In my personal experience (but I'm usually more focused on Powerplay abuse) you should try sending some tickets time to time, especially if something especially anomalous happens. Apparently they mostly check when asked in particular situations.
Livestream Discovery Scanner 6 - Reshaping the Simulation
I got the impression that it wasn't an assumption but based on an "anonymised" snapshot of data over a given period of time (4 days in the example given). Perhaps, because of it provocative nature, you're over-sensitised to botting activity, or perhaps large ship botting wasn't occurring during the snapshot period?
Great stream by the way! Thanks Frontier - I love these Discovery Scanner episodes. As ever, my main takeaway from these is "it's complicated" - far more complicated than a casual armchair observer of the game might think. Mind (once again) blown!
"The deputy hamster herder sends their regards". Almost a year ago, someone in the forums asked who maintains the servers while Dav is on a Livestream, and "the deputy hamster herder" was the reply.
Yeah, as I said it was tough to distinguish commas and dots, which may have introduced errors
I also didn't remember that exact quote so I tried to get something close to what it should be
S-curves everywhere.... I knew it!
Thanks Dav and Will.
Thanks Dav and Will.
Robert Maynard
Volunteer Moderator
Transcript from YouTube:
Code:
05:54 tonight's elite dangerous livestream we have a discovery scanner cooked up for
05:59 you tonight and I'm about to introduce a man who needs little introduction I of
06:04 course and will Flanagan but that's not who I'm talking about
06:07 talking about my guest tonight who is this wonderful man right here who are
06:13 you good evening my name is dev I work on the server team here at frontier
06:17 primarily on the elite dangerous project so all of the all of the servery
06:23 witchcraft when things go a little fiery absolutely the team that works on elite
06:28 dangerous is server team are we're responsible for the galaxy simulations
06:32 making sure that everybody gets a good experience making sure that good Souls
06:38 the missions work to make sure that everybody gets a good experience
06:42 hopefully with minimum disconnections and networking problems but you know we
06:47 wanted to the server's we did what we can and yeah we keep an eye on things
06:51 just to make sure that the galaxy stays ticking along with all of you very many
06:55 plays flying around within it so I'm gonna put a little anecdote because it's
07:00 quite nice when I think when distant worlds to started this this year and we
07:05 were all there you were there you were absolutely were logged in I was logged
07:08 in ready to ready to jump everybody jumps and servers go and we were all we
07:17 were all texting like dev dev servers down yes
07:23 me and the number of people from the team were definitely online that night
07:26 we were keeping a very close eye on the very first mass jump from distant Wells
07:30 to for the weeks leading up to it the team and I brainstormed what we thought
07:35 might be stressed by a large number of commanders jumping at once we spent some
07:38 time engineering certain areas that we thought would would maybe be stressed by
07:42 that and yeah it didn't go entirely smoothly because of something completely
07:47 unexpected just another thing just a dip and of course it was fixed for the next
07:51 week's mass jumping yeah so we tried the best we can but players they just
07:55 continually surprise as illegal in exciting ways and that's that we're
08:00 going to be talking about that tonight actually it's a nice little segue so
08:03 tonight's discovery scanner is a reprisal of what you did a lathe on this
08:08 year and it's going into a little bit more detail about that I'm I could
08:11 ramble on about what it's what this discovery scanner is about okay and I'm
08:15 gonna pass over to you dad because now you're now the star of the show for me I
08:20 cut I've got okay I've got the title reshaping the simulation right dad so
08:26 I'm handing over to you why was this discovery scanner all about absolutely
08:30 so what we're going to talk about this evening is a recap of the background
08:37 simulations within the elite dangerous galaxy and in particular we're going to
08:41 look at the faction simulation which is one of the bigger ones and it is often
08:44 called the background simulation but there are a number of them I'm going to
08:48 recap a little bit about what the what the design is what what we believe the
08:52 the principles that drive the background simulation are and also we're going to
08:56 talk in some detail about the changes that we made last December in beyond
09:01 chapter four and some of the changes that we made using server-side changes
09:05 since then it might give you an insight into maybe some of the design feedback
09:09 some of the ways that we monitor the live galaxies and we iterate upon it
09:13 we've talked about this on live streams before now but this is probably the most
09:18 detail that we've been into for some of the design processes with the way that
09:22 we look at some of the maths and some of the equations I know you like numbers
09:26 well we will do what we can to to present those in a way that might might
09:30 give some insight so there are a number of background simulations in Italy
09:36 dangerous the ones that you'll interact with the most often are things like the
09:40 commodity simulation which says we've got these these star ports and these
09:44 planet bases with some economies we know what commodities are created and
09:48 consumed economies nearby might import and export different commodities and
09:53 plays when they trade in their ships we will record that and set up trending
09:57 trade routes or the galaxy maps so that players can see what's hot and what's
10:00 not maybe identify some trading routes that haven't yet been exploited and
10:05 therefore there might be some really good supply and demand to get like good
10:08 prices we also have the outfitting simulations
10:11 these decide what ships are available in a shipyard or which modules are in stock
10:16 at the shipyard for that week yes they do very occasionally based on some
10:20 situations and the investment state in particular can in which what's available
10:24 through outfitting I would love to talk a little bit about some of the tricks
10:28 that we taught the thyroid simulations in chapter 4 but this evening we're
10:31 going to focus on the faction simulation because it's it's really the biggest
10:35 topic and it's the one this players interact with most often it's worth
10:39 mentioning that one of the keywords for here is these are all background
10:42 simulations they are designed to bring the galaxy to life based on player
10:47 activities they are not necessarily intended to be a end goal in and of
10:53 themselves they are supposed to enrich the game using gameplay loops that are a
10:58 little bit more immediate these are a little bit more gradual they work in
11:02 concert they work across many different players on many different platforms in
11:05 all of the different game modes so they're bringing the galaxy to life but
11:10 they have never been the most visible in terms of their details although we hoped
11:17 to correct some of that in chapter 4 because well we'll come on to some of
11:21 the details in a bit and so dev I'm gonna before we get onto
11:25 that and we're going to just quickly jump to a back swing screen there's a
11:28 little audio issue that we're going to just fix and so we'll be back in about
11:31 two minutes now you know what it's about think about what's going to come
11:39 [Music] [Music]
13:48 and we're back and hopefully you can hear have a lot absolutely hopefully we
13:53 now have sound I mean as fun as this would be to do using mime and
13:57 interpretive dance I think maybe a line yeah no let's go with audio hopefully
14:02 this is a little bit more if you're not do let us know we'll try our best but
14:08 yeah we've tweaked where we can and we don't see an issue on our end so no
14:12 here's hoping right we haven't yet turned it up as far as 11 but got
14:17 megaphones yeah right so I interrupted you davon so sorry but we're looking at
14:21 ultimately we're looking at the the factions of the background sent
14:24 absolutely so a quick recap of the faction simulation so this is the design
14:30 principles that the faction sim has been implemented in since version 1.0 in 2014
14:36 so the the idea is that all player actions can have an effect on the share
14:42 galaxy a faction in the game and there are dozens of thousands of factions
14:47 across populated space and there were a lot of populated systems so a faction
14:52 represents the interests of the people and the population who are in that star
14:56 system they are tied to markets and assets and assets can come in all
15:03 different shapes and sizes and as the game has progressed and we've added new
15:08 content to the game in new games activities to the game we have added new
15:11 assets for factions to control so I've included screenshots of star ports of
15:16 some of the new space installations of a planet installation which and a planet
15:22 ports but the important thing to take home here is assets can be large they
15:26 can be small they can control the system or that can control nothing in
15:29 particular they might not even be dockable in a ship but they're still
15:33 then performing an economic function for the for the state of the galaxy things
15:38 like science we outposts or industrial installations they all take part in the
15:43 economic simulation of the game and when a faction changes ownership of it
15:48 sometimes the decals might change to indicate that the ownership has changed
15:53 to a different superpower maybe the available ships might change
15:56 for instance the Alliance will have Alliance ships and in depend
15:59 might not or yeah things like that but the idea is that that's a way of having
16:13 politics represented as entities within the game but that would be less engaging
16:20 as a galaxy if the factions were static and they just didn't really change over
16:24 time so this is where the faction simulation comes into its own so a
16:27 faction can have a state now a state could be one of different categories it
16:31 could be in economic States or like boom or bust or that could be famine or it
16:37 could be in lockdown factions should be able to move between star systems as
16:41 their fortunes wax and wane so there was the expansion and the retreat and
16:46 finally when a faction wishes to seek dominance over another either just
16:51 politically or maybe wrest control over the star pot that controls the star
16:55 system then conflict happen these could be peaceful like election or they could
16:59 be armed like war in civil war so prior to beyond to chapter 4 only a single
17:05 state could be active for a given faction no matter how many star systems
17:09 that faction is present in but obviously if a faction is present in 10 or 20
17:13 different star systems then that's an awful lot of different player inputs
17:16 that we need to try and represent so we had a Q and the Q had priorities so for
17:21 example in a bust might be stuck behind a boom but if a movement States was
17:27 queued up if it became eligible for an expansion then that would take priority
17:30 over a boom and yet if a violent conflict came along then no matter how
17:34 much the faction might be in an economic boom if they're fighting with guns and
17:39 spaceships then that takes priority over all of the other ones that was how it
17:43 was before chapter 4 then we changed it up then we changed it up a bit so also
17:53 as part of the design mechanisms we we designed it around the majority of
17:57 player activity which still today is in the smaller ships and for smaller
18:01 factions we also received the feedback that conflicts could be difficult to win
18:08 in some situations conflicts are triggered by the influence in a star
18:12 system and the influencer mechanic was changed
18:15 to win or lose a war overtime or an election overtime but we received the
18:22 feedback that was not always the most predictable gameplay activity for people
18:28 who were trying as part of their primary gameplay activity to influence a war we
18:33 received a feedback that it wasn't as predictable as perhaps it could be so we
18:37 wanted to try and address that we also heard the feedback that factions were
18:43 not as straightforward to control when they're in a larger number of star
18:47 systems originally back when the game released a faction might be in two maybe
18:51 five maybe ten different star systems and still today the vast vast majority
18:55 factions are still only present in their home star system but over time through
19:01 the actions of players because nothing else drives this over four thousand
19:06 factions are now present in ten or more star systems and there are a handful of
19:10 factions between more than 60 different star systems and we took the view that
19:14 this was perhaps not scaling as elegant as elegantly as players would have come
19:19 to expect and likewise it was time for a change up in the design and chapter 4
19:24 gave us an opportunity to do that so we took the opportunity to retune the
19:32 simulation so we were looking at the balances of different activities and
19:38 different states we took the view that if a faction is present in 40 50 60 star
19:48 systems than having a single state that represents the fortunes of that many
19:53 people was perhaps losing some semantics it was maybe not the most realistic the
19:58 most dynamic experience that people would expect from a game that is
20:02 affecting that many different star systems because this is still a game it
20:06 must be a realistic simulation but it also must be enjoyable and be
20:09 predictable so we wanted to redesign the simulation so that a faction could be in
20:15 different states in different star systems which gave us a great deal of
20:20 opportunity to redesign a number of things we also wanted to make it more
20:25 explicit which states works cluesive of each other and which ones
20:28 could safely coexist previously we had aq estates and these were these were
20:33 hidden this was a server function it wasn't really displayed in the game
20:36 client anywhere so we wanted to make it a little bit more directed having say
20:42 civil unrest and boom along the same axis and they they were both broadly
20:47 economic but we felt it would be more clear if we were to split out economic
20:52 state some security states so we did so at Chapter four
20:57 economic states could be in famine boom bust or investment and we have a concept
21:04 of an economic points system which moves up or down along the axis so we can now
21:08 represent which activities move the pointer and what states are triggered by
21:13 that pointer we added the positive security state of civil liberty because
21:19 that we only ever had negative security states before and we felt that if a a
21:23 star system was feeling well secure if your pardon the turn of phrase then we
21:28 needed a way to to represent that in the game movement states were largely
21:33 untouched with the retreat and expansion previously a failed expansion would have
21:38 directly triggered investment but we've now moved investments so it's an
21:41 explicit really really strong economic state and we're also able to add some
21:47 rare estates some ones that occur a little bit less often just to add a
21:51 little bit of interest to the game add a little bit more contrast to the
21:55 geography so the outbreak state was augmented by the poet attack state which
22:00 could have their own effects and their own triggers and their own consequences
22:04 within the game so in addition a chapter four we added a number of new activity
22:11 types number of core gameplay loop so that players can engage in on a
22:14 minute-to-minute basis so I've got an illustration here of a mega ship that
22:19 has been damaged by thyroid attack and it is asking for players to come and
22:23 help defend it and also to repair it this is just one example of some of the
22:27 new activity types that we added at chapter 4 there's a very large number of
22:30 installations that the design team and my team helped add to the game conflict
22:35 zones we've redesigned mega ships now had object is depending on their
22:40 situation if they been broken down although under attack
22:42 by pirates or perhaps you want to raid it and steal the cargo that Sears cargo
22:46 bays all of these new activity types needed to be included into the
22:50 simulations design with these new activity types we were able to add these
22:57 to the user interface so I've got an example here which is shown just xeno
23:01 hunting and exploration activity types but now that we have that explicitly
23:05 announced to the players as part of the targets in the navigation panel we were
23:10 able to illustrate what actions were possible in a place and therefore what
23:16 consequences they might expect from the simulation from engaging within that we
23:22 also wanted to bring greater balance to player actions for a given day between
23:28 the different activity types we already had the previous activity types we've
23:32 added a large number more so given that we were rebalancing them we might need
23:36 to take a more holistic view than simply adding the new activity types and
23:39 getting on with it and we also looked at wanting some more design inputs and I'll
23:45 definitely come on to how we've sorted out the design inputs a little bit later
23:48 on however one of the major pieces of feedback that we heard and we believed
23:55 ourselves was that the the function simulation is can be a little bit opaque
24:00 even today it is perhaps not as visual and visible as it could be the problem
24:04 is that it is a very large amount of day to --'tis necessarily complicated and
24:09 affects a large number of player actions as soon as we display a state in the
24:16 game it is interesting the outer date as another player has contributed towards
24:19 it but with the new economic point security points and the political
24:24 influence system we were able to add States directly to the user interface so
24:30 the right-hand cockpit UI panel is now much more descriptive about what the
24:34 current state is maybe what spending which direction it's going in and we're
24:38 able to include some descriptive text for what it might mean what some of the
24:43 consequences might be we were also able to ride on the tails of the squadron's
24:49 feature squadron's was an amazing new social feature that we were really
24:52 excited to bring to the game but it also gave us an opportunity to
24:56 change the dimensions to how we were visualizing this previously up until now
25:02 the biggest place to get information about the factions was in the local news
25:06 and our port services and we had some some digesting text where you had like a
25:12 newspaper article saying what that a faction is expanding or retreating or
25:17 the war didn't go so well for them but if we're adding multiple states and
25:21 multiple systems that very quickly fell down especially given that the user
25:25 interface panel for that is about yay big so we were able to bring together a
25:29 user interface pane which brings out the detail for every single star system that
25:36 faction is in the limiting factor here is that we can only display one faction
25:39 at once which is why the squadron feature letters engage with that by
25:43 tagging a squadron - a faction which was then the selection for which faction of
25:48 the dozens of thousands we displayed so these were ways that we could bring out
25:53 the state I mean we've just made the state a lot more complicated because
25:56 you've got different states in different star systems we needed a way to to bring
26:00 this bring us out from the game otherwise it would be less meaningful
26:03 for players we also took the opportunity to use the objective system that we
26:12 brought a chapter four to rebuild how conflict zones work so previously
26:17 conflict zones were relatively Undine ammok affairs where we spawned two
26:23 populations of npcs and two factions they were hostile to each other players
26:27 joined in and they received combat bonds for participating with the objective
26:33 based system were able to bring something a little bit more guided in
26:36 particular we're able to bring some optional objectives and it also meant
26:40 that a given instance of a conflict zone could suddenly have an outcome as
26:44 opposed to just being an ongoing representation of factions in armed
26:48 conflict a battle could be won or lost but the war might still be ongoing so
26:53 this was one of the biggest things that we were able to bring to this so we have
26:57 objectives and we have optional objectives
27:01 in addition to this we were able to spend some time developing the
27:04 technology for how we spawn how we place scenarios within space and we were able
27:10 to bring a political element to where we place to conflict zones within the game
27:14 that previously was mostly Geographic in nature so for instance if a factions
27:20 asset was at risk were able to spawn conflict zones closer to that asset as
27:25 opposed to being around an asteroid base maybe three or four planets away we
27:30 wanted conflicts to have fixed durations so that it was a little bit clearer for
27:34 what to expect and when but also so that players who could maybe only play
27:39 certain times of the week still had an opportunity to contribute towards their
27:42 their faction and as I mentioned previously faction conflicts used to be
27:48 based on the influence mechanic and whilst today they are still triggered by
27:51 influence we have now given them their own scoring mechanism so at chapter 4 we
27:57 now have the concept of a day so if a conflict lasts for five days at the end
28:04 of any given day using the same tick that we used to advance the rest of the
28:07 simulation a we tally up who who got how many plus points who else got how many
28:12 plus points on we declare a winner so it was a race to the top and then whoever
28:17 won that day got a point in the one day's whoever won the basin whoever won
28:21 the most number of days that's easy for me to say won the conflict in addition
28:27 there was some uncertainty from plays this was one of the things that we we
28:31 looked at quite a bit was plays who won a conflict were often surprised by the
28:36 outcome they might have been expecting to get a large coriolis star port but
28:41 instead their faction took control over a time doctor or planet installation and
28:45 this was maybe surprising to them so this we were able to declare at the
28:49 start of the conflict what they were fighting over we've been that we could
28:52 bring it out in the user interface and the news reports as well this makes it a
28:55 little bit more visible for what's going on the conflict scoring mechanism is
29:03 still fundamentally based on the same actions that it was previously
29:06 so elections are done combat actions only so anything that will boost the
29:11 economic status of a faction will go towards contributing towards it
29:15 school like wise if it's in a war or a civil war then only combat actions will
29:21 contribute towards it and yes the the new conflict zone objectives are one of
29:26 the stronger ways of doing it combat bonds are not quite as strong but
29:31 they contribute towards it to a certain extent but also the mission system do
29:35 violent crimes do assassinating other plight other NPCs but selling
29:40 exploration data doesn't help a faction if they're busy fighting for their very
29:43 presence in that system which kind of makes sense to me yes from a glance and
29:49 if anybody is just joining us and midway through and dava's for reading off all
29:54 of this interesting information of how factions and the background simulation
29:58 this stream in particular is a discovery scanner where we take an opportunity to
30:02 look behind the curtain sort of what goes into developing and Lee dangerous
30:06 and to night we're discussing factions on the background simulation you may may
30:10 as you may have guessed absolutely so this is where the technical part begins
30:17 we've reviewed some of the design concepts this should hopefully give you
30:21 a indication of the objectives that we are trying to achieve now I'm going to
30:26 talk a little bit about some of the implementation details this is where
30:29 some of the database work comes in the may even be sand while certain wild
30:32 statistics ahead so please do be warned so the illustration that we have on the
30:39 screen at the moment is the style of the database structure that we used prior to
30:44 chapter 4 so for a given faction which was cleared down during the daily tick
30:51 excuse me we stored the sum total of player actions but they were converted
31:00 in units to do with the state that that action effective so this was very much a
31:04 real-time simulation in terms of monitoring players activity player
31:08 transactions player actions so at the time the action happened we would
31:13 categorize it we would transform it from saying this is a murder this is a
31:17 mission completing the system acceleration data and then we'll take a
31:21 look at what states were active at that time and then we would convert that into
31:25 this might increase our influence or this might contribute to
31:28 woods towards lockdown and they moved add that to these buckets let's call
31:33 them inside the database and these were tallied up over the day and then at the
31:37 end of the day or the start of the next day every 24 hours or so for those who
31:41 like to track the tick these were tallies and then we would look at what
31:45 the overall efforts were for the last 24 hours and that would form the new
31:49 situation now that wasn't going to accurately record the player activities
31:57 now that we've suddenly added an certain amount of complexity we have multiple
32:02 states where previously those just 1 states could be different between
32:05 different star systems where before they were not and in particular we wanted to
32:10 look at how we can better balance different activities between each other
32:17 from a design point of view we wanted to take more of a aggregated approach over
32:23 the day for how much players have affected a certain faction as opposed to
32:28 find our best to do it in a real-time situation so this much more complicated
32:34 database structure and again this is this is only indicative this is this is
32:38 not exactly a screenshot from the live in the live long ghost cluster of every
32:45 faction for every day we now store a number of metrics we've mentioned
32:49 economic points security points and influence we wanted to be able to better
32:54 balance positive actions against negative actions as opposed to being a
32:58 tug of war that could very quickly become undone for example exploration
33:02 data is an extremely high discord number just because of the value of credits
33:06 involved whereas smuggling is often done in small ships which means you have a
33:11 number of relatively small transactions so we wanted a way to better balance
33:15 those we also needed to record a which star system that effect was what was
33:20 happening within and missions they can affect more than one star system for a
33:25 single mission because it can start and end in different places and we wanted to
33:29 be able to better balance different activity types against each other so for
33:33 example if we look at the example on-screen ignore the actual numbers but
33:38 the unit's where we put the decimal point is important so for example Mich
33:42 over a day might be a +16 expiration data over the day might be plus 300,000
33:47 if people have been doing some of the core mining that was included in chapter
33:51 4 that plus could be tens of millions those were obviously very different
33:55 orders of magnitude to try and balance against each other so we wanted to have
33:59 some more design controls about one of the other questions that frequently
34:06 comes up is exactly which action affects which which state so this lovely world
34:12 of text that I've just put in front of you and I'm sure is being fairly
34:15 screenshot it is a illustration of a chapter for which actions affect which
34:22 of those dimensions it doesn't necessarily directly affect
34:25 the state but it tells you how we have intended for instance trade versus
34:31 smuggling to have an effect on the economic welfare of a faction don't
34:38 worry too much about the details here because the core data structures that I
34:41 paraphrase this one are several hundred lines long there's a lot of detail in
34:45 there there's a lot of design coefficients it's a lot of different
34:48 functions and algorithms involved but in general terms if you do a positive thing
34:52 to a faction positive things should happen to it if you do a negative thing
34:56 to a faction bad things will happen to it
34:58 or perhaps you're reining in some of the positive actions that are happening to
35:02 it it's difficult to summarize some of the details because again they are in
35:08 very different orders of magnitude but for example a typical scenario worth of
35:15 combat bonds might be worth say twenty or thirty percent of a medium conflict
35:20 zone but a high-intensity conflict zone might be worth more than that
35:23 simply because eight is either harder to achieve or it takes longer to achieve or
35:27 it takes the same amount of time but a player must have spent more time
35:32 progressing their ships power and therefore we need to recognize that
35:35 progression as well so that whilst the galaxy simulation must and does respond
35:41 to every new player in their sidewinder we needed a way of quantifying people's
35:46 progressions if they're in say an engineered fer-de-lance which takes a
35:49 much greater investment in terms of time and progression and player skill as well
35:55 so missions and scenarios are just too varied to include on the table if I did
36:00 it would still be scrolling for whilst I've been talking about it there's just
36:03 too many variations but needs us to stay this is roughly how we deal with this
36:10 however this is just which inputs we would like to hook up to which outputs
36:15 how do we convert between those activity units and those state units because if
36:20 you remember at chapter 4 we are now storing numbers in the database in units
36:25 of the activity type from the player not the faction state that it's affecting
36:29 and this is where the the maths begins so one of the things that we wanted to
36:35 try and do to make sure that we didn't introduce unexpected change during the
36:38 design process for chapter 4 was to base our design exercise on what the inputs
36:46 and outputs were achieving today the inner workings were changing
36:50 significantly you saw the different design dimensions in the database but we
36:54 needed to make sure that the net change was as small as we could possibly manage
36:58 it that way we could trust that the implementation of the new states and the
37:02 new dimensions was broadly correct from a strictly science point of view
37:10 the type of curve that I've Illustrated there it is a straightforward
37:14 logarithmic progression and it illustrates along two axes we have a
37:17 fictional state axis and a fictional activity axis and it shows you one way
37:22 of illustrating how we can convert from one into another
37:26 now many activity types are by their very nature logarithmic in their
37:31 progression as a player power expands or as many players over the course of a day
37:36 contribute towards one action for example if somebody has just done ten
37:40 career missions or somebody has found a lucrative trade route and they are
37:44 quickly running backwards and forwards until somebody else finds it or the
37:47 supply and demand is exhausted then each of those is a relatively linear
37:52 progression whereas if you're trading in say a sidewinder you might have five or
37:58 ten units of cargo if you're doing it in a type six you might have say 80 or 90
38:02 if you're trading in a type nine you might be up as many as six or 700 units
38:07 of cargo each of those is a logarithmic rushon which means that when you take
38:11 that into consideration our inputs against outputs our activities versus
38:16 States is actually a relatively linear progression and that way it leads to
38:22 relatively easy to understand design constraints because we need to be
38:26 confident that what we want to happen is relatively close to what actually
38:30 happens there is some uncertainty because no matter how hard we design the
38:35 game how rigorously we try and train it and validate the model is based on
38:39 player activity there will always be some player actions that surprises that
38:43 are creative or that touch the simulation in ways that it doesn't
38:48 behave particularly elegantly and we'll come on to that in a little bit later on
38:53 but in order to redesign this we looked at the live game because one of the
39:01 luxuries that we have from a design point of view is we can take a
39:04 anonymized section of time from the live game service transaction logs and we can
39:09 take say a four week chunk we can anonymize it to make sure that we don't
39:14 risk any person other than to follow them personally identifiable information
39:17 leaking we can then aggregate it per faction per system per day which are the
39:24 new units of measurement and then we can quantify them so for example the
39:28 histogram that's being shown here is a opinion in activity units so this is the
39:35 outputs of the simulation of the influence gained from trade cell so
39:41 along the y-axis that is the count for that population sample so we can see
39:47 that the majority of days most factions had relatively little influence gained
39:51 from trade cell but some days there was a very very very large amount of
39:55 influence gain from trade cell and again it doesn't look particularly interesting
39:59 it's just a minute return but if we go and apply logarithmic function we
40:03 suddenly have these very clear quantifies of sections within it so this
40:10 gives us a concrete mathematical goal to aim for when we are measuring the
40:15 results of our newly designed simulation so this is what the results want to be
40:21 but one of the things that we need to model
40:23 is what all the different possible inputs might be and again we can use the
40:27 same data set to do that so what we have here is a histogram which I've overlaid
40:33 with an approximation of the distribution of that thanks octave and
40:37 its historic function so these are populations again summed perfection per
40:45 star system per day and again I've used to trade cell there's an example here so
40:51 we can see that again by and large most transactions are round about say 0 to 10
40:57 to the 3 10 to the 4 and this is because the vast majority of trading is either
41:02 small ships or which is people selling salvaged or just relatively unconcerned
41:07 efforts as opposed to getting a cutter filling it with cargo bays and shuttle
41:13 running that's relatively uncommon but it does happen and we must recognize
41:16 that and this also gives us a indication of what the majority of typical faction
41:25 inputs are likely to be perfection per system per day and to that goal I've
41:33 used 2 Sigma which is a way of quantifying this is the majority of
41:38 typical of likely inputs which we need to put as being where the most of the
41:45 rough of variety in outputs comes in we were still recognized very small and
41:49 very large outputs but the majority of them fit within this this bell curve and
41:53 this is where we were designing for it in statistical terms you might say this
41:57 is overfitting but because we had very specific design objectives in mind this
42:02 was actually a desirable outcome perhaps we still have to make sure that the
42:08 design allows for every single possible players activity so this is one of the
42:14 examples of the outputs so obviously it doesn't make a huge amount of sense and
42:20 you'll forgive me for not running this through a logarithmic x-axis because
42:24 it's possible to make an economic loss when you're doing commodity trading I'm
42:27 afraid I couldn't put a log scale on it because you can't do a log of negative
42:32 numbers without getting into imaginary numbers and no but
42:35 once imaginary numbers in the galaxy it's complicated enough thank you very
42:38 much however what it does illustrate is the majority of the active area so we
42:45 can see that the response between say in this case minus 25 and plus 25 has been
42:50 fitted to the majority of those player inputs and we then went through this
42:56 process and validated it for each of the different activity types that we have
43:00 within the game and then this was where we got to four chapters hall then it
43:06 released in early December and then very quickly we were monitoring the live game
43:12 we were taking the live transaction logs again through the same anonymizing and
43:19 aggregation process and we then checked the results was the live simulation
43:24 behaving in the way that we designed it to and some things were perhaps not
43:28 quite as respectful of what we were trying to achieve as we're aiming for so
43:33 very very quickly I think it within the first week expiration data and missions
43:38 were both very quickly tweaked they had different orders of magnitude out in
43:43 addition the new activity types because we didn't have a realistic model of
43:49 typical player inputs we weren't sure exactly how players were going to engage
43:52 with the new activities we just did the best we could and then we watched it so
43:58 it was very quick quickly apparent that the rate of change of influence for that
44:04 first week was too high factions were gaining and losing 10 15
44:08 maybe even 20 percent of influence a day that that rate of change was just too
44:12 rapid for players to have a reasonable control over it so we reigned in some of
44:17 the day the daily caps this calmed things down and gave us the opportunity
44:21 to reconsider things from a design point of view and this is where we started in
44:29 the new year this was back in January 2019 so as opposed to using a
44:34 logarithmic progression like we saw before we decided to evaluate a
44:39 different style of equation so what we have here is an extremely rudimentary
44:43 one this was drawn with a ruler but it gave us a very quick way of iterating on
44:48 some design ideas for how we could transpose activity units
44:52 into state units for the mission system and I've already mentioned the mission
44:56 system is extremely varied therefore the responses from it can be extremely
45:00 varied so what we have here is an illustration of the type of curves that
45:07 we're looking at and it gives us an idea about the design constraints that we
45:10 were trying to design controls that we're trying to express so we still have
45:15 the concept of an early ramp up respecting every single action we still
45:19 have the concept of the diminishing returns maintaining a reasonable
45:24 response for potentially very very very large inputs but the majority and I do
45:28 mean the vast majority is in this active phase where we have the most contrast
45:32 between different different typical inputs also very quickly we received
45:39 feedback from players that they were finding difficult to move influence for
45:44 factions now that factions could have multiple states and multiple systems
45:51 previously where there was a single conflict allowed active and other
45:54 systems that were not caught by the conflict conflicts were now blocking
45:59 movement in all systems and this was proving too cumbersome for players to
46:03 try and effect and direct especially when factions are in 20 30 40 50 60 star
46:09 systems so we took the opportunity to slightly tweak how we allocate influence
46:15 and also how we were rebalancing influence up to 100 percent because
46:19 whilst most of the dimensions and most of the effects are independent one
46:25 factions economic points is unrelated to another factions economic points the
46:29 influence within a system is always close to some it must add up to 100% we
46:34 also modified how we were scoring conflicts a little bit and this was some
46:40 of the major design changes that we iterated upon after receiving player
46:44 feedback and monitoring how the galaxies really responded so we discovered that
46:51 conflicts well not discovered we decided that conflicts that did not have
46:55 something at state at stake we're no longer as interesting as ones that had
46:59 so in order to let factions move up beyond
47:03 each other because some systems might have six or maybe even seven factions in
47:07 them that we would only start a conflict if one or the other or both of the
47:11 factions had a asset at risk we also had the concept of bonus days at the release
47:18 of chapter 4 so that the the design idea behind that is if a faction was losing
47:23 four days down to nil that through heroic effort on those remaining three
47:27 days they might still force a draw the feedback that we received in the plays
47:32 was that this was too opaque maybe a little bit too surprising so instead we
47:37 removed the function of bonus days but instead if it was no longer possible for
47:42 a faction to win that war then we would simply end them end the war earlier
47:47 which means that Wars can suddenly end after four days rather than going to
47:50 going the full duration and this may seem like a small thing but when
47:55 factions are in conflict because if you imagine we've got two faction influences
47:59 moving around in a system a conflict is triggered when they get close to each
48:04 other and influence and then they lock together for the duration of the
48:06 conflict when those are locked together their influence cannot change which
48:11 means that the other systems in the factions in the system have less
48:14 influence to move within because these two must remain locked together we then
48:19 unlocked that for the day when they're recovering from the conflict which just
48:22 when taken in aggregation over the game actually freed up the the influence
48:27 simulation by quite a lot we also received good feedback from our missions
48:33 change for how we were equating the activities into the States so we took
48:40 the design decision that all of the activity types would really benefit from
48:44 this style of design control as opposed to the previous shape of equations we
48:49 also had a number of times when the galaxy did perhaps not elegantly respond
48:54 to certain situations such as when the distant wails - expedition visited
48:59 explorers anchorage for the mining community goal the the player activities
49:03 for there were I would say unprecedented with some with some confidence the
49:09 amount of exploration data that were sold to explorers Anchorage was I would
49:14 say very very high but in addition the there were a number of other players who
49:19 were supporting the other factions in that system with a very large number of
49:22 credit donation missions and again the number of those being completed was of
49:26 an unprecedented level and the galaxy just didn't give us a elegant response
49:32 for how they were coping it tried his best but it wasn't giving us the results
49:37 that we were looking for so we redesigned all of our activities and
49:45 this was a complete redesign from those same mathematical models for how all of
49:50 the activities were balanced from activities into States so what I've
49:54 Illustrated now is a logistic curve this is an s-curve this is a smooth response
50:01 for how inputs and outputs are equated together you can see it's broadly
50:07 similar to that approximation that I drew with a ruler a few slides ago but
50:11 it gave us a more continuous response it has infinite bounds I mean you can go
50:16 below zero if you really want it to what I wouldn't recommend it which is one of
50:20 the reasons why we have tallied up the negative inputs separately from the
50:26 positive input it means that we can use the same design functions and then
50:29 balance them against each other during the tick as opposed to it being a real
50:33 time function this gave is much greater design control it also meant that we
50:37 doubled the number of activities that we needed to have equations for because
50:39 some went up as well as some went down so we used the design tool and we placed
50:48 the inputs if you can remember that histogram I drew earlier on such that
50:53 the again that same two signal worth of inputs were now placed into the majority
50:58 of the contest from the outputs so this was the the modeling part of the
51:02 exercise so we we trained some models we looked at the design coefficients and
51:06 then we we monitored it and we found it gave us a good result so as part of some
51:10 of the server-side updates that we made without interrupting the service to the
51:14 galaxy this was played out in the early months of 2019 and now this is how the
51:20 majority of activity types of scored today there are a few that are not using
51:25 this conflict scores is another good example of something that's a more
51:28 real-time function but anything that is tallied
51:31 today he's now using an equation such as this and we found that this gave us very
51:36 elegant responses even for unreasonably strong or unreasonably small inputs and
51:42 that is the secrets of how we redesign is the faction simulation for beyond
51:49 chapter four dad thank you so much I won't claim to be an expert of the vgs
51:55 and when we started to talk about graphs I was like yes I'm following I think no
52:01 wait no I'm but I know there are people out there who are really interested in
52:05 this stuff I'm there are a lot of BGSU experts here who are furiously probably
52:10 rewinding or waiting to rewind video will be available to go and pour over
52:14 these details but yes you have a DAB here so it's it seems fitting you know
52:18 that we have a little bit of the Q indeed you collected some of the
52:21 freshmen I did I'm afraid I wasn't able to keep track of questions that were
52:25 raised during that during the stream on the chat channel I'm afraid I couldn't I
52:29 couldn't read that quickly yeah but you don't was presenting to camera but I
52:33 have noted a few questions already that I can maybe address from the forum
52:36 threads and if somebody can shout out some other ones from the chat then we
52:39 can maybe look at this so one of the very first questions that
52:43 we received a lot is how are factions that were placed in the game from the
52:48 request of player groups different to the NPC factions that we designed and
52:52 the short answer is they're absolutely identical in every possible regard
52:56 except maybe we answer more questions on the player factions and we do on the NPC
52:59 ones there is absolutely no other regard the game doesn't even know there is a
53:03 difference they are just the same let's see what else did we have trends for
53:11 many people asked about why say negative states were difficult to achieve
53:17 why retreat was maybe difficult to sustain how faction system happiness
53:23 could maybe engage with movement states such as expansion I want to touch
53:26 another little bit because this was another opportunity that squadrons
53:29 brought to us we needed a way of supporting how successful a faction was
53:35 being rather than just its size because it is arguably more difficult to control
53:40 a larger faction but if you have a larger group of players then
53:44 so you've got more options so this is where the happiness mechanic was
53:47 introduced in chapter 4 and at the moment it is being used to contribute
53:51 towards the political leader board for a given faction a happy faction is going
53:56 to give you a better position on the leader board than an unhappy faction you
54:00 can still do it with an unhappy faction it just takes more effort and it's
54:03 harder to achieve people also mentioned that factions with high influence can
54:11 sometimes get stuck in a chain expansion if a faction is at a high influence in a
54:15 system and it's it son contended then it can get stuck in that direction and
54:19 we've also heard that players would maybe prefer missions to have a greater
54:23 impact for a given investment in time and progression so some of those were
54:28 design constraints that we designed into the system in particular how actions
54:32 were balanced against each other but a lot of these things that I've just
54:36 reeled off and these were asked by large number of commanders so apologies for
54:39 not name checking each of you these are all things that we've talked about
54:41 internally we've heard the feedback and we would like to consider in a future
54:46 update however whilst we are actively talking about them at the moment we have
54:50 nothing to announce today about it but it is under active consideration if we
54:54 can maybe fit it into another another update in the future so on the forms
55:01 commander rubber nuke name-check mentioned that earlier on doing the
55:06 background sim he'd seen complicacy they had seen conflict zones pop up in
55:11 asteroid belts and they've not seen one in the life billed over again can
55:15 conflict zones appear anywhere so I mentioned briefly in passing how the
55:21 galaxy simulation in Chapter four now knew about the political landscape of a
55:25 star system as well as the geographic one I took a look at the design
55:30 preferences for where we're placing complex owns and as well as spawning
55:34 closer to faction assets each of the low medium and high conflict zones have
55:39 their own sets of primary and backup preferences for whether try and spawn
55:42 today they will try very very hard indeed to spawn in a ring system around
55:48 a star but it is possible although very very unlikely that they'll spawn around
55:53 the planet ring there are a number of reasons for that some technical you may
55:57 have seen some fame reference bugs shortly after Chapter
55:59 four that we've since fixed but we agreed that it would lead to some cool
56:05 flying experiences but for the moment it is unlikely that a conflict zone will
56:09 spawn around a planet ring we'll never say never but no changes to announce
56:13 today commander samurai 83 ITA asked when you leave a conflict zone how does
56:21 this count for the days war so at the moment it doesn't a conflict zone even
56:26 though it won't have any ships in it is there for everybody on all platforms for
56:32 that system for that day so the objectives count towards it so a win
56:35 will count a loss will count but if it's left to its own devices then only the
56:40 player activities will count towards it so if you've collected some combat bonds
56:43 whilst you in there but perhaps you had to leave before the objective was
56:46 completed then we'll still recognize those combat bonds when they're redeemed
56:49 but without the objective no score commander Missoni event apologies for
56:56 mangling your name there mentioned that static capital ship installations are
57:01 based upon the default owner of a system and they don't seem to turn transfer
57:05 ownership the answer to that is yes that is indeed the case capital ship
57:08 installations don't change ownership if a federal star system contains a capital
57:14 ship that is building Faragher to class capital ships it would make little sense
57:18 for an empire faction to control that that's not something that the simulation
57:22 is able to influence today we can't despawn the Farragut's and spawn some
57:25 interdict as for example that's that's not something that we have so yes for
57:29 the moment those don't those aren't considered assets from the point of view
57:32 of the faction simulation but all of the other installations that you'll see
57:37 around the galaxy are at least anything that has a as a jurisdiction around it
57:43 great thank you thank you deferens that we gathered from the forum's we will try
57:48 our best answer some right here but if there isn't anything we get to today and
57:52 yeah and of course we can't answer everything but we will do a follow-up
57:54 yeah I will try to follow up next week absolutely so first question we have
57:59 here from NIC bolts any hopes that population numbers in a star system will
58:03 vary so at the moment the answer is no it's it's obviously a cool idea it could
58:07 lead to some spectacular gameplay but it's not something that we are talking
58:11 about implementing the game today just because
58:16 that's just it has a great deal of design complexity you can see where the
58:21 design might lead but speculating on design might might have had ideas are
58:26 never the hard part trying to build them into a cohesive whole and then schedule
58:29 time to implement them that's that's what we're trying to work within yeah
58:33 there any more questions I know there's some some scrolled up we've chatted um
58:37 even quite quickly so if you have more questions do pop them in we'll try our
58:41 best to answer them yeah we're here for another five five ish minutes so are we
58:47 going to track and remove in active player miner factions the problem with
58:51 that is how do you quantify inactive could it be that the players just simply
58:56 haven't touched them for a time so one of the design constructs within the game
59:00 is that a faction can never die it is always present in its home system even
59:04 if it has no influence and that's the same for player sponsored factions as
59:07 well as the other factions whilst we have a very large number of
59:10 player miner factions in the game that they're still vastly dwarfed by the
59:14 number of NPC factions we are talking dozens of thousands of NPC factions and
59:21 yeah so there are a number of faction of questions about negative states yes we
59:26 hear the feedback that you would like negative states easier to influence we
59:29 hear that feedback it's something we're talking about I'm afraid there's nothing
59:32 to announce today about how we might achieve that what design we might change
59:35 to make that easier yeah it's it's not something you can just decide here in
59:40 hearing no I would like to but then I would be lying because production and
59:44 design would say no and that's it's highly reasonable because I'm just an
59:48 implementer and my team are just an implementer yeah anything else here grab
59:54 some so yeah expansion tax yes we hear the feedback it's if something we've
59:59 heard in a number of occasions it's something we're actively considering
60:02 somebody mentioned briefly the I how thyroid conflict zones was born yes
60:08 that's part of some of the thyroid simulations that happened within chapter
60:11 four I'd love to go into the detail now but then would have to have another
60:13 hours livestream and I don't think anybody's prepared for that this evening
60:17 have to try and get you back for another for that yeah absolutely
60:20 likewise people asking about future features such as free carries we are
60:23 going to be talking about them in the future I'm sure
60:25 that time is not today not today it's not this kind of livestream we are gonna
60:29 we're gonna go all into the details dive rightly says but yes today we're talking
60:33 about factions yeah and and the background simulation and anything that
60:37 weaves things together indeed so yeah okay yeah okay so retreat
60:42 famine yes we hear the feedback nothing to announce today negative states
60:45 somebody mentioned briefly in passing if a criminal faction has control of a
60:50 black market does that aid or hinder them so the way that that's represented
60:56 in the galaxy simulation this is actually part of the economic simulation
60:59 not part of the factions political simulation so when a criminal faction
61:05 takes over control of a market which goods or which commodities they deem
61:10 illegal changes so whilst a social faction might deem trading in weapons to
61:17 be illegal and therefore smuggling him on the black market harms their economic
61:21 interests when a criminal faction takes over that is no longer illegal and
61:25 therefore you can no longer smuggle them on the back market instead of selling
61:29 them on the open market becomes a net positive impact that's how it's modeled
61:32 and the thing that you smuggled on the back market by by by its very nature is
61:38 something which they've deemed illegal and that's always there for a negative
61:41 effect I mean I'm absolutely over hassed for
61:48 this indeed so I would love to talk about missions but the mission system is
61:52 it's very dynamic it has at the last count 110 120 different mission
61:57 templates to choose between each template can have a number of different
62:00 variables that are generated based on the market in that system with those
62:05 factions at that time for nearby factions the number of variables are
62:09 high so the trying to quantify those on the live stream for you would maybe take
62:17 again all evening likewise people who are asking about general networking
62:21 questions so for instance the game has a debugging mechanic for bookmarking a
62:28 network crash we if somebody has taken the time to do that and has logged a
62:35 support ticket and that makes its way through to us as a bug report
62:39 that we have investigated looking at those Network bookmarks is incredibly
62:43 helpful because it gives us a snapshot of the peer-to-peer matchmaking network
62:49 connections for that group of players it is not something that we draw attention
62:53 to in and of itself because it only has to work within the context of a wider
62:57 bug report having it there is invaluable and we're grateful to those players who
63:01 use it but it in and of itself is not something that will attract our
63:04 attention to it because it's too it's too prevalent and also it doesn't give
63:09 us a context of whether it is a good thing a bad thing or somebody miss
63:12 clicked on the keyboard okay that I think that's really all we have time for
63:17 today as yeah one final question are bounties and bonds on the same s-curve
63:21 if it's affecting the security state yes it is if it's affecting a complex state
63:27 no it isn't I want to take it away from you over to you
63:35 that wasn't a veiled threat I mean it promise
63:39 thank you thank you Dad for joining for joining us tonight thank you to
63:43 everybody in the chat came along this is a this is a reprise of what I've went
63:48 into at Lake Como this year that's another question I wanted to answer you
63:52 can answer it I'm not going to deny that one final question I promise I promise
63:56 undone after this will do his best to give a closing report so one of the
64:00 Platts comments that's just scrolled past was somebody asking about markets
64:05 that are in different positions within a star system and this was something that
64:09 we added to the economic simulation in Chapter four so where previously the
64:14 prices of a good is affected by the supply and demand what we were able to
64:19 do was reflect the distance you had to fly in SuperCrew from where you entered
64:24 the star system into that equation the logic being that any players either an
64:29 npc or a player who doesn't have to travel very far might not get such a
64:34 good response from the market as somebody who's just flown to Hutton
64:38 orbital and therefore Hutton orbital will blaze natures naturally have better
64:42 prices both because they have to pay a premium because they're more
64:45 inaccessible and also because fewer players make that journey and therefore
64:49 there is a greater amount of supply or a greater amount of demand
64:52 and therefore the price isn't naturally higher this was a change it's not
64:55 political it's more economic in nature but it was something that we changed in
64:59 Chapter four okay probably some done mail and oh and we're
65:02 gonna have to close up because otherwise Steven over there might yank dad off
65:05 with one of those old-fashioned hooks yeah and nobody wants to see me exiting
65:08 stage left right I'm gonna bring it back round thank you thank you dad really
65:13 thank you for going into detail about that again it's it's it's a quite a long
65:19 monologue there I mean I'm not it as best I could that there you go and thank
65:24 you to everyone again who dropped by so that's what we have time for that has
65:27 been our sixth discovery scanner and I think and we'll be revisiting or
65:31 visiting more discovery scanners in the future dev will likely show his lovely
65:35 face on the stream again I have no doubt of that and as I said if there are any
65:40 questions that you have that aren't on the forums we are going to try and get
65:42 to as many as we can so yeah pop and pop on the chat and
65:46 thanks again that we're going to be signing out right now so see you next
65:50 time goodbye everybody goodbye
Last edited:
Play that hamster tune again!
O7,
O7,
Play that hamster tune again!
O7,
POWOWCHAKAWOW POWACHAKAWOWOW BUMCHAKAWAWO [10 minute drum solo] bewwwweeeaaaaahhh BOMWOWCHAKAWAOWOW
Quoting Dav, "combat bonds might be worth say twenty or thirty percent of a medium conflict zone but a high-intensity conflict zone might be worth more than that simply because it is either harder to achieve or it takes longer to achieve or it takes the same amount of time but a player must have spent more time progressing their ships power and therefore we need to recognize that progression as well so that whilst the galaxy simulation must and does respond to every new player in their sidewinder we needed a way of quantifying people's progressions if they're in say an engineered fer-de-lance which takes a much greater investment in terms of time and progression and player skill as well so missions and scenarios are just too varied to include on the table..."
I just want to make sure I'm reading this right. The BGS favors harder missions and combat zones, but please tell me it doesn't favor ship builds. In other words, if I run a courier mission in my Sidewinder, I'm making as much a difference as if I ran that same mission in an engineered Corvette, right? Otherwise that would seem pretty screwed up. I purposefully like using some of the "early game" ships because I find them fun, and I'd hate to think I was being penalized for that.
I just want to make sure I'm reading this right. The BGS favors harder missions and combat zones, but please tell me it doesn't favor ship builds. In other words, if I run a courier mission in my Sidewinder, I'm making as much a difference as if I ran that same mission in an engineered Corvette, right? Otherwise that would seem pretty screwed up. I purposefully like using some of the "early game" ships because I find them fun, and I'd hate to think I was being penalized for that.
I'm pretty sure that it means a High CZ is likely to be tricky to do in the Sidewinder, so it gives more effect because you generally need a better ship for it - not that the ship you use directly makes a difference, or that you'd get penalised (beyond it taking longer) for clearing the High CZ in a Sidewinder.
I'd read that as.... high-intensity CZs are worth more 'points' towards winning the day because to succeed you need to either; a) fight harder (be more skilful?), b) fight longer or c) be in a bigger better ship.
And this is why i make a huge fuss about being unable to submit bonds if you become hostile to the station owner during a conflict.
Frankly, submitting bonds should not count to war at all. If FD want a "consolation prize" from bailing on a CZ before it finished, it should be the actual kills, not whether you get paid or not.
He didn't cover mine, but then he didn't really answer any more detailed questions. I liked the stream, because well, it is Dav.
But it was more an introduction to 3.3 BGS rather than new revelations about mechanics. Those most into the BGS didn't really learn much they didn't already know. The curve bit was useful, though.