0:52
But, now the problem is that each of you is on the same air.
So, your signals are naturally going to collide with one another.
And caused interference. Now you're probably thinking, well this
is really bad, right. Because our signals are hitting each
other which is naturally going to cause them to get jumbled up?
But at the same time, shouldn't your cellphone, your talk, be coming into my
phone. Shouldn't this conversation be coming
into my phone? This one, too?
And at the same time, shouldn't all of my conversation be going into the output of
your phone and so on? Shouldn't we all be able to hear each
other's conversations then? Well, clearly you know that that's not
the case, because you've been next to someone, I'm sure.
While you been, both been on the phone before and that doesn't happen.
But why doesn't it happen? I mean, clearly it looks like it should.
1:43
Well, that boils us down to whole idea behind sharing.
So network operators have had to deal with different ways of getting us to be
able to share the network medium, and it's kind of similar to if you've ever
shared an ice cream with someone else. So each of you will split the ice cream
cone maybe in half, and You can each eat half of it and then you're both happy.
Now, the problem is that what happens if one or more people want the same ice
cream cone? Then each of you gets a smaller and
smaller share and that's the same idea with sharing the air is that as we have
more and more users, each of us gets a smaller share of that resource.
And so we have to do the same with less. So a similar thing would happen to you if
you're ever on your laptop and you go to a local coffee shop.
If you try to connect to the internet you're trying to go to this point over
here with your transmission that's going to connect you to the rest of
everyone on the outside world. But you've probably noticed that lots of
times your signal when you're sitting in a coffee shop is a lot worse than it is
when you're at home. But why should that be if you're on wifi
in both places? Well, the reason is that a lot of other
people have the same idea. They all want their coffee too.
So they come to the coffee shop and each of you is trying to talk to this point.
[SOUND] And at the same time this point is sending its transmissions back to each
of you. And so, those are naturally going to
collide and cause a lot of lost messages and require devices to have to retransmit
as we say the messages and will look at ways the engineers help make it that.
So let's move on to our second principle now, the idea that consensus is hard.
And again, this will be covered in the material in the course.
So, what I want to start off by doing is considering three different examples of
cases where we have to find consensus, and how it's hard in each case.
First one is ranking set of items. So if we items A, B, C and D, say, we
have to come up with a consensus as to which one is the most important and which
one is the least important in the ranks accordingly.
So maybe B is most important, A is the 2nd, C is the 4th, D is the 3rd.
So while we need some information to be able to come up with this consensus, it
might come in the form of something like this, where we have relationships between
the items. Here those relationships are the forms of
referrals. So if I'm on this smiley face, which is
one of these items, then I could get referred to either the yellow item, the
red item, or the green item. And likewise if I'm on this red item, I
could get referred to this item or this item.
And we could use those referral relationships to establish an idea of
what's called importance. And then we use that importance to come
up with this consensus idea. But clearly, there's no right answer
because there's no saying what we should and shouldn't do with that information.
And also, as we get more and more people involved from more and more items the
answer gets much more more complicated as to what the right consensus is.
The second example I want to consider is the idea of an auction.
And with an auction, we have a number of bidders, maybe we have a few, one, two,
three bidders here, and you can have a single or multiple item.
So I'll draw three items on this side and we want to determine what the proper
matching is between bidder and item and that will depend upon what they bid.
But again there's no, necessarily right answer.
Maybe the matching has a set answer. The highest bidder will get the most
valuable item. Second highest will get second most
valuable, and so on. But then, what do we charge them?
And that's something again, that doesn't have a right answer.
And it gets more complicated as we have more and more bidders.
The third thing I want to consider is something similar.
So, in an option is when we have voting. So, when in a voting system, we have each
person submit a vote which is like a ballot where they rank each of the
candidates. Maybe we have candidates A, B, C and D
and then we have to use these votes here to determine which person wins, which
person loses. So, in this case maybe we say that C
wins. B gets second place.
D gets third place, and A gets fourth. I'm just making this up.
And so again, as we have more and more people voting, and we have more and more
candidates and so on, this is going to get much more complicated.
So this is all great, and now you can see why we're finding consensus in these
situations and why it's hard, but how's this relate to networking?
Well, turns out that companies on the web every single day, like Google and
Wikipedia, actually use these systems to come up with a consensus on things that
we deal with. So, if we consider a standard Google
query. We get a standard results of ten items on
the first page. And let's not look at the colored items
right now, or the items on the side. We'll look at those in a minute.
And these are pages. And Google has to, uses this process
called page rank. Which is very similar to some of the
ideas we were talking about up here conceptually to determine the rank order
of those pages. And so each of the pages has a referral
which is in the form of a hyperlink in which they connect to each other.
And it uses that to establish a notion of importance and rank the pages
accordingly. At the same time, Google uses the idea of
an auction. To determine who gets allocated to which
ad space. And the way that that's done is each
website here, submits a bid to Google, and Google uses those bids to determine
which person gets which ad space. So, someone gets ad space one of over
here, someone gets ad space two and so on.
And again, they have to determine what to price, accordingly.