00:11
Hello, Len. Hello again, Asami. All right, so we have talked long, on and off mic, about
our frustrations on inaccessibility of some of the reading materials,
particularly scientific textbooks, journal articles, not the most accessible. Yeah.
Or reader-friendly in any sort of way. I don't think they even try to be most of the time.
There are a number that don't try to be. Yeah, that's true.
So how, but we still have to read. And we know that the knowledge is there. And you know,
when you take yourself back to maybe early grad school days, like you're getting used to this
new field of research that you know very little about, what sort of things helped you in sort of
get a little crack into this new genre of research, new field? Yeah, getting into a new field
is tricky. I admit, I was sort of prepared and thinking of how to approach any challenging
type article. But let me, let me try to answer this question first. The idea of
entering a new field, I find one of the most, well, helpful, but also somewhat satisfying,
is that I try looking for an assortment of articles and mainly just reading their
introductions. I find that to be useful. I'm not necessarily looking for like what's happening,
but I'm looking for how others are framing their problem, are framing their context,
are framing like the research as a whole. This is because it's, it's sort of... I like that.
It's sort of interesting to me that, which it helps to, you know, kind of contextualize everything.
That's... Like you learn how and what sort of like a lingo that people use to even begin
thinking about this problem. Right, and like depending on how I've arrived with this,
you know, here's a new field that I just opened the door on, right? Maybe I come into it with
some sort of framework to build off of or I don't. And so that's kind of to give me a framework or
a place to start. Okay, so and how do you go about collecting these assortments of articles?
03:04
Yeah, if I'm entering into a field, like I guess I have sort of done recently,
this is maybe not the way that many people start. But I had, in the case that I'm having here,
which is maybe we'll talk about these ones later, right? With some LLM and education stuff.
I have somewhat of a framework to at least begin poking my nose around. I have some,
I don't, maybe I shouldn't call them buzzwords, but they're kind of buzzwords, which are,
you know, the types of words people toss around. They're keywords. Keywords feel almost
not sufficient, because buzzwords are like, please stop saying that word. I need you to like,
but yes, I have keywords to go off of in my mind. And I start thinking, to me, you know,
what are the types of questions or topics that I think is interesting about this particular topic?
And I just start, I just start plugging them in. I just start looking, databases, you know,
even something as simple as like, a Google Scholar search, just to see what starts coming up
without, without much of a true goal. And I don't think this is the best approach. But I think
when I'm just starting off, it's kind of nice to get this feeling for what's out there a little bit
of, you know, what types of titles am I seeing? Oh, my God, I've seen so many of them mentioning
things like this. Okay, I've seen a lot of this now. I think I don't want these types of things.
So I'm going to start removing that from some of my search criteria, right? I'm going to start
taking things out that I want. And so it's a bit of a, an indirect approach to
stepping foot into a field. If I want to be more rigorous, right, it's always great.
If you want to be more systematic.
About it. Honestly, I would, I would just talk to people. That's even, I don't even know if that
counts as systematic, but I'd talk to somebody that either is related to the field, or if I
couldn't find that, I would look for, you know, intended introductory-based material on that field.
Yeah. Yeah, yeah, yeah. No, and I was going to say the same thing, basically, that even though
in this day and age, AI does a really good job of, you know, letting us have access to
these knowledge or letting us sort of, you know, get a foot inside, like they'll summarize it
nicely. It makes us feel like we can sort of understand it. When you are starting something
that you have very little idea about. Yeah. I find it to be most efficient to rely on people
who spent decades already in it. Yeah. Yeah. And again, I don't always have access to these people.
06:07
Right. Typically, if you reach out, they're also not going to kind of deny it, you know,
and I feel like this is the case, especially when you are young and just starting out.
If people have their, like, you know, if people come up to you and say you're interested in
something, they're interested in something you're doing, very few people are going to be like,
no, go away. You know, they would typically have like a go-to, maybe it's a textbook or like
a couple of seminal papers that like people always talk about. And specifically regarding
academic subjects, I try to find the review paper that is published within the like the past five
years. And this number can change depending on a field, right? Like, you know, if it's a fast-paced
moving field, it could be less than five. If it's like a more slower-paced field, it might be more
than five years. That window you need to kind of judge for yourself. But I typically go for something
like a five years, past five years review paper that is most cited. I don't even care like who
is writing it or if it's like a good person, because I don't have a judgment, like I don't
have a compass yet of like whether it's a legit or not. So I just go for what's most cited as a
review paper, and then kind of skim through. And that's basically what I'm doing with
like what you're doing with introduction reading. Like I try to understand what is the general
problem that this field is concerned about, and what sort of questions should I have in mind going
in, right? And at this point, I'm not trying to learn any new knowledge. As a very first goal,
I try to look for review papers that have very high level scan of the field or of the particular
subfield. Like you said, familiarize myself with the language of the way people talk about things.
And usually there's like a handful of golden principle type of things in the field that
people always come back to, or one or three unsolved ultimate problems that people are
trying to tackle at. It gives yourself an upper bound and a lower bound of like the kind of
problems that you can have. That's what I'm trying to gauge by reading these like frequently cited
review papers. And from there, I try to go into like different sections of these review papers
and look at the references in it. Again, at that point, it's very helpful to have an expert eye.
09:06
Again, like I really cannot emphasize enough the importance of doing science as a team,
as a community, and not in isolation, especially when you're starting a new field.
You could waste a lot of time, I think, in kind of spending a lot of time getting familiar with
the peripheral knowledge that you don't really need. And again, if you have infinite time,
please, by all means. But most of us don't. And if you do, please let us know. I'd like to know
what your secret is. Like most of us don't. So I think trying to have a good qualitative limit of
your scope, I think, is a very good first step for learning a new field. And then from there,
you do need to go sort of, you know, at some point, you cannot just do just high level
understanding, right? You do need to get to the nitty-gritty part of it. And I mean, I've only
graduated from grad school less than two years ago, so what do I know? But I find most of the
nitty-gritty foundation comes from my actually undergrad education. Like having that foundation
helps me dive into the nitty-gritty more easily. Because I think, at least in my experience,
grad school was already specialized when I started doing it. The framing was kind of
decided for me by me choosing a particular lab. And at least in my case, I was given a project
that already kind of limit my search on these things. And of course, I had to do a lot of
learning in that field from the foundations. But that was based on a lot of what I did in undergrad.
So I actually found myself coming back to quantum that I did in undergrad,
or linear algebra that I did in undergrad, to refresh my memory and go deeper into
some of the subjects. In that way, I feel like I learned very few completely brand new things
in grad school, in terms of sit-down classroom learning.
Yeah. As you described, this idea of undergrad, sort of where this foundational bit came from,
12:03
or where you have a place that you can... Or even high school.
Where you can investigate the nitty-gritty stuff. It could be from any, right? These were
from possibly high school and undergrad. It's a lot of... It's different for everyone, right?
What their experience was like. But I think there was quite a bit in those moments where you're still
essentially mainly focused on learning, if you're lucky enough and privileged enough to be in that
space. And you're also invested in setting a foundation, maybe even without knowing it,
that that works, right? If you...
Yeah. I didn't know that's what I was doing in undergrad.
Yeah, it's...
Like it kind of... The realization came in a retroactive fashion, where like,
oh, I didn't know that I've heard of this concept before and would need to revisit.
Yeah. Yeah, yeah, yeah. This... I think... So in your case, you had some of those foundational
pieces to kind of go back to. And whether intentionally, unintentionally, right?
It worked out to have developed that framework you needed, right? To approach the nitty-gritty stuff.
If somebody doesn't have that, which is actually kind of what happened to me,
because I had a lot of background within the field, but not within the specific field and sort
of this branch that I had entered. So I didn't really have a great foundation in things like
quantum, for example. I see.
And so what... One thing to do, and this might be different for everyone,
is that you almost have to fill that in, right? You have to go find something that
does give you some sort of grounding. I don't think I had the perfect findings.
I think for a number of reasons, I was challenged to sort of do so in that space.
But it did really change things, even in noticeable amounts, when I found a textbook
that was approachable and leveled. It didn't say obviously and naturally all the time, right?
Something that gets you going, and then you have a little bit more to stand on.
The more you can hold this picture of all of that information in front of you,
I think the more readily, and I don't necessarily mean quickly.
It sounds like you have gotten a feel for what are good questions.
And I think when you start to see the field and say,
oh, well, it seems like you've got all these pieces, but I see you,
a person on the other side of this discussion or paper or textbook,
sort of not mentioning something that feels like it would be really useful.
15:03
And usually, if somebody is not answering that, it might be because there is not an answer,
right? Or there is something missing for that. And that seems to be what you're able to pick up on
really well. And that comes from...
And because I've done this many times.
Yeah, it comes from the process of seeing and asking and questioning.
Right. And this really comes with practice. The first, really first rendition, let's say,
was grad school or maybe even undergrad research. It took me a long time to wrap my head around
what even is the problem that they're trying to solve. What can we do and what can we not do,
right? These are things that I didn't know at all.
Second time around in grad school experience, the research was bigger in scope. So it took me
longer time even to wrap my head around and figure out what's possible, what's not,
what's being answered, what's not being answered. But I've done that. And then now in my postdoc,
again, completely different field where I would say maybe only 20% of my background
really directly applies to this field. To some, it's extremely uncomfortable situation.
Sure, yeah.
But for me, I was honest to them. I don't know anything about machine learning. You still want
to hire me? And they're like, it's fine. It's fine. People learn.
Figure it out.
And then they were fine with it. But I feel like the process was much faster
and less arduous than when I first did it for grad school time. Because again, I have this
circuit that I've activated before going from zero knowledge to one knowledge. That process is
much faster. And one to 10 is much faster than 10 to 20 is much faster and so on and so forth.
Despite only having worked in this new field for about a year at this point,
I feel like I have a better intuition on the scope of what is capable and what is
not or either computationally just too expensive and not practical.
And that I take as a win because I didn't know what even was possible.
Yeah, yeah. I want to validate that process because I imagine those listening are either
experiencing something similar or maybe have experienced or will experience if they're going
18:01
into these types of research fields. It's part of it, right? You're going to build things up
as you go along. And everybody has gone through some version of this. And I want to maybe even
at this point, take it and attach it back to something we talked about in the last episode,
which is that there's somebody who's made this writing. There's somebody who chose to
write the word obviously and naturally and created these, right? There's a person who
has gone through their own version of that experience of building up this idea and this
framework and this meaning. And basically what I'm saying is it's good to remember that every
once in a while and if not more often almost all the time after maybe you've gotten your feet
ready in a new field to remember this especially and go like it's now time to look at this as
people trying to communicate to other people, not here is a statement I'm trying to embed it
in my brain or this statement is a truth that I need to understand, right? That feels very,
I think, connective between that. Recognizing that everybody has those experiences. They begin
to learn things. They develop. They can move faster and faster and faster. Okay, that person
is now doing this task. Right. It's so fast. It's so rapid and efficient to the point that it's so
obvious to them when it's not for the people who are looking at it for the first time. Yeah. So
here's that obvious and, you know, naturally sort of thing and it reminds you. Okay, I take back.
Koen Tenuji, you did all right. It was obvious to you. Oh no, I just helped you like take down
the take down. Not for me. Seven years ago. Right. Which, yeah, which is important, right? It's
the you of seven years ago didn't need that, right? Which is fine and you can recognize it by
saying that person is at a different stage or is not speaking to me at a stage that is equivalent
to mine, right? I need to find something else here and that's fine. You have to find something
else because you're trying to also join the discussion, right? And sometimes, yeah, that's
what you have to do. Yeah, yeah. So again, we tried to aim for a short episode and we didn't,
obviously. Real professionals here. Real, real professionals. But professional academics,
we just talk. We try when we are entering practical tidbits from you then is kind of just like
string together your keywords and see what appears and also talk to people. Yeah. And
21:02
I said, yes, also talk to people to expedite that process. But I also try and take advantage of
review papers. Yeah. And like without even judging which review paper is good, I would just go for
the one that's most cited because that's one easy metric. And after you have an understanding of
the field, you might decide that that was a terrible review paper, but you don't know that
yet. So that's okay. Right. Exactly. So that's, if you're poking around and you're not sure,
you know, pick, just pick a direction and attempt to get a feeling of what is being discussed and
how it's being discussed. This is also a place where you can think about the person behind it.
What is the author of this thing? Like what do they think about this topic, right? What are they
believing or are convinced about after 20 years of research, right? You look at their research
backlog and it's like all about this, you know, they've never changed in 20 years, right? Then
that could be really useful. This has varying degree of difficulties depending on like how much
prior exposure you have and how deep you're trying to get into. Or maybe it's a very,
very fundamental science field that requires like a whole new degree, you know? And I really do think
it's just kind of practice for this. Like there's no magic to it. It just, you have to make your
brain go through this several times in order to start feeling like you can get it. Yeah. So
look at that. Satisfying end. Only triple our original title.
That's definitely within an hour of our time, right?
That's it for the show today. Thanks for listening and find us on X at Ego de Science.
That is E-I-G-O-D-E-S-C-I-E-N-C-E. See you next time.
