WEBVTT

00:00:05.100 --> 00:00:09.080
ERIK DEMAINE: So with every
sort of advanced class

00:00:09.080 --> 00:00:11.640
that I teach there's
a final project.

00:00:11.640 --> 00:00:13.250
And the goal of
the final project

00:00:13.250 --> 00:00:15.660
is for students to
somehow get their feet

00:00:15.660 --> 00:00:17.830
wet with the material
and sort of experience

00:00:17.830 --> 00:00:21.490
it at a more researchy level.

00:00:21.490 --> 00:00:23.750
In general, this can be
things like surveying papers

00:00:23.750 --> 00:00:25.750
that I didn't cover in
the class because there's

00:00:25.750 --> 00:00:28.290
only so much you can
fit in one semester.

00:00:28.290 --> 00:00:31.250
So they'll go and read other
material and kind of aim

00:00:31.250 --> 00:00:33.760
to teach that to the students.

00:00:33.760 --> 00:00:36.270
So there's a written
project part,

00:00:36.270 --> 00:00:38.780
and then there's also a
presentation in class.

00:00:38.780 --> 00:00:42.700
So this is an opportunity
for students to learn more.

00:00:42.700 --> 00:00:45.849
But actually most
students in this class

00:00:45.849 --> 00:00:47.640
aim for a particular
type of project, which

00:00:47.640 --> 00:00:50.310
is to solve an open problem.

00:00:50.310 --> 00:00:52.870
So actually advance the
field and then present

00:00:52.870 --> 00:00:54.100
their solution.

00:00:54.100 --> 00:00:56.930
Now, of course, not
everyone succeeded.

00:00:56.930 --> 00:00:58.300
And that's fine for a project.

00:00:58.300 --> 00:01:00.050
You don't have to
solve the problem

00:01:00.050 --> 00:01:01.520
that you set out to solve.

00:01:01.520 --> 00:01:03.410
But at least you try
different approaches

00:01:03.410 --> 00:01:05.300
and see what doesn't work.

00:01:05.300 --> 00:01:08.020
But actually most students
actually solved their problem.

00:01:08.020 --> 00:01:08.900
Often in groups.

00:01:08.900 --> 00:01:10.930
So the group size is unlimited.

00:01:10.930 --> 00:01:14.110
So sometimes 2, or
3, or maybe more,

00:01:14.110 --> 00:01:17.530
I think maximum 4
students got together

00:01:17.530 --> 00:01:19.610
and solved their
problem jointly and then

00:01:19.610 --> 00:01:21.740
present their solution.

00:01:21.740 --> 00:01:26.189
So I think, I mean it's
really exciting for me

00:01:26.189 --> 00:01:27.980
to show that they
engaged with the material

00:01:27.980 --> 00:01:31.480
so much they can actually
add new material.

00:01:31.480 --> 00:01:34.797
But it's also, I think, really
exciting for the students

00:01:34.797 --> 00:01:36.880
to show, hey, I can actually
solve these problems.

00:01:36.880 --> 00:01:39.070
Many of the students were not
theoretical computer science

00:01:39.070 --> 00:01:39.570
students.

00:01:39.570 --> 00:01:41.540
This is not their
kind of main area.

00:01:41.540 --> 00:01:46.800
But they were able to conquer
it anyway and have fun doing it.

00:01:46.800 --> 00:01:50.130
So I think it's
a nice confidence

00:01:50.130 --> 00:01:53.460
builder that actually--
you know, it sounds hard.

00:01:53.460 --> 00:01:55.230
The field is called hardness.

00:01:55.230 --> 00:01:57.740
But actually it's
pretty accessible.

00:01:57.740 --> 00:02:00.720
And I think it gives
them the confidence

00:02:00.720 --> 00:02:03.130
that they can go back
to their research fields

00:02:03.130 --> 00:02:05.910
and say, hey, how can we prove
that our problems are hard.

00:02:05.910 --> 00:02:08.340
And justify why we're
using heuristics

00:02:08.340 --> 00:02:11.249
or why we're using
exhaustive search.

00:02:11.249 --> 00:02:12.540
Maybe there's something better?

00:02:12.540 --> 00:02:13.160
Actually, no.

00:02:13.160 --> 00:02:15.590
We can show that's the limit.

00:02:15.590 --> 00:02:17.629
We encouraged, for
all the projects,

00:02:17.629 --> 00:02:19.420
not only to collaborate
with other students

00:02:19.420 --> 00:02:22.210
but to collaborate
with us, the teaching

00:02:22.210 --> 00:02:24.510
staff, the TAs, and myself.

00:02:24.510 --> 00:02:28.090
So we would give them advice,
especially in the beginning.

00:02:28.090 --> 00:02:30.300
So there's like a
project proposal stage,

00:02:30.300 --> 00:02:32.500
we would confirm that,
to our knowledge,

00:02:32.500 --> 00:02:34.842
this problem had not
been solved before.

00:02:34.842 --> 00:02:37.300
And then we would give them
lots of pointers for, you know,

00:02:37.300 --> 00:02:38.690
you might try this technique.

00:02:38.690 --> 00:02:41.440
You should rewatch
lecture seven video

00:02:41.440 --> 00:02:44.230
and see how those
techniques apply.

00:02:44.230 --> 00:02:47.500
Or maybe you could reduce
from this or reduce from that.

00:02:47.500 --> 00:02:52.210
Lots of different ideas for how
they might solve their problem.

00:02:52.210 --> 00:02:54.380
And then some of them
came to us afterwards

00:02:54.380 --> 00:02:58.050
like, well, I've gotten this
far but I'm stuck again.

00:02:58.050 --> 00:02:58.860
Any ideas?

00:02:58.860 --> 00:03:01.250
And so I think the
more you can talk

00:03:01.250 --> 00:03:05.310
to people about where
you're stuck the less stuck

00:03:05.310 --> 00:03:06.550
you will get.

00:03:06.550 --> 00:03:10.250
And so that helped a lot,
I think people generally

00:03:10.250 --> 00:03:11.550
found the solution.

00:03:11.550 --> 00:03:12.980
Some people brought
their problems

00:03:12.980 --> 00:03:14.580
to the open problem
session and had

00:03:14.580 --> 00:03:17.080
all the students look at it.

00:03:17.080 --> 00:03:18.890
So there were lots
of ways to do it.

00:03:18.890 --> 00:03:22.540
But generally, collaboration
was the answer.

00:03:22.540 --> 00:03:24.620
So the last part of the
project is for students

00:03:24.620 --> 00:03:27.560
to present their work in class.

00:03:27.560 --> 00:03:29.980
Usually it's a whole bunch
of student presentations,

00:03:29.980 --> 00:03:31.270
one after the other.

00:03:31.270 --> 00:03:34.320
So each presentation
is relatively short.

00:03:34.320 --> 00:03:36.870
I forget if it was
like 10 or 15 minutes.

00:03:36.870 --> 00:03:39.710
But I think it's really
valuable experience.

00:03:39.710 --> 00:03:42.230
Presenting is also hard.

00:03:42.230 --> 00:03:44.440
And the way to get good
at presenting your work

00:03:44.440 --> 00:03:45.760
is to do it a lot.

00:03:45.760 --> 00:03:49.050
And so, generally,
I find MIT students

00:03:49.050 --> 00:03:50.102
are quite good at this.

00:03:50.102 --> 00:03:51.810
They've had, I think,
a lot of experience

00:03:51.810 --> 00:03:53.500
presenting their work.

00:03:53.500 --> 00:03:56.310
But one of the things I
like to stress in my classes

00:03:56.310 --> 00:04:00.120
is, because most of
the projects were

00:04:00.120 --> 00:04:02.250
joint with multiple
students, how

00:04:02.250 --> 00:04:05.387
to navigate a joint talk
where one person talks

00:04:05.387 --> 00:04:07.220
for a few minutes or
one slide and then they

00:04:07.220 --> 00:04:09.100
switch back and forth.

00:04:09.100 --> 00:04:10.559
And I give general
advice about how

00:04:10.559 --> 00:04:12.850
to do this but for the most
part the students just kind

00:04:12.850 --> 00:04:13.558
of figure it out.

00:04:13.558 --> 00:04:18.310
And by practicing they
find a good way to do it.

00:04:18.310 --> 00:04:20.790
There are no videos of
the student presentations

00:04:20.790 --> 00:04:23.734
because I think that would be
a little bit too intimidating.

00:04:23.734 --> 00:04:25.400
It takes a lot of
practice before you're

00:04:25.400 --> 00:04:30.110
comfortable being video recorded
like this and in your lectures.

00:04:30.110 --> 00:04:34.460
But they are totally
of video quality,

00:04:34.460 --> 00:04:36.010
so it's a shame we
can't share them.

00:04:36.010 --> 00:04:39.260
But it's a chicken and
egg problem, I guess.

00:04:39.260 --> 00:04:42.600
And so at the end of the
class the TAs and I each

00:04:42.600 --> 00:04:45.440
write a bunch of bullet points
about their written project,

00:04:45.440 --> 00:04:48.870
and the work that they did,
and their presentation.

00:04:48.870 --> 00:04:50.730
And I pay particular
attention to like

00:04:50.730 --> 00:04:52.489
how strong their
voice is, do they

00:04:52.489 --> 00:04:54.280
have good presence,
and that kind of thing.

00:04:54.280 --> 00:04:56.230
And try to give
them that feedback

00:04:56.230 --> 00:04:58.380
so they know where
they need to improve,

00:04:58.380 --> 00:05:01.170
or where they're
especially good,

00:05:01.170 --> 00:05:03.050
or that the slides
needed more figures.

00:05:03.050 --> 00:05:04.620
That's my number one comment.

00:05:04.620 --> 00:05:06.520
Usually the answer is yes.

00:05:06.520 --> 00:05:08.860
Can never hurt to
have more figures.

00:05:08.860 --> 00:05:11.660
And also, on the
research side, we

00:05:11.660 --> 00:05:13.690
think-- we look
at their results,

00:05:13.690 --> 00:05:15.190
tell them whether
we found any bugs.

00:05:15.190 --> 00:05:17.394
But for the most
part we're thinking

00:05:17.394 --> 00:05:18.560
what would be the next step?

00:05:18.560 --> 00:05:20.250
Like, oh, here's
another open problem

00:05:20.250 --> 00:05:23.080
that would advance your
problem a little farther.

00:05:23.080 --> 00:05:24.630
And here's some
ideas for solving it.

00:05:24.630 --> 00:05:28.660
And that often leads to
papers that we publish.

00:05:28.660 --> 00:05:31.480
And some final projects
are basically done.

00:05:31.480 --> 00:05:33.247
Their papers are
ready to submit.

00:05:33.247 --> 00:05:35.830
And then we give them feedback
on where they should submit it,

00:05:35.830 --> 00:05:38.320
what would be a good
conference, a good journal,

00:05:38.320 --> 00:05:41.360
we can basically
help them publish it.

00:05:41.360 --> 00:05:45.104
So that's part of
the full service.