Galaxy Exile System

[Appocomaster]

typically, this should be with the whole galaxy / universe setup, but I know that several people have objected to the galaxy exile system round on round, and so am creating a separate thread for suggestions and discussion .

We're assuming that the galaxy setup is similar to now, with buddy packs and randoms in each galaxy, and c200.

(bare in mind that buddy packs could have self + galaxy exile turned off or on).

I know some people have recently come to me suggesting exile be based on "activity". The problem lies in how easy it is to fake activity, and how this can be abused to get a better exile than you should.

Firstly, how a new suggested system would work:

Take a galaxy slightly higher than average (probably about 1/3 of the way through the non c200 galaxies, around 66th or so).

Any exile moves (generally) towards that galaxy rank with each self exile, depending on the difference in "activity" to the current galaxy.

The general idea is that if you're really active and in a galaxy ranked 200, then you'll shoot up to a galaxy around ~120th or so, and then maybe 85th, 70th, and so on.

If you're not active and in a top galaxy and get exiled, then you'll land say 35th and then 45th and so on.

You have to be 2x more active than your galaxy (i.e. activity rating is twice your galaxy average) to move "up" in rank of galaxys, and 2x less active to move down in rank of galaxys. (ish).

However, this can be abused. To stop this being abused, it'll get more complicated. Basically, one of the nearest galaxies to the "target galaxy" with less than the universe average number of planets / galaxy should be selected (out of 3 or 4 at random)


Now, how is activity measured?
The best formula I can come up with is:

(XP + login_ticks/72 * value/100)/ (ticks_planet_existed_for)^0.5

where "login_ticks" is the number of separate ticks the planet has been logged in for in the last 72 ticks (this does have problems as you don't login every tick, so perhaps it should be the number of ticks that you've loaded a page while logged in, in the last 72 ticks).

we have the planet existed for modifier so that newer planets have a chance of getting into slightly better galaxies.
XP is probably, especially later on, one of the better indicators (for all races but Zik) of activity. I've included value so that Ziks aren't hurt too much, and I used the planet dump to work out that (roughly speaking) value = 100*xp by this time in the round. As value is only of any use when you login, and number of ticks logged in is important, I modified value with the login frequency.

This is obviously slightly more complicated than previously, and the formula for selecting a galaxy isn't worked out yet, but while not being too abusable it does mean that activity is rewarded. It does admittedly disadvantage small players, but the activity formula does mean that they'll probably get quite a high activity rating (I'm not sure that the exponent of the number of ticks the planet has existed for is quite correct, but it's probably about right).

Edit:
I forgot some things!!
Firstly, galaxy exiles weren't mentioned. They'd be placed somewhere about 0.7-0.8x their activity.
Secondly, self exile costs. Due to the way you leap through the universe towards activity, the costs should probably rise in terms of number_of_exiles^2 - say 0.1*value*(number_of_previous_exiles)^2 of each resource.


Do notice that if everyone thinks "wow I can get into an active galaxy" and start exiling, the not being able to land in a galaxy above the universe average rule effectively means that you just build an active layer of galaxies centered approximately on the galaxy 1/3 of the way from the top of the universe, meaning elite galaxies at the top, a big bulk of OKish active galaxies in the middle, and a number of small inactive galaxies at the end.

I know that this doesn't deal with inactive galaxies (indeed, galaxies whos average planet score is less than the average planet score of a planet in c200 should be disbanded and the planets moved to c200 automatically), but it does minimise the losses of active planets in the c200 reaches.

[mist]

(XP + login_ticks/72 * value/100)/ (ticks_planet_existed_for)^0.5

should that be:

(XP + login_ticks/72 + value/100)/ (ticks_planet_existed_for)^0.5 instead?

as modifying the value by activity, but not the XP seems rahter strange as you've mentioned a ratio between the two.

also, sandmans claims that the 1000th ranked planet has 14K experience. even if a planet has done something every tick for 3 days they're going to get 1 point for their logins. against 14000 that seems pathetically small, tho i guess that's assuming i'm right about the typo.

i'm not entirely sure about your thoughts on the cost of self exile either. if the tendency is to move towards a galaxy of your own activity level then it seems simply to be there to penalise those who were unlucky and ended up in a galaxy which means they have to make more hops - as the idea of this system seems to be to avoid penalising people for their ending up in inactive gals i'm rather unsure as to the point of this.

on a different note - after all that's been said about getting new players in to gals with active people etc etc, this seems a bit counter to what you're trying to achieve. i think it's reasonable to assume that someone who's new to the game won't dedicate that much time too it untill they've 'gotten the bug', by which time they'll have been exciled down to a galaxy where noone else is active and then they'll rapidly loose interest. i'd have gone with an approach whereby the more active galaxies get the inactive exciles so that things even out - this would then mean they had an incentive to get them going. this approach would require a change to the gal setup at the start of the round tho, which i'm assuming you don't want to do.

oh, and would your formula include stored resources in its calculations? 'cus if not it might be an idea to mod it so it does. i'd imagine they contribute quite a lot of value to inactive planets, and are more a sign of inactivity than activity.

[[hydra]rain]

mist, you shouldn't have changed ur first post
now you sound exactly like him

[Appocomaster]

No, the formula was right. You're welcome to suggest a different formula

I didn't want to modify both formulae. It is a bit flawed, but generally XP is a sign of past activity, whereas value is your network and not so important unless you're active enough to use it. I wasn't thinking too much about zik planets, so maybe it should read login_ticks/72 * value/50 .

This does discourage new players that start the round in a bad galaxy, as they won't be able to do so well and won't know how to get high enough XP / value early on, as well as not logging in enough to do great with exiling.

However, later in the round the scale factor should mean that new planets are exiled into better galaxies.

There is then the issue about spy planets being signed up, but as long as free accounts can't self exile still (?) this won't be such a huge issue.


I know that people have had issues with the exile system and want it to be more "activity related", and this is the best way I can think of it being activity related while still benefiting new players signing up mid round. It does mean that players signing up at the beginning of a round aren't so advantaged, but you can't really cover all bases

[mist]

why not have a different formula for xan, if the same metrics don't work for them? i'd imagine you're more interested in long term activity than short term anyway, so wouldn't it make sense just to have a correction constant that equals things up for ziks instead of trying to bodge them in to a single formula?

also, has anyone done anything about exponential growth yet? if not you might want to think about the effects of value/score increasing by x^(time played) and your activity formula including (time played)^-0.5

Originally Posted by rain

mist, you shouldn't have changed ur first post
now you sound exactly like him

maths geeks, sorry :/

[Banned]

I I have one question that doesn't seem to be addressed anywhere. Will number of planets in a galaxy still count for eligibility to receive an exile?

[mist]

Basically, one of the nearest galaxies to the "target galaxy" with less than the universe average number of planets / galaxy should be selected (out of 3 or 4 at random)

i assume so, tho i do get the feeling that this system would lead to a top and bottom end with smaller glaaxies.

[mist]

another thought that occours is what happens to buddy packs under this system?

say rod, jane and freddy (w00t!) form a buddy pack together, but end up in a fairly inactive gal. the only actives in their gal self excile up to a more active one, leaving them in deadsville. obviously, rod and co can't excile as their buddy pack'll be broken up, but as things stand they don't have a prayer of getting an active gal as all the actives are headed in the opposite direction, so any inactives they get rid of will just be replaced with more of the same.

[Kal]

As appcco knows i'm against an exile system that moves planets to the middle of the universe or to galaxies with similar activity.

The reason being that this then dooms the worse galaxies to complete failure. Now I am entirly aware that there will allways be worse galaxies, but my feeling is that if we prevent eciling on a large scale people will be more willing to try and make their galaxies work.

I propose the following:
disable self exile of buddy pack members
disable galaxy exile of buddy pack members.

This means each galaxy will have a core of 5ish people under the current system.

I think under that system the current exile system of putting planets in the galaxies with the fewest planets will be much more effective.

[Paisley]

It is a bit of a tricky one... One recommendation I would say is that 1 person cant be exiled into the same gal... I had to exile the same guy twice because somehow he managed to end up in my gal again and the reason we booted him was ingal spying for his ally (free jpg) and not deffing ingal (a 3 fleeter) So log ins wasnt the problem it was the fact he wasn't good gal m8 material.

[mist]

given how much people love that system, you must truely think that appoco's system sucks

[Kal]

Originally Posted by mist

given how much people love that system, you must truely think that appoco's system sucks

heh, well I admit I'd really like to get rid of exile entirly, unfortunatly it does have legitimate uses.

[mist]

why not run a system whereby the top x% of gals can only receive exciles below a certain activity?

they'll hate you for it, but it'll mean that exile gets used for what it was intended a lot more.

[Kal]

Originally Posted by mist

why not run a system whereby the top x% of gals can only receive exciles below a certain activity?

they'll hate you for it, but it'll mean that exile gets used for what it was intended a lot more.

why not design it so exiling re-distributes planets so they all have the same "activity" level then?

[mist]

Originally Posted by me in an earlier post on this thread

i'd have gone with an approach whereby the more active galaxies get the inactive exciles so that things even out

[mist]

i had a good rant about the ineptitude of pa typed out, however i'll be niceish.

one of you's talking about how to impliment an idea. the other's talking about whether the idea's really a good idea. one of you is wasting the time of anyone who reads this thread. either you're deciding what you want exile to do, or you're deciding how to do it. both at the same time is tricky. left hand, right hand. horse, cart. etc.

when i was asked about it, i was told you had a development methodology, this thread rather purveys the feeling that you couldn't find your arse with both hands, so you might want to review it a little.

a while back i posted a thread asking what your aims for pa:n were. can we take the lack of response to mean that you don't have any, or that you don't know what they are? 'cus this thread really feels that way.

seriously, i really do want you to succeed, but you don't fill me with confidence :/

[Kal]

sorry, I must have missed your other thread - i'll take a proper look at tomorrow night (bed time now).

I agree that our development hasn't been going quite how we wanted it to, the varying acitivty levels of people, particularly at this time of year is being problemamtic.

[mist]

how much of your contincency time have you used up, are there prospects for gaining more and what do you plan to do when you run out?

[Kal]

Originally Posted by mist

how much of your contincency time have you used up, are there prospects for gaining more and what do you plan to do when you run out?

I think its fair to say that unfortunatly we have used up all of our contingency time. For this reason we have reduced the scope of the changes in PAN - people who saw earlier versions of FAS will know that it had significantly more things in it much have now been postponed to PAN 2.0 - a good example of something that has been delayed is the pa bot (though in this case we also have yet to obtain approval from netgamers).

Before xmas I plan to redo our 1st half of 2006 schedule with the aim of rearranging some events to fit in more development and coding time without shofting the end of PAN 2.0 into the 2nd half of 2006 - this may proove impossible, in which case we do have contingencies to keep the community entertained before PAN 1.0 starts even if we have run out of time.

[mist]

i'd have thought that with it being a recode anyway that whatever changes you want would take about the same amount of time to code. as you have only one coder and at least two people who can work on deciding what should be coded, would how much code you need to write not be a sensible focus for timesaving measures?

as far as i'm aware, pax runs on mysql, which is what i assume you'll be using for pa:n as well. therefore, would it be possible to, say, use the alliances code from the old version with the game code with the new version? if not, might this kind of thing be something to look at?

[Kal]

Originally Posted by mist

i'd have thought that with it being a recode anyway that whatever changes you want would take about the same amount of time to code. as you have only one coder and at least two people who can work on deciding what should be coded, would how much code you need to write not be a sensible focus for timesaving measures?

as far as i'm aware, pax runs on mysql, which is what i assume you'll be using for pa:n as well. therefore, would it be possible to, say, use the alliances code from the old version with the game code with the new version? if not, might this kind of thing be something to look at?

its more than a code change though, one of the biggest changes is the enxt gen user interface, and in the old code there isn't a clear distnction between fucntionality and interface so its rather difficult to use the old code.

Our problem isn't so much coding time its that design time has eaten into coding time

[mist]

a distinction between functionality and presentation would indeed seem like a good thing.

however, could you not use the functionality *and* presentation from the old alliance system along with the new game, for a time? or for that matter, use the old 'front page' for a while with the new game? both would seem feasable with the currently releasted cunning plans.

[Kal]

Originally Posted by mist

a distinction between functionality and presentation would indeed seem like a good thing.

however, could you not use the functionality *and* presentation from the old alliance system along with the new game, for a time? or for that matter, use the old 'front page' for a while with the new game? both would seem feasable with the currently releasted cunning plans.

technically yes - but wouldn't it be better to release a finished product than one thats bodged together? We do afterall want people to pay for it.

[robban1]

remove exile all together and its all about luck if you get a decent gal or not

[mist]

Originally Posted by Kal

technically yes - but wouldn't it be better to release a finished product than one thats bodged together? We do afterall want people to pay for it.

blah blah blah volunteer labour blah blah blah

i doubt jolt will fund free rounds for ever, so it would seem to be a choice between a semi updated product and a not updated at all one.

[Appocomaster]

The idea behind mine is that, no matter how the shuffle works in practice, you won't ever get galaxies that are "equal". There's no way you can use exile to do anything like keep all galaxies "equal" in activity or anything else.
I think the best way of doing things would be to accept that bottom galaxies exist and to try and keep them as small as possible. The last few rounds, whenever I've looked, the galaxys have been almost a straight line in terms of score (if sorted from top to bottom). I think it'd be better to have a bigger "bunch" in the middle / upper middle section. The majority of the players would then be in the active universe.