0:00:23sure
0:00:35a
0:00:36you
0:00:37i
0:00:39um
0:00:41for
0:00:42yeah
0:00:43i
0:00:44why
0:00:45um
0:00:46i
0:00:49a
0:00:50we
0:00:51oh
0:00:53okay
0:00:53oh
0:00:54right
0:00:55and
0:00:56a
0:00:57uh
0:00:58you
0:00:59no
0:01:00i
0:01:01no
0:01:03a
0:01:03yeah
0:01:04i
0:01:06i
0:01:07and
0:01:09oh
0:01:12oh
0:01:15a
0:01:16you
0:01:18a
0:01:18and
0:01:20uh
0:01:22i
0:01:24oh well
0:01:27i
0:01:28oh
0:01:29oh
0:01:31a
0:01:31was
0:01:34i
0:01:35oh
0:01:35i
0:01:36i
0:01:38and
0:01:39two
0:01:40and
0:01:40i
0:01:41i
0:01:43and
0:01:46ah a
0:01:48i
0:01:48oh
0:01:49right
0:01:51a
0:01:52yeah
0:01:53uh
0:01:55oh
0:02:01where
0:02:02uh
0:02:04i
0:02:05a
0:02:06a
0:02:08yeah
0:02:09like that
0:02:10yeah
0:02:12one
0:02:13i
0:02:15for
0:02:17a
0:02:18but
0:02:20right
0:02:21i
0:02:22i
0:02:23right
0:02:25the moment i
0:02:26so
0:02:27oh
0:02:28a
0:02:29thank Q
0:02:30um and hands and use was that can be enforced and then any moment in time
0:02:35as both the server as a klein
0:02:37i'm going to model this in the following way i'm going to assume that i have a client which can
0:02:41quest a file of some form of service a server
0:02:44which are is an upload quest
0:02:47and and i'm going to assume that
0:02:49um i can model this
0:02:51as shown in here
0:02:52and i going to call is a gift giving game
0:02:55so i'm going to assume that the client has no action
0:02:58he just a request at some content
0:03:00and the set that has
0:03:02an action he can choose to serve and
0:03:06i'd
0:03:06thus that of is to the client
0:03:08or not to serve
0:03:10i'm going to assume that it the step but decided to set the clan to get the if you don't
0:03:14be and
0:03:15otherwise is this that but it and i to serve you to get the benefit of zero
0:03:19so clearly thus client has the benefit
0:03:22from
0:03:22being served by the server
0:03:24a set on the other hand in the cost of C so has a to do would be minus C
0:03:29and
0:03:30the um in the case of doesn't
0:03:31serve
0:03:32the cost will be zero
0:03:34so that a yeah why actually can action a one prime in that action there would be no incentive for
0:03:39the server to a
0:03:41to to to two to of and hands she would like to not serve and fear
0:03:47so what is important in this context is to try to bring some form of incentives
0:03:51to can is the users
0:03:52to serve each other
0:03:53and that is the goal of the proper design of let's call or than that design
0:03:58yeah are many ways to implement incentives to not lex
0:04:01the most studied within the out of engineering a a see propagation for example in P two P a was
0:04:06people assume of his study of past see propagation out that users
0:04:11how about in the current and vitamin we can see that that is not a good model because i may
0:04:15not have a his study of past see propagation of the particular person
0:04:19so i know a couple of you in the audience but moment ready to of you i don't know
0:04:22however i may be interested in getting some content from you and you might be interested in getting some content
0:04:27from somebody else
0:04:29but because of that fast see the of interest
0:04:32as as well as the that of being dynamic can use of starting and going i may not be able
0:04:36to have a he story of few are particular
0:04:39um
0:04:39a
0:04:40personal his story of reciprocation
0:04:43has in such situations
0:04:45we use social of
0:04:46um
0:04:47if C propagation
0:04:48and this also reciprocation skins so a from based on reputation and going to discuss that in know to do
0:04:53that
0:04:54so like a like less you will is you are going to implement this
0:04:57source show
0:04:58all if you provocation in a rigorous way as follows
0:05:01i'm going to assume that the of code design or would try to design a social norm
0:05:07and he's going to do that the by introducing the following elements for he's going to a to and you
0:05:13here were every agent
0:05:15a set of imputation labels L
0:05:17but number of L would be part of the design
0:05:20um then a peer cannot but the a G
0:05:24a web are given a particular reputation a label for the
0:05:28so that for the client with from seen that that he be can decide to starve not this
0:05:34uh
0:05:34that the protocol designer can
0:05:37design a social norm as follows
0:05:39she can sell like the set of reputation label
0:05:43she can determine i mean the social strategy what you would like to see it in that actions and merging
0:05:48so given the reputation of the sort of a and the reputation of the client be if i'd upon be
0:05:54a at the G to start models that or
0:05:57and if the agent was a P D and
0:06:00he's its reputation would be increased
0:06:02otherwise it would be decrease
0:06:05not to though that we deal with that of fish agent here so they may decide that it's not in
0:06:09the set of interest
0:06:10to form of this also
0:06:13um so i'm with don't you one more thing and went to assume that you can though maybe somebody comply
0:06:19with the social norm
0:06:20that maybe be a
0:06:22i the reporting errors
0:06:23or
0:06:24something has happened and i i'm a not product the update this reputation
0:06:30um let give an example of a source of i'm going to use in the talk
0:06:34um
0:06:36i'm having to san ninety as i'm going to have a and a to you are going to consider a
0:06:39for base or shows that they G
0:06:41which has a service to old age
0:06:44if the server has a class hold below age
0:06:48and the kind and has a reputation of by of age i'm going to assume always
0:06:53that
0:06:53that was that of so we've one condition or not that happened that would serve
0:06:57well yeah at hand
0:06:59they are not going to serve if the server has a reputation that is higher than that of the plot
0:07:06uh yeah
0:07:07you and this type of test whole base also of the G
0:07:10uh i i can have much more sophisticated social that i give a talk earlier today day which can that's
0:07:15a more sophisticated strategies
0:07:17but that makes a noise is more difficult
0:07:19so here and once the ring just a simple they G
0:07:22if they you was that's are going to follow the social strategy that put patient is going to increase
0:07:28an and went to assume a marketing a passionate stations
0:07:31if at any moment in find somebody body all bay
0:07:34it's going to be punished
0:07:36or the way the reputation
0:07:39um so
0:07:40they the use that have on the ability to decide that you want to solve all this type of
0:07:46um
0:07:48that's for based scheme
0:07:50so what if you doing kids optimization one
0:07:53so the not would be its own reputation label
0:07:56um
0:07:57reputation
0:07:58distribution of the community which i'm going to consider consider
0:08:01i'm going to call let's community configuration
0:08:04and also the a one and a quotation you may want how that all this information is known
0:08:09well this information could be no maybe there is a track
0:08:12you can come up with distributed versions of the but they are more complicated
0:08:15to
0:08:16to really formal lies and and come up with a eighty percent you
0:08:22oh or going to introduce a simple belief that users are going to hold
0:08:25so the use that's are going to assume that
0:08:28use as that are at a reputation zero
0:08:31um
0:08:32will
0:08:34higher than zero will follow the social more
0:08:36and you data reputation zero they do not provide them
0:08:40now in in order to determine whether a user should for this sort of thought that are not i need
0:08:44to determine not only the for kind a D V D of the user
0:08:48but also the long-term utility of
0:08:50are we use a given the community configuration and the specify social norm you can this actually i yet or
0:08:56not and keep data is that based on its performance in the long term
0:09:00in that you could would be just one shot did you need to go to T decided nobody will comply
0:09:04everybody will feel i
0:09:07so this is the
0:09:08um
0:09:09a problem of that values use that needs to solve
0:09:11so you can use to decide actually what be its best response at every period in time that maximise is
0:09:18it's a long time at that it is a can used to select this at signal which is to leading
0:09:22to the optimal long term that data
0:09:25no that's not from a system point of view what they would like to her
0:09:28so from a system point of view what they would like to have
0:09:31is he would like to have an a me go back to to this first setting with like to have
0:09:35a problem was that
0:09:37is
0:09:38source should the optimal
0:09:40so that we would like is that is not a code design of cell X that's also one
0:09:43that's so long as form on what time by the set of degradation label
0:09:48the reputation update scheme
0:09:50and the social strategy
0:09:53so is to send that this and i mean that
0:09:55such that
0:09:56the community are rice at the unique stochastic a stable like will there is i want to have something like
0:10:02that is if i don't have a unique a key
0:10:06the community make over to different and it different um
0:10:10types of configurations some of which may be efficient
0:10:14and in part because can that i would like that this unique
0:10:17a Q blue which i'm going to define in just one second
0:10:20is corporate
0:10:21so in this unique i Q label everybody will cooperate with the job
0:10:25what you talk i speak at least a be like in meets up quite complicated concept you why there no
0:10:29with already you word this would be kind of a very small somebody you of it
0:10:33but the idea here is is that um
0:10:35i have a strategy at you provide sick to get together with a comedy configuration you
0:10:40and this should be a stochastic a least table like you do if and only if when the at
0:10:45is update errors
0:10:46of excitation go to zero
0:10:48seek my is that that's this response of the use as against
0:10:51the overall population
0:10:53configuration
0:10:55and use new is time-invariant then is that points than i make in do by this particular signal
0:11:01and more than that that the community stay set me with a positive for about
0:11:05okay
0:11:06now that you look so this is a protocol design a problem i'm going to
0:11:10go back to it all the way the end for now let me concentrate on the use
0:11:15and you know what that for this again to be a stochastic a least like you leave room
0:11:19the best is one dynamic when need to ended cold first to decision able actually
0:11:24i for this that to sort is best response then i mean in to solve an in P yeah or
0:11:29finding me with and P the only thing i want to point out here is that we can see that
0:11:32in this and D that this state
0:11:34is the reputation of the user
0:11:36and the configuration of the components
0:11:39given that at additional of they was
0:11:41so best on this
0:11:43big for lies the of a to functions such that we can do that best response than i mean
0:11:48and we want to now look at a given that this and and
0:11:51what that the structural properties of these test is point by not
0:11:55and if is that we have which is very important is that the solution of the previous and be problem
0:12:00in this type of social norm governed
0:12:02all um
0:12:03community is that that's all they the G
0:12:06this is the optimal strategy selected what it means is actually
0:12:09that given the configuration of the client which which i mean that acting
0:12:13i'm going to have a a S for age
0:12:15a that as for than going to start the users and below and not
0:12:20therefore in order to determine the optimal point see we can
0:12:24only it's sufficient to determine the optimal test hold at at very
0:12:29um
0:12:30state
0:12:30yeah so the optimal set service test for that age
0:12:34yeah i can be formulate
0:12:36this and D
0:12:39and and if i do that
0:12:40based on this information is structured is desired
0:12:43i can come up with one more
0:12:44or out
0:12:45also also some form of structure
0:12:47which is that this that's for of remote point
0:12:50has the following property i will now a one two
0:12:54of
0:12:55some body
0:12:56a but have the source norm so the source norm that is me go to so that of people have
0:13:00a bad reputation five
0:13:02i would never go and set some by they shall for
0:13:06that's a first thing
0:13:07a not that is that the service is user provides one one only increases with a but
0:13:12you can think about that as some to think it'd the above a certain and that's about but as the
0:13:17reputation increase is you don't want to get in the start this you have in the system it becomes increasingly
0:13:22that you meant for you
0:13:23to to cheap
0:13:26no that's a look
0:13:27do you this type of structure properties how as the community of or
0:13:31because you would like to solve this from a proper code design a point of view
0:13:35so for thing is we can prove that the community of all following a markov of chain
0:13:39and that is that's these phones and i'm mean of the user
0:13:42in each state of the markov chain is a community configuration
0:13:46more of what we can show is that is not of jane P as the following properties
0:13:51that exist actually is T you are shown we have stability and a D
0:13:55and i in a some of that just to be able to define a early meeting configuration is if you
0:14:00are shown which i'm going to court it all make a hat
0:14:03to which the community will
0:14:05oh four to as this at all
0:14:08oh reputation errors
0:14:09then the words your
0:14:11so that has a configuration is stochastic least stable when it's probability in on that data
0:14:17star or or make a on had well my um
0:14:20bar should say
0:14:21is a larger than zero
0:14:22that's one i'm going to say that the configuration is to classical state
0:14:28you that what i can can in terms of gorgeous of the community the following first is that the limit
0:14:33in configuration distribution diffusion unique
0:14:36has the comedy converges to early make it's a top of the degradations in the long down so i just
0:14:41i want give a small set of configuration
0:14:44and this is very good because otherwise the design space will explode
0:14:47a second the and says that the set of stochastic least able configuration and the set of configuration be long
0:14:53it to this stochastic take a like people are the fame
0:14:56what does that really mean
0:14:58it means that we can and otherwise
0:15:00only the stochastic or stable like you be a
0:15:03we can use yeah like the stochastic least able it can either by analysing think the that of stochastic least
0:15:08stable configuration quotations
0:15:11okay so i can have a weight to
0:15:13conduct that i Z school version of the community
0:15:16what i want to do
0:15:17about
0:15:17when they want to do the design no
0:15:20for we can cut that i've barely meetings set
0:15:23all of this
0:15:24stochastic stable configuration and if we do that we have a first you're that which says
0:15:29well as anything in this network will converge i that two
0:15:32zero which means never cooperate operate everybody here writing
0:15:36or or always corpora
0:15:38and thing in the middle of from one to a minus one
0:15:41it's not going to get a present
0:15:43so i that everybody serves
0:15:45or nobody sir
0:15:47yeah the question becomes of the proper up design of or
0:15:50how what he design
0:15:52this type of interaction between users how should he determined the social strategy the G that it could station
0:15:58a a data would and the set of excitation
0:16:01to let use that as she as the competitive out
0:16:04other then the defect about
0:16:07for that what he is to do is to set that that's hold in the also at
0:16:12depending on the pot i'm it that's of the form
0:16:15and is to set it in such a weighting which
0:16:18depending on the cost of the benefit ratio
0:16:21as well as the discount factor of the
0:16:23on utility of the users
0:16:25this test is smaller than is about
0:16:28so the design becomes
0:16:30mainly focus on the H
0:16:32well as that really em you know what that is by as
0:16:34let's look at that in in state of result
0:16:37what actually in here is the following thing first i show that
0:16:41how the social well if there very is for a community as the cost increase as the cost of service
0:16:47okay
0:16:48that as increases
0:16:49i assume a i peak the network
0:16:52with the following parameters
0:16:54and what i want to show with the following first let's look at the blue lines
0:16:58what used the blue as is you optimal corporation this is something like a network utility maximization of do if
0:17:05i have self less use that's for be the end user
0:17:09they would always but it they were always feel a protocol
0:17:12and this is the best i can do
0:17:13that is in is goes down is because the cost increase so
0:17:17even we user corporate
0:17:18the cost increases as well
0:17:21hence
0:17:22what i want to show here is
0:17:24if you have a B D and is that everything goes fine
0:17:27one yeah hand if i need to determine the optimal
0:17:32um um
0:17:33talk
0:17:34which is sustainable
0:17:36meaning the optimal protocol which is
0:17:39oh bait by D set interested users is what i'm getting is this to be an bar see here
0:17:45so what is see that to be a design
0:17:47two things happen
0:17:48for is the cost becomes too high
0:17:50but one want to out that it that a will collapse you cannot apply to that interested users to in
0:17:55is high cost
0:17:56relative to the band
0:17:58i se
0:17:59a thing is you can see a as the cost increases
0:18:02you are starting to have gap
0:18:04then if you have this gap with as opposed to the optimal cooperation is because you have a trade
0:18:10between one hand optimality
0:18:12the that was incentive
0:18:14so in order to make this protocols protocol sustainable about you need to give a something in terms of form
0:18:19that we say well this optimization is that in ask and that it difficult what with i why would i
0:18:23go out that of this very difficult
0:18:25um
0:18:26a complicated optimization problem can and not just to was a particular fixed
0:18:29so shows that they using the particular that's for which makes sense
0:18:33if you do that to see that is that in red
0:18:36so what you would see is first the network community will collapse much faster
0:18:40because of the cost
0:18:42increasing and this stuff that you being chosen randomly
0:18:45well so you we see that as the cost increases is that gap to the optimal corporation
0:18:50a what we take away from here is to messages and this would be the last thing and not even
0:18:54going to show the crucial
0:18:56the that's the thing i like to to see is that
0:18:59you can not talk yet about a network to delete the optimization need to talk about sustainable that the direct
0:19:04optimization
0:19:05because the one
0:19:07type
0:19:08problems cannot piece
0:19:10really a really for forty used in this type of environment that is that a selfish
0:19:14and secondly if you want to go well
0:19:16a head and just to something a hall
0:19:18like a lot of the literature are currently in engineering is trying to do to cut people up
0:19:23social strategies and policies is for that but they up the that that ad hoc
0:19:29you performance with for a lot
0:19:31thank you that
0:19:41yeah
0:19:45no
0:19:45i i have every bite your bus has some
0:19:48information some knowledge let's say and we change with each other
0:19:52so i mean change it with you you make change it that on a make change changes if you can
0:19:56and so on
0:20:01no so what happens in here is because we don't have the few story of fife the see location what
0:20:06i had is i have a blob his study
0:20:09you're are study in the that
0:20:10so you don't have a for his study will need that i in the beauty of this work
0:20:14so in an that that it which i that i with the all the time
0:20:16this will not be that i model
0:20:18but an an use dynamic users coming go in the use of have been in different things
0:20:23i would not have the P and to necessarily have met to before
0:20:27and and
0:20:27be a the his study with you
0:20:30so it is a they she's a global level of reputation after it in that action
0:20:34you will have a different excitation
0:20:36of course you can come up with a reputation scheme or maybe be are part of a club of the
0:20:39signal processing a and we should behave in a signal processing cost but there we moved to the information that
0:20:44the club we only to behave was we don't meet these people very often
0:20:47that would be extended
0:20:49i