.280 --> 00:37:44.480] think that can help you unblocked, that can help you grow.
[00:37:44.480 --> 00:37:47.040] That could be very technical questions.
[00:37:47.040 --> 00:37:48.480] Like I don't understand this math.
[00:37:48.480 --> 00:37:50.720] I don't understand how you implement that thing.
[00:37:50.720 --> 00:37:53.840] Or it could be more philosophical, more career growth.
[00:37:53.840 --> 00:37:57.920] Oh, I want to intern in your group in the next summer.
[00:37:57.920 --> 00:37:59.360] I like what your companies do.
[00:37:59.360 --> 00:38:02.160] I like your past intern projects.
[00:38:02.160 --> 00:38:03.120] Can I work with you?
[00:38:03.120 --> 00:38:08.160] I think all this are more-- I think people will appreciate that.
[00:38:08.160 --> 00:38:08.480] Yeah.
[00:38:08.480 --> 00:38:12.160] - I think just reaching out to people saying, I want to collaborate with you,
[00:38:12.160 --> 00:38:15.760] it needs to be sort of lower level than that.
[00:38:15.760 --> 00:38:20.880] So I think to your point about, I read your paper and I didn't understand the math.
[00:38:20.880 --> 00:38:24.000] That is a perfectly great question to ask in my mind.
[00:38:24.000 --> 00:38:27.440] And in fact, people will respond to that question.
[00:38:27.440 --> 00:38:30.560] Whereas if you write something generic, like I want to collaborate with you,
[00:38:30.560 --> 00:38:31.840] well, you're not going to get anything.
[00:38:31.840 --> 00:38:42.720] So often papers are missing details, mathematical details, programming details, et cetera.
[00:38:42.720 --> 00:38:46.480] And yeah, ask those questions.
[00:38:46.480 --> 00:38:50.560] - It's quite funny how much people will share if you off the paper.
[00:38:50.800 --> 00:38:54.400] On the paper, everything is perfect and everything worked 100%.
[00:38:54.400 --> 00:38:58.240] And then you pull them off the side, either an email or in person,
[00:38:58.240 --> 00:38:59.680] and you say, oh, how did you do this?
[00:38:59.680 --> 00:39:04.160] And they'll just kind of like sigh and be like, well, this and that and that.
[00:39:04.160 --> 00:39:06.800] And if you look at it from this angle, it works even better.
[00:39:06.800 --> 00:39:10.560] And you're like, OK, there's a lot of extra knowledge that wasn't written down,
[00:39:10.560 --> 00:39:14.080] but it still works as intended, maybe in the paper and whatnot.
[00:39:14.080 --> 00:39:18.880] And they're way more willing to share that in these sort of less formal settings
[00:39:18.880 --> 00:39:22.640] or over email when you're telling them I'm struggling because I want to do this too.
[00:39:22.640 --> 00:39:24.320] And they're like, oh, yeah, try this trick.
[00:39:24.320 --> 00:39:26.160] We found that helped in a lot of different papers.
[00:39:26.160 --> 00:39:29.760] - We need a rotten tomato for archived papers.
[00:39:29.760 --> 00:39:31.920] - Exactly. That would be so useful.
[00:39:31.920 --> 00:39:40.400] - Yeah. My other point I wanted to share, which relates to Rohan's point of
[00:39:40.400 --> 00:39:42.880] long-term investment in yourself, like be patient with yourself.
[00:39:42.880 --> 00:39:45.120] I am a very impatient person.
[00:39:45.120 --> 00:39:48.240] I want to see the result immediately.
[00:39:48.240 --> 00:39:52.400] I want to get instant gratification, like whether I write a half-edged structure
[00:39:52.400 --> 00:39:55.600] or something else, I just want to see the result immediately.
[00:39:55.600 --> 00:39:58.240] And the way I'm keeping myself, like,
[00:39:58.240 --> 00:40:03.680] patient with myself and investing myself is just to do the things that really motivates me.
[00:40:03.680 --> 00:40:08.080] Like, for example, when I in my PhD second year, maybe,
[00:40:08.080 --> 00:40:11.200] I realized I'm not really that into math.
[00:40:11.200 --> 00:40:12.960] I'm not good at it at all.
[00:40:12.960 --> 00:40:15.360] And a lot of my collaborators, they were good at it.
[00:40:15.360 --> 00:40:19.600] And they helped me with all the math derivation and stuff.
[00:40:19.600 --> 00:40:22.640] But I realized I'm more interested in the physical part of things,
[00:40:22.640 --> 00:40:25.280] the implementation and eventually into 3D printing.
[00:40:25.280 --> 00:40:31.520] Like, I enjoy playing with the objects and anything I learn on that direction,
[00:40:31.520 --> 00:40:34.560] it's just, I can work till really, really late at night.
[00:40:34.560 --> 00:40:35.600] I don't feel tired at all.
[00:40:35.600 --> 00:40:38.640] I grow really fast in those domains.
[00:40:38.640 --> 00:40:43.120] And quickly I became the 3D printer guy in the lab.
[00:40:43.120 --> 00:40:45.600] Better for worse, I'm pulling to all these repair jobs.
[00:40:45.600 --> 00:40:53.360] But my point is, I made very little research progress on those areas for two years.
[00:40:53.360 --> 00:40:58.240] And I was fine with it because I'm really motivated on 3D printed stuff.
[00:40:58.240 --> 00:40:59.440] And eventually all paid off.
[00:40:59.440 --> 00:41:02.320] Fortunately, all paid off.
[00:41:02.320 --> 00:41:04.880] And I got papers out of it and made friends.
[00:41:04.880 --> 00:41:07.200] And it led me into internship and so on and so forth.
[00:41:07.200 --> 00:41:11.600] So if I had to stick to more mathy projects,
[00:41:12.240 --> 00:41:15.520] maybe I would not have completed my PhD.
[00:41:15.520 --> 00:41:20.800] Maybe I would have quit it to pursue other direction.
[00:41:20.800 --> 00:41:23.760] I'm glad that I picked something that I'm really motivated
[00:41:23.760 --> 00:41:28.320] and then get to just grow slowly but surely in those directions.
[00:41:28.320 --> 00:41:34.800] One way that I realized you can also manufacture motivation in some sense
[00:41:34.800 --> 00:41:40.240] is to work with students who are sort of of your age
[00:41:40.240 --> 00:41:42.320] or in a similar sort of bracket.
[00:41:42.320 --> 00:41:48.320] Because weekly meetings with advisors, you might not make progress over a week.
[00:41:48.320 --> 00:41:54.320] But with students in your lab, it's just a lot more informal.
[00:41:54.320 --> 00:42:01.760] So just day-to-day conversations, you're often at the same level of understanding.
[00:42:01.760 --> 00:42:04.320] If it's like a dense mathematical topic,
[00:42:04.320 --> 00:42:08.240] which might be easier for your advisor to understand.
[00:42:08.240 --> 00:42:10.800] But you folks might be at the same level.
[00:42:10.800 --> 00:42:13.680] So just silly questions go back and forth.
[00:42:13.680 --> 00:42:17.440] And then ultimate clarity comes out of it from those silly questions.
[00:42:17.440 --> 00:42:23.920] So really talking with people your own age in your lab
[00:42:23.920 --> 00:42:31.360] without worrying about who's going to be the first author on this project or whatnot.
[00:42:31.360 --> 00:42:32.320] Put all of that aside.
[00:42:32.320 --> 00:42:33.360] It doesn't really matter.
[00:42:33.360 --> 00:42:37.360] What matters is sort of the longer term collaboration.
[00:42:37.920 --> 00:42:38.400] Right?
[00:42:38.400 --> 00:42:41.520] If you're talking with someone in your first year, second year,
[00:42:41.520 --> 00:42:44.080] and you're just really hitting it off,
[00:42:44.080 --> 00:42:48.400] then you can have three or four years worth of a really good time
[00:42:48.400 --> 00:42:52.240] where two people or three people are being simultaneously productive.
[00:42:52.240 --> 00:42:57.920] So yeah, chat with your own lab mates.
[00:42:57.920 --> 00:43:01.440] I think that is something that really shouldn't be discounted.
[00:43:01.440 --> 00:43:03.840] And chat technical things.
[00:43:03.840 --> 00:43:07.040] It pays off, I think, completely.
[00:43:07.520 --> 00:43:08.720] Yeah, I agree.
[00:43:08.720 --> 00:43:12.160] I think chatting with your own seniority group,
[00:43:12.160 --> 00:43:15.680] like you're taking the same course, you're learning the same thing,
[00:43:15.680 --> 00:43:17.600] I think it's really helpful.
[00:43:17.600 --> 00:43:24.480] Do you think it has to be physically the same lab or like a remote would also work?
[00:43:24.480 --> 00:43:26.480] Because, for example, working in industry,
[00:43:26.480 --> 00:43:29.360] one thing I realized is that most of my colleagues, I don't see them often.
[00:43:29.360 --> 00:43:31.520] They're not based in Seattle.
[00:43:31.520 --> 00:43:32.960] And even for those who are in Seattle,
[00:43:32.960 --> 00:43:35.120] we don't come into the office on the same days.
[00:43:35.120 --> 00:43:36.720] So I never get to see them.
[00:43:36.720 --> 00:43:42.000] But I still get to learn from them and pick their brains whenever I need to.
[00:43:42.000 --> 00:43:45.120] And for PhDs, and I think it's for universities in general,
[00:43:45.120 --> 00:43:47.360] it's less remote driven.
[00:43:47.360 --> 00:43:48.400] It's more physical.
[00:43:48.400 --> 00:43:49.840] We're on campus, we're in the lab.
[00:43:49.840 --> 00:43:56.000] But I wonder, since what you're working on might not be exactly the same as your peers,
[00:43:56.000 --> 00:43:59.520] because if you're working on the same topic, you're kind of in the bad situation.
[00:43:59.520 --> 00:44:01.200] You don't want to work on the same topic.
[00:44:01.200 --> 00:44:04.880] But you might be working on similar topics with somebody else across the globe.
[00:44:05.920 --> 00:44:09.200] And do you think it'll be helpful to set up like interest group,
[00:44:09.200 --> 00:44:11.600] not based on university, but based on,
[00:44:11.600 --> 00:44:13.840] hey, you both like to work on half-edge structure.
[00:44:13.840 --> 00:44:14.400] Let's talk.
[00:44:14.400 --> 00:44:18.720] I mean, I can give sort of my two cents here.
[00:44:18.720 --> 00:44:22.720] Sort of towards, you know, I feel like I had projects
[00:44:22.720 --> 00:44:25.440] with sort of different group sizes in some sense.
[00:44:25.440 --> 00:44:30.960] And what I found was sort of, you know, the larger and larger the group,
[00:44:30.960 --> 00:44:33.600] sort of the less productive the meetings would get.
[00:44:33.600 --> 00:44:38.720] And even if this is just interest groups and students, you know, a bunch of them.
[00:44:38.720 --> 00:44:43.520] What all towards, I don't know, I felt what sort of started working well,
[00:44:43.520 --> 00:44:47.440] and this really depends on the project and, you know, the scope of the project was,
[00:44:47.440 --> 00:44:53.920] you know, if you have like two students who are sort of equally invested, right?
[00:44:53.920 --> 00:45:01.600] And an advisor or two advisors, those groups of three and four, you know, work sort of really well.
[00:45:03.280 --> 00:45:08.720] So it was almost like, yeah, find a friend, you know, you can find many friends,
[00:45:08.720 --> 00:45:12.320] maybe find one or two friends who are sort of equally invested.
[00:45:12.320 --> 00:45:17.280] I think that sort of, you know, there's a little bit of luck involved here,
[00:45:17.280 --> 00:45:18.720] finding the right person, all of that.
[00:45:18.720 --> 00:45:26.240] So I would add to that and say, even beyond the projects and the actual work that you're doing,
[00:45:26.240 --> 00:45:31.600] find friends of your age in your group, find friends of similar age, maybe a year or two up.
[00:45:31.600 --> 00:45:35.120] So you can ask more like mentor type questions and definitely find friends
[00:45:35.120 --> 00:45:39.680] somehow in other groups, if you can, to collaborate, to throw ideas.
[00:45:39.680 --> 00:45:44.720] And essentially it's a six year project, like process or five or four, however long it is,
[00:45:44.720 --> 00:45:50.480] you're going to need someone to help you just to commiserate just for mental health purposes,
[00:45:50.480 --> 00:45:52.480] just to complain about stuff at some point.
[00:45:52.480 --> 00:45:56.640] And it's easier if that person's around and you're going to go to these conferences
[00:45:56.640 --> 00:46:00.080] where you get there and everyone already knows each other above you.
[00:46:00.080 --> 00:46:03.120] All these professors are off doing their own thing and catching up.
[00:46:03.120 --> 00:46:06.080] And it's intimidating because you see all these other people catching up
[00:46:06.080 --> 00:46:09.200] and you're sort of just there meeting everyone for the first time.
[00:46:09.200 --> 00:46:13.360] And so if you recognize someone from when you were visiting schools, right,
[00:46:13.360 --> 00:46:18.000] and trying to decide where to go, and now you guys can sort of communicate and become friends,
[00:46:18.000 --> 00:46:21.360] it makes the whole like conference floor that much less intimidating,
[00:46:21.360 --> 00:46:24.080] that much more exciting, fun, and just like relaxed.
[00:46:24.080 --> 00:46:28.160] If you're there watching talk with someone else and sort of catching them up and other people are
[00:46:28.160 --> 00:46:32.080] seeing, oh, you have friends at this university or whatever, or now you can collaborate and maybe
[00:46:32.080 --> 00:46:36.000] you don't work on the same thing, but one day someone else does and you get to be the person
[00:46:36.000 --> 00:46:38.320] that connects them. And that's a great feeling also.
[00:46:38.320 --> 00:46:43.920] And so friends across the board, wherever you can make them and hold on to them because they're
[00:46:43.920 --> 00:46:47.920] going to help you when things get dark and when you're not making any progress for a while,
[00:46:47.920 --> 00:46:51.600] or when everyone else is publishing paper that year and you're just attending the conference,
[00:46:51.600 --> 00:46:54.800] that's okay. Like that happens. There's good years, there's off years.
[00:46:54.800 --> 00:46:59.840] And so, but having these people around sort of helps push through all of that.
[00:46:59.840 --> 00:47:08.000] Yeah. I wonder if some PhD student would set up like a Gather Town page and then just open that
[00:47:08.000 --> 00:47:12.480] out as they're working. And then if somebody else wants to ask them a question, they would just walk
[00:47:12.480 --> 00:47:16.080] up in the Gather Town and then like interrupt. Because right now we have this same thing in
[00:47:16.080 --> 00:47:20.480] physical space. I'm in my lab and then Hickey and I, we used to just interrupt each other all the
[00:47:20.480 --> 00:47:26.640] time to bounce the idea off. But now with all this remote technology, I wonder if that will change
[00:47:26.640 --> 00:47:32.560] how this same interaction will happen. I think also with larger groups, then, you
[00:47:32.560 --> 00:47:38.080] know, sort of the usual thing sets in that you feel embarrassed to ask a question and whatnot.
[00:47:38.080 --> 00:47:46.080] So whenever the groups are smaller, I feel like you feel less embarrassed. You're sort of willing
[00:47:46.080 --> 00:47:53.200] to ask things and share things. Whereas, you know, public Slack channel or public Discord channel
[00:47:53.200 --> 00:47:56.960] with lots of people in it, even if it's an interest group.
[00:47:56.960 --> 00:48:02.160] Like you don't want to ask, I don't know, it sort of naturally is like, I don't want to ask
[00:48:02.160 --> 00:48:07.600] the dumb question. Yeah. It's also, you can create some sort of like weekly meeting or
[00:48:07.600 --> 00:48:11.600] monthly meeting, but then it just becomes like a chore, like this check-in where you feel like
[00:48:11.600 --> 00:48:17.600] you have to ask certain questions. So it's possible it's just like spontaneous or just like vent
[00:48:17.600 --> 00:48:22.720] about what happened, what your day or something like that. Then that's when you really build a
[00:48:22.720 --> 00:48:27.040] connection. And not when it's like a plan, like meeting regular meeting that you're checking in.
[00:48:27.040 --> 00:48:30.640] And so if you have someone that you're slacking from another group, or you're just going to like,
[00:48:30.640 --> 00:48:34.720] Hey, you want to hop on a call for five minutes? I'm like a little bit stuck in this. How are you
[00:48:34.720 --> 00:48:39.840] doing? Also tell me about all this stuff. Like the more irregular type meetings is when you create
[00:48:39.840 --> 00:48:43.760] these sort of like connections. And that's when you can sort of learn that everyone else is going
[00:48:43.760 --> 00:48:47.680] through the same stressful and intimidating process of trying to come up with ideas and
[00:48:47.680 --> 00:48:53.680] getting stuck behind things. And they'll give you their tips and all that stuff will help.
[00:48:53.680 --> 00:48:57.440] Even if you're remote, I feel like that's when you can sort of just like send them
[00:48:57.440 --> 00:49:02.000] 15 emojis in a row that make no sense, but at least you can sort of like express yourself
[00:49:02.000 --> 00:49:07.840] a little bit. And they'll just like it. And that's all you need is that someone else is out there
[00:49:07.840 --> 00:49:14.320] listening. Yep. Yep. I like today's conversation. I'm looking at our
[00:49:14.320 --> 00:49:17.600] outlines. Is there anything else that we want to add that we didn't get a cover?
[00:49:17.600 --> 00:49:25.600] Uh, I don't know. Brohan, you mentioned something about this quote you really liked on,
[00:49:25.600 --> 00:49:28.480] on how to like start off or what is it?
[00:49:28.480 --> 00:49:36.800] Oh, it's sort of a two minute YouTube on Ira Glass talking about how to be creative
[00:49:37.440 --> 00:49:43.600] and how, you know, you, you, you get into the business of being creative or, you know,
[00:49:43.600 --> 00:49:49.120] doing something that you're passionate about because you have good tastes, but, uh, you can't
[00:49:49.120 --> 00:49:55.440] sort of execute and produce things of equal quality, uh, as a sort of the taste that you have.
[00:49:55.440 --> 00:50:01.280] So, you know, just, this sort of goes over how, you know, that just requires time and patience
[00:50:01.280 --> 00:50:05.920] with yourself and just keep working. And the quality of the things that you produce sort of
[00:50:05.920 --> 00:50:11.440] improves over time. So just two minutes, definitely worth a watch. Maybe you can sort of put it in the,
[00:50:11.440 --> 00:50:18.160] um, in the, yeah, in the YouTube. Yeah. I'll put in the description. Yeah.
[00:50:18.160 --> 00:50:21.520] I think that makes sense. Cause when you, when you're reading all the papers, trying to come up
[00:50:21.520 --> 00:50:25.760] with ideas or trying to catch up to the state of the art, you're reading people who have polished
[00:50:25.760 --> 00:50:29.600] their work, submitting it and talking about it and discussing it. And you're like, this is really
[00:50:29.600 --> 00:50:34.480] impressive. I want to do that. But that sort of tastes develops over time. Like it's almost,
[00:50:34.480 --> 00:50:39.280] you should try and read some first author papers from people who have like no background, whatever,
[00:50:39.280 --> 00:50:44.240] uh, and just try and get a better sense for how it works. Um, like papers of different conferences
[00:50:44.240 --> 00:50:48.640] of different quality and stuff like that. Um, just so you can sort of build up to confidence
[00:50:48.640 --> 00:50:53.200] and build up your own taste and see how things are done differently. And then in your, your second,
[00:50:53.200 --> 00:50:58.160] third year, fourth year, you started, you know, what your style is and what you prefer to do.
[00:50:58.160 --> 00:51:05.360] And so allow yourself to build that over time. Yeah. Yeah. I think, I think maybe in a future
[00:51:05.360 --> 00:51:10.400] episode, I think what I get from the hourglass quotes, I think everybody gets different from
[00:51:10.400 --> 00:51:18.480] that quote. Um, my take is that we should do a lot of repetition on the stuff we're interested in,
[00:51:18.480 --> 00:51:24.080] like in just writing paper, we should just keep writing papers. And until we figure out what,
[00:51:24.080 --> 00:51:29.760] what makes it, what makes a good paper and so on and so forth. And, and if we want to write more
[00:51:29.760 --> 00:51:35.120] papers, we need to kind of iterate on project faster. If we take 10 years to write a paper,
[00:51:35.120 --> 00:51:39.520] you're only going to write a few papers in your lifetime. Instead, if you were at a paper every
[00:51:39.520 --> 00:51:45.040] single day, you can print a lot of paper, but then the quality of the paper is, is, is, yeah,
[00:51:45.040 --> 00:51:49.120] it's not very good and you don't learn from it. So, but what's the right balance. I think even
[00:51:49.120 --> 00:51:55.280] within our computer science field, like there are some fields, some sub fields that are project
[00:51:55.280 --> 00:52:00.080] cycles. It's a lot shorter. And there are some project fields, maybe C graph is one of them,
[00:52:00.080 --> 00:52:06.400] or systems paper are much longer. It takes a year or two to, to, to, to complete a project and what
[00:52:06.400 --> 00:52:11.200] that may affect the disputes and their development. I think that's an interesting topic in the future.
[00:52:11.200 --> 00:52:16.560] If we want to discuss it, yeah. Cause that's only affects how I chose project and how I
[00:52:16.560 --> 00:52:22.640] pivot. Yeah. Yeah. Well, that's, that's why SIGGRAPH has the conference track now, right?
[00:52:22.640 --> 00:52:27.280] Because too many people complaining that the SIGGRAPH is too much. Yeah. Yeah. That's new
[00:52:27.280 --> 00:52:32.320] this year or last year. Is it last year? Okay. Yeah. We should, we should talk about that in a,
[00:52:32.320 --> 00:52:42.720] in a coming episodes. Yeah. Yeah. All right. Taking hands of the devil. Yeah. Yeah. All right.
[00:52:43.440 --> 00:52:48.560] If there's no other thing you want to add, I think this, this wrap, wrap up this episode really well.
[00:52:48.560 --> 00:52:56.800] Hope our, our PhD listeners can learn something. Let us, let us know what you think or what worked
[00:52:56.800 --> 00:53:04.000] for you. If you finished off. All right. All right. See you guys next time. Bye.
[00:53:04.000 --> 00:53:06.080] Thank you.