Press "Enter" to skip to content

Are We Hearing The Best Ideas At The Table? | Simon Cullen | TEDxPrincetonU


many researchers are working at the
moment on really interesting questions
about how technology and communication
interact so there are really fascinating
questions about what happens to online
communities when people can communicate
anonymously or what happens when people
communicate via text or by say 140
character chunks these questions are
fascinating but it’s really striking
that when we want to have a discussion
about something that matters we actually
don’t really rely on technology we still
just sit around a table open the holes
in our faces jibber jabber and that’s
pretty much how we do it when it matters
so that’s pretty much the state of the
art in technology for in person
discussions this image captures nicely
the theme that technology is really
drawing us away from the things that
matter if you search your favorite image
search engine for a distracted person
about half of the images will show you
people being distracted by their gizmos
the same thing is true if you search for
a distracted student it’s inevitably
something to do with technology so today
I want to talk about one way we can use
technology to improve the quality of the
discussions that we’re having in the
classroom that is to say in person
discussions so group discussions are a
very natural place to begin because
whenever more than five or six people
get together in a group to have a
discussion they face a problem they need
to figure out a way to decide who gets
to talk next and for how long they need
to decide on who’s voice will be heard
the usual way of doing this probably
hasn’t changed in 50 thousand years that
is to say some sort of senior person
maybe the tribal leader or a professor
will stand at the front of the group and
will pick people who have opted into the
discussion by dangling their hands in
the air above their heads that’s pretty
much the way we do it now but that
generates some really big problems so
all of these problems derive in various
ways from the fact that the moderator
the person choosing who gets to talk
next has no
information that’s actually relevant to
promoting a good discussion so the
moderator doesn’t know what anyone wants
to say doesn’t know how strongly they
want to say it doesn’t know how valuable
this contribution would be instead of a
different contribution that we might
choose instead so the moderator is in
this information poor environment I
thought a really nice way to explain
this to you would be to look at some
photos of students in discussions but
then I discovered that if you just went
by the brochures that colleges and
universities produce you would be very
misled and deceived about the nature of
college so one thing you would go
thinking is that everybody asks
questions in class or nearly everybody
but a more subtle way that you would be
deceived is you would think that
everyone either puts their hands up nice
and high
I’m ambiguously wanting to contribute or
they put their hands down really clearly
and low unambiguously not wanting to
contribute actually luckily I got a
photo of the audience earlier today and
so we can look at what real people
actually look like so here you are
earlier in an earlier talk and as you
can see many people are doing what the
people in the stock photos do they have
their hands up nice and high maybe it’s
that guy from your philosophy precept or
they have their hands down really
clearly but they also assume a variety
of intermediate and ambiguous poses as
you can see at least person up there is
you know pretty low hand this one down
here actually it looks like from the
blur like this person was actually going
to reach up higher they they really
wanted to express a strong desire to
country but they were just a bit slower
so this illustrates that when we use
hand raising to distribute discussions
to moderate and decide who gets to talk
things like brute speed how fast you
raise your hand will determine whose
voice gets hurt a host of really
interesting other features will also
play a role
things like how introverted or
extroverted are you how comfortable do
you feel in social situations and then a
really interesting one how socially
desirable is it for you to communicate
to the whole class that you really want
to contribute that last one is really
interesting because we know that it
varies between different groups so here
is some data from a study that to
Harvard Akana
did it looked at Dadar on how popular
kids are in high school objective
measures of popularity for 90,000
students and what they look at here is
the relationship between how popular you
are and your GPA your grades at high
school so what do you find for white
kids is roughly what you would expect
the kid with the 4.0 and 300 is got more
friends about 1 1 or 2 more friends then
the kid who’s flunking this however is
not true for black students there you
can see there’s very little benefit for
clearing the first three points on the
GPA scale and then actually it might
even turn around for the fourth point
and the situation is even more dire for
Hispanic students so this suggests that
different students from different groups
will incur different costs when they
raise their hand to indicate that they
wish to contribute and they will incur
those costs whether or not they actually
get to contribute simply putting
yourself forward having a chance is
enough to broadcast to all of your
friends that you’re pretty eager to
contribute to class that you’re a
studious person and that may have
different costs for you ok so the sorts
of factors that we should probably
expect will play a big role in the
psychology of hand raising or
determining which of those many
intermediate and ambiguous posers people
assume well it shouldn’t come as so much
of a surprise given that these are the
sorts of things that are doing the work
that actually who gets to talk in a
classroom might depend on prejudice and
bias this is an extremely widespread
practice it’s important to remember this
is a photograph from a seminar which or
a conference that i sat in not too long
ago this is professional people a whole
lot of people who’ve got PhDs in
philosophy and politics and they do the
same thing as these kids who are
preschool aged they raise their hands
and someone who doesn’t know anything
picks people arbitrarily well I want to
think about how we can do that better
but just to give you a sense of the
scale it’s about 1.3 billion students in
the world if each one of them raised at
their hand about once every 5 hours of
instruction and you’re the about half a
meter up and half evaded down you can
calculate
but actually in an average year the
world students move their hands through
a distance of 234 billion meters which
to put that into perspective is six
thousand times around the earth or one
and a half times the distance to the Sun
it actually takes about a hundred
billion seconds or forty human lifetimes
every year for just students dangling
their hands that’s asbestos a really
conservative estimate okay so where else
do we use this
well education extensively teachers love
hand raising Q&A boardrooms political
debate and ethical inquiry these are
things which matter deeply so why are we
still seeing caveman technology to do it
the question bothered me and I started
to think about a better way to moderate
group discussions and because I work in
education I’d be thinking about this in
the context of classrooms and so in the
remainder of this talk I just want to
tell you a little bit about how I’ve
been doing that humans are remarkably
good at many many things but we are not
good at the sorts of things you need to
be good at to moderate a discussion for
example no one in this room can keep
track of how many seconds each of 15
people has spent contributing to a
discussion for the last 30 minutes no
one in this room could even know much
less calculate and remember what sort of
contributions a variety of people want
to make so like I said a few times we
just yes well computers are extremely
good at this
computers are extremely good at
measuring accurate things and doing
calculations very quickly if we could
yet software to take care of this part
of moderating a discussion we would free
the humans to do the thing which they
are uniquely good at listening really
carefully to our students thinking hard
about what they need to hear to learn
more so I’ve been thinking about
building an app to do this the app is
extremely simple it’s currently a
prototype it’s called pallava and the
main way of interacting with it for
students is just by dragging this slider
to indicate how much they want to talk
cost they can do this very privately
just using their cell phones so when you
drag the slider the app will detect that
you want to talk and it will then enter
you into the queue if you’re not happy
with where you ended up in the queue
you can say you want to talk more
urgently and see if that bumps you up
further you shouldn’t try that trick too
often though because it is normalizing
the values that come off these sliders
so overly dominant and overly shy
students are over time treated quite
equally so the moderators interface if
you know the professor our graduate
student it is also dead simple it is
basically two buttons the button to
select the next speaker and the button
to start a new topic so when students
enter the discussion they indicate
whether they want to make a comment
following up on something someone has
already said or whether they want to
contribute a substantially new idea and
so the moderator can select when they’re
ready to go down a new pathway the
moderator will see the queue updated in
real time they don’t have to juggle any
of this information anymore early in the
semester of course we really nice not
having to remember students names which
always terrifies me and generally they
will be free to now focus on being a
better teacher so let’s just step
through how it works when it’s finally
your turn to talk the app will tell you
that you can set a warning if you’re
someone who maybe tends to ramble a
little bit or your againg that guy from
the philosophy precept you might
actually no want to be that guy and so
you might elect to have a timer set
which alerts you whenever you’ve been
talking for 30 seconds say and so that’s
really easy to configure this really
isn’t at all rocket science so the app
will tell you 30 seconds the moderator
the person running the discussion
decides in a hard limit how long do you
want to let people ramble on for let’s
say we set it to 90 seconds then when
you get close to that limit it will
start counting down and then it will cut
you off so it is never going to happen
in a preset or a discussion section
where they’re using this that someone
gets to take over and dominate the
entire conversation it’s extremely
difficult for moderators especially
early on in teaching careers to say to
that student I think we’ve had enough of
this idea now let’s put it on the table
and maybe we will return to it if we
have time it’s really hard to confront a
person like that this means we never
need to do that again the other really
cool transformative effect of
switching away from hand raising and
using these globally interconnected
supercomputers in our pockets is that we
now get analytics you know like Fitbit
but for your discussions so we get a
second-by-second accurate measure of who
spoke and who wanted to speak and for
example you could get an email that
automatically notifies you you spoke
more than 96% of students at Princeton
University this week if you got that
message
you might modulate your own behavior
without anyone having to confront you in
this awkward way and say you know you’re
really smart no one else can understand
what you want to tell them
so whining just come to office hours I
tried that once it actually worked so
this is a system which can automatically
send people non-judgmental feedback on
how they’re performing in discussions of
course this applies for shy students as
well maybe you are able to sit in
discussions hardly talking at all but
you convinced yourself I don’t know
maybe I’m in the bottom 15%
well if you learned that you were
actually contributing less than 96
percent of students at your clich maybe
you would automatically try to
contribute a bit more so this could have
a really transformative effect and the
answer is we don’t know whether it will
do anything we don’t know how any of
this will work because we’ve never
really studied it so how does this app
work the queuing algorithm is really
flexible it basically takes metadata
about the discussion to determine who
gets to talk next so it’s really not
rocket science it counts the number of
times you’ve wanted to contribute to the
discussion it counts how long you got to
talk for it counts the strength of your
desire to speak but again remember you
can’t game it too much by just if you’re
someone who just always thinks what you
have to say it’s unbelievably important
it must be right now it will just start
to ignore you
okay and other discussion metadata so an
effective and fair moderation system has
got to be informed by a study of real
discussions that students actually have
in classrooms when students start to use
this app they will have the opportunity
to contribute to scientific research
project where we will ask them a whole
lot of introductory questions questions
about their sex and gender and race and
Nationality educational background
socioeconomic background and also
interesting questions about things like
their openness to experience their
experience of social anxiety and will be
able to see what happens when you look
at different groups of people and put
them all together in a room is it true
that the group composition really has an
effect on how well the discussion goes
can we not just randomly put students
into precepts but begin with some
predictions about which groups will have
the best discussions none of these are
questions that we can answer now but
hopefully we’ll have a data set very
soon that will allow us thank you very
much and I hope I’ll get to update you
on the results of this study soon
[Applause]
Please follow and like us: