00:00:00
Dan Loney: Well, despite numerous
challenges both here in the
00:00:02
United States and Europe, Google
remains at the top of the ladder
00:00:07
in terms of the world of search.
But a recent paper looks at just
00:00:11
how that is occurring. Pleasure
to be joined by Leon Musolff, who
00:00:15
is an Assistant Professor of
Business, Economics and Public
00:00:19
Policy here at the Wharton
School. Leon, great to have you
00:00:21
with us today. Thanks for your
time.
00:00:23
Thanks so much for having
me, Dan. Always happy to talk
00:00:26
more about this fascinating
antitrust case.
00:00:29
It is, and it's obviously an
interesting topic, especially
00:00:33
when you think about how long
Google has had this position in
00:00:37
the world of search. Let me
start, I guess, with how you go
00:00:40
about researching
something like this.
00:00:44
So the core idea of our paper is
to run what we call a field
00:00:48
experiment. So we want to
actually observe how people make
00:00:51
their choices between search
engines in as natural
00:00:55
environment as possible, so
literally on their computers.
00:00:59
For this, we developed a little
browser extension that can
00:01:02
basically tell us which search
engine they are using at any
00:01:04
given time, and then we walk
them in a survey through various
00:01:10
sort of what we call treatments.
So for instance, we might pay
00:01:14
some users money to try out a
different search engine, but
00:01:18
that money, that incentive, runs
out after two weeks. And the
00:01:22
question is, you know, once that
incentive runs out, if I paid
00:01:26
you some money to try Bing, for
instance, are you then going to
00:01:29
switch back to Google, or are
you going to stick with Bing?
00:01:32
And as you might imagine, the
answer could depend on whether
00:01:35
we remind you, you know, "Hey,
you are now not being paid
00:01:39
anymore to try a different
search engine. So do you want to
00:01:43
switch back?"
00:01:44
How does, then, the quality of the
experience play into the
00:01:49
decision of which search engine
they might use?
00:01:54
Right. So the idea is that— that
choices between search engines
00:01:58
could be driven by a couple
of different things. So one of
00:02:01
them is, like, the quality of the
search engine. But also, there
00:02:05
might be an aspect where there
is a search engine that's just
00:02:08
the default in your browser.
Indeed, we know that Google pays
00:02:12
around $26 billion a year to
remain the default on various
00:02:16
browsers and devices. So you
might think, you know, a lot of
00:02:19
people are just going to go with
whatever the default search
00:02:21
engine is. And then the last
thing that could play in is, you
00:02:25
know, not the true quality, but
sort of perception of quality
00:02:29
that might not necessarily be
accurate. If I ask you, you
00:02:32
know, "Have you ever used a
search engine that is not
00:02:35
Google?" The answer might be, "No.
I've never even tried this." And
00:02:41
because of that, you might not
actually know, you know, what the
00:02:44
quality is of the competition.
00:02:46
So you mentioned using Bing in
this research. What was the
00:02:50
general consensus from the
people that you had in this
00:02:53
research about Bing? And did
they want to stay with Bing, or
00:02:58
did they feel more comfortable
or just fall back in terms of
00:03:02
using Google after the
experiment?
00:03:05
Right. So the core group, where—
this is an interesting question,
00:03:09
is the people that we tried to—
that we paid some money to try
00:03:13
Bing, but then we reminded them,
and we said, "Hey, now that
00:03:16
you're making an active choice
again, now that we're not paying
00:03:18
you money again, do you want to
switch back?" And what we saw
00:03:21
there is that about a third of
these users stuck with Bing. And
00:03:25
that surprised us, because that
suggested that, you know, after
00:03:29
experiencing both search
engines, there was actually a
00:03:32
sizable minority that seemed to
prefer Bing to Google. Now I
00:03:36
want to be careful to say, you
know, a third of users is
00:03:39
definitely not 100% of users.
The vast majority of users
00:03:42
still prefer Google, even if
they are informed. But it is
00:03:46
interesting to know that, you
know, about a third of users can
00:03:50
be shifted in this way if they
just were to experience the
00:03:53
other search engine.
00:03:54
But there— there is that component
of people just kind of logging
00:03:57
on to, probably, whatever search
engine they are using that is,
00:04:02
you know, first up kind of in
the rotation on their— on their
00:04:05
device. And they stick with it,
and they don't really consider
00:04:09
whether or not it's Google,
Bing, Microsoft, whatever it is.
00:04:13
One other thing that we did as
part of this experiment is, for some
00:04:16
people, we paid them some money
to switch the default search
00:04:20
engine sort of very briefly, and
then we just never reminded them
00:04:24
of this fact again. And one
thing that was maybe less
00:04:28
surprising but still very
interesting, is that these
00:04:31
people, at very high rates,
stuck with that changed search
00:04:35
engine, even though they had
previously been using a
00:04:38
different search engine, right?
So this default does seem to
00:04:42
have an outsized influence on
what search engines people
00:04:45
choose. But there is an
interesting subtlety here,
00:04:47
because the default for most
people is Google. Because people
00:04:50
use Chrome, or on their mobile
devices in Safari, always, Google
00:04:54
is the default. But Google is
also the search engine they
00:04:58
would choose if they were to
sort of actively think about it
00:05:02
without experiencing a different
search engine. You might think
00:05:04
that's because Google has a
really high brand value or
00:05:07
something like that, but that
means that just changing what is
00:05:10
the default search engine would
work. But just having people
00:05:14
make an active choice rather
than, like, giving them the
00:05:16
default of Google would not work
in changing Google's
00:05:19
market share.
- I know, in the terms of
00:05:21
this research, you're talking
00:05:22
about Google, and you use Bing
as the example. But it makes me
00:05:26
also wonder about something like
Microsoft with Edge, and whether
00:05:30
or not we should expect, either
already or down the road, that
00:05:35
Microsoft would be able to gain
a certain advantage because of
00:05:39
Edge, and it going into so many
of their devices.
00:05:43
100% agree, and I think that has
already happened. So in our
00:05:46
study, we explicitly separate
out Chrome users and Edge users.
00:05:50
And that's actually really
useful for us, because, like you
00:05:52
say, the Edge users have a Bing
default. And so that allows us
00:05:57
to sort of, you know, get some
more sense of how much defaults
00:06:01
influence choices. And we do
find that Bing has a huge
00:06:05
advantage because of Edge. And
indeed, you know, I keep
00:06:08
referring to Bing and Google as
the only two search engines, when
00:06:12
we all know there is a bunch of
other search engines. That's
00:06:14
because they jointly have a
really, really high market
00:06:18
share. You know, on the order of
95%. And that is, we think, exactly
00:06:22
because Google has the default on
the Chrome browsers, and Bing has
00:06:27
the default on the Edge browser.
00:06:29
What do you think, then, that this
research could mean for the
00:06:34
policy path? Because certainly
there is still lots of
00:06:37
discussion about the potential
of monopoly that Google has in
00:06:43
the area of search.
00:06:45
No, I totally agree with you,
and indeed, Judge Mehta is
00:06:48
deciding right now what remedies
to impose on Google in this
00:06:54
antitrust trial, after ruling
last summer that Google is a
00:06:57
monopolist and has sort of
violated anti-competitive
00:07:02
clauses. And what we think would
be most useful is any policy
00:07:08
that has sort of two key
features. The first feature is
00:07:12
that it must respect users
choice. So the dangerous thing
00:07:15
about defaults is that users are
not making an active choice.
00:07:18
They're just, sort of just
sticking with what they're
00:07:21
presented with by the browser.
So we need to do away with that.
00:07:24
But the interesting additional
fact that our research uncovers
00:07:28
is that that's not quite enough,
because users do not know the
00:07:32
quality of all the products. So
somehow a successful policy
00:07:37
needs to expose users to
alternative search engines. To at
00:07:40
least one more search engine,
not just Google. And one way one
00:07:44
could accomplish this is by
defaulting people into, say,
00:07:49
Bing or some other search engine
at random, and then after a
00:07:53
certain period of time, say, two
weeks, have them make an active
00:07:56
choice. So then you would have
both of those features where
00:07:59
they experience— at least a bunch
of them experience a different
00:08:02
search engine because they just
stick with the default. But then
00:08:05
we also sort of rescue them from
that other search engine if they
00:08:09
really dislike it. And that
second part is really important.
00:08:12
Just making some other search
engine the default is really
00:08:15
bad, because we find that users
massively prefer Google, on
00:08:18
average, even if there is some
exceptions to that rule.
00:08:22
And probably at this point in
time, with how much we use the
00:08:27
main search engines out there,
it might be hard to expect to
00:08:32
see a third company, kind of
outside of Microsoft or Google,
00:08:36
being able to gain enough market
share where you can have all
00:08:41
three of them on a level playing
field.
00:08:45
Yeah, so I totally agree with
that. One core complication
00:08:48
there that we haven't mentioned
before is that there might also
00:08:51
be what we call feedback
effects. So basically, if you
00:08:54
have more users, you learn more
about what results users like,
00:08:57
and so it becomes easier to
serve relevant results. And we
00:09:00
find in our paper that these
effects are not so important for
00:09:03
the competition between Bing and
Google, because they are both
00:09:05
already quite sizable. But if
you were a small company
00:09:09
starting a new search engine,
that sort of effect would
00:09:11
certainly limit your ability to
compete with the big boss. Yeah.
00:09:16
Leon, great to have you with us
today and get your insight.
00:09:18
Thank you very much, sir.
00:09:20
Thank you so much
for having me. Have
00:09:21
a great rest of your day.
00:09:22
Thank you. Leon
Musolff, who is an Assistant
00:09:25
Professor of Business Economics
and Public Policy here at the
00:09:28
Wharton School.
