Darwinbots Forum
Bots and Simulations => DNA - General => Topic started by: Jez on December 14, 2006, 02:46:06 PM
-
Firstly, if you could spare the time and have an idea about which bots use viruses, could you glance at the F1 league table or even better the F1 bestiary and name any bots which use viruses but don't have the 'vir' tag in the type/league part of their name. (I run stuff with video disabled most the time so am unlikely to pick up the ones I've missed) I think the two DIN bots should have (vir) added for instance
Secondly, do you think viruses are overpowered?
Seems to me that they are, it seems a pretty free way to gain a great advantage over your fellow bots, I've been wondering recently if there is a relative disadvantage to the advantage they gain for using viruses. I'm all for equality, every method, ties, shots, viruses, even MB's (in the future maybe) being pretty equal with the programming being the key to success.
If you do think they need an added disadvantage for using viruses then what should it be?
Cost, time, perhaps even a better defence against them (1G bots are the best defence atm) or something else?
If you think they are balanced at the moment then what do you think the balances are?
-
I dont find viruses very effective in the leagues which is why I dont use them, whenever I have played around with adding a virus to a bot it always seems to perform worse, but that may be down to my design philosophy
-
In my experience viruses are a sort of zen art form. Viruses which instantly kill another bot generally are only going to be marginally effective. However, viruses which dampen the host or makes it lame are very effective.
The question in my mind isn't if viruses are overpowered, it's if slime is underpowered as a method to stop them. Someone should try doing a cost analysis of slime vs. viruses.
-
Other than the time taken to make a virus, which I didn't think stopped a bot doing anything else in the meantime, or the cycle needed to fire a virus, don't know if it can still feed through a tie while firing virus, I didn't really think there was a cost imposed on viruses.
Good point about slime maybe being not effective enough v viruses. If I knew how to do a cost analysis of viruses v slime then I would, instead you'll just have to wait until I've finished the leagues and I'll play about with a slime bot v a top virus bot and take a guess.
If you can enlighten me on the cost of firing a virus then I'll have a better idea, certainly a bot can spend a fortune on making slime.
-
When you do 100 .vshoot store to shoot the virus, you are using 100 energy units, also its direction is random and a bot with no slime still seems to have a 50%ish chance of stopping the virus.
-
“good cost-benefit or cost-effectiveness studies are complex, require very sophisticated technical skills and training in methodology and in principles of economics, and should not be undertaken lightly” (White, 1988)
Hmm, methinks me is a little underqualified for this but nm.
I could test some of this a bit more if I wasn’t running leagues atm but thought I’d sketch out my thoughts anyway;
1. Slime degrades over time (not sure what rate maybe 1 per turn (Can’t run sim slow enough so hard to see))
2. Max level of slime is 5000
3. What is the effectiveness of slime, 50% resistance for none? 100 or 99% for max?
If max level of slime is 5k with 100% resistance and viruses are 100% effective, (as the best ones are) using 5 bots per side, gaining 1 nrg per turn, Bot A virus attacker Bot B defender:-
Let’s set the probability of Bot B of hitting a virus at 0.1% all viruses shot for 100 turns.
Bot A fires 1 virus every 100 turns which survives 100 turns so after the first 100 turns there is always a 0.1% chance of Bot B hitting a virus. (0.2% effective so every 2000 turns destroys bot B )
Bot B spends 1 nrg per turn making slime which is worn away at the same speed so always has 50% chance of surviving and no chance to reproduce.
Bot B loses
Nrg gained = 2 per turn
Bot A fires 1 virus per 50 turns; chance of Bot B hitting is 0.1% after 50 turns and 0.2% after 100 turns. (0.1% effective so every 1000 turns destroys bot B )
Bot B spends 4999 turns making slime and another 1000 turns to reproduce.
Bot B loses (using 50% rather than incremental %’s I admit) (50%/5000nrg = .001 resistance gained per turn?)
Where’s MP? I know he likes this maths stuff! Not saying that the starting figures are correct but certainly at 1 nrg per turn it would be unfair. Giving both bots the same amount of nrg input per turn and allowing for slime degradation I think that (using these figures) it is more cost effective to use viruses than it is to use slime.
I don’t know how high you’d have to raise the nrg per turn (or lower the chance of a virus hitting Bot B ) to make slime cost effective but figure that it’d have to be higher than 100, the point at which Bot A could fire a virus every turn.
The acid test will be if (the F1 league rerun finishes ok) the effective (working) virus bots all end up in good positions (not respecting 1G antiviral bots)
Marks out of ten?
-
You want to make viral defense even stronger? It's already possible to add a viral defense to a bot that is 100% effective and doesnt cost any energy by reducing the bot to an SG.
-
I know, the top bots are looking like being all 1G anti viral and virus bots. Last time I tried a rerun of the F1 league, got to match 14, the top six bots were 1G and Virus bots. The only bot I thought was viral and didn't manage to stay up top was DIN2.
Perhaps the F1 league should turn into 1G and viral league? It seems rather a speciality field, not something to encourage new visitors to DB when it comes to writing their first bot. Lots of experienced bot writers don't like writing 1G bots either, they can be a bit of a pain.
I've always liked the way bot's abilities were balanced, paper scissors stone. Not just V or 1G with all the others 2nd class robotizens.
What do you think?
-
1. Slime degrades over time (not sure what rate maybe 1 per turn (Can’t run sim slow enough so hard to see))
2. Max level of slime is 5000
3. What is the effectiveness of slime, 50% resistance for none? 100 or 99% for max?
1. Slime currently degrades 2% per cycle.
2. Max slime level is 32000
3. Slime effectivness against virus shots is a function of the shot's "power" which in turn is a function of the shots nrg (proportional to .vshoot) range (how far the shot had to travel to hit the victim - further means less effectiveness) and value (which appears to be the gene number usede = go figure).
I think this whole area has seriosu bugs. The way shots are powerred and effectiness gugaged is incredibly convoluted. I've been reluctant to mess with it. Perhaps in the new year I (or someone else) will spend some time on it.
-
Did the bug where they would vshoot other genes than the intended one(original gene deleted or virus added prior to) ever get fixed? Caused some really weird runs.
Another problem is the plants, with no slime and infinite nrg they are pretty much sitting ducks.
Err... the original Din doesn't have viruses, only Din2 does.
-
Wow, I didn't realise that slime was that expensive. Maybe why the bot I added an 'always make loads of slime' to never got past 5000 slime.
That is a convoluted way to work out virus v slime effectiveness, perhaps also why sometimes viruses work really well for one bot and not another (the gene number value bit )
Err... the original Din doesn't have viruses, only Din2 does.
Thanks for the info. Din looks like it may end up the top non virus/1G bot then. We'll find out one day
Being able to freely infect plants is indeed another advantage for viruses.
-
you could try adding hunter 2.3, excalibur 1.11 and ymir into the F1 league run, I would think they would all do quite well
-
I'm doing a league rerun under VB atm to try find that overflow bug otherwise I'd try them out. I wouldn't have any problem putting old bots back into the league though.
-
I'm doing a league rerun under VB atm to try find that overflow bug otherwise I'd try them out. I wouldn't have any problem putting old bots back into the league though.
So, after 6 days, I gave up trying to find the memory leak under VB. I ran for 6 days straight without a crash. I bet the memory allocation strategy used by the intereptor is different from the executable or the problem is something other than a memory leak - something that does not manifest when running in the debugger. I'll have to find it some other way. So, you might want to bail on running in VB...
-
I ran for 6 days straight without a crash. I bet the memory allocation strategy used by the intereptor is different from the executable or the problem is something other than a memory leak - something that does not manifest when running in the debugger. I'll have to find it some other way. So, you might want to bail on running in VB...
You are the expert, it's taken 3 days to show up before when not running the debugger though. I can't seem to do a full rerun of the league with out hitting that bug either. I'll leave it a couple more days just in case. If it doesn't show up using VB it might be the quickest way for me to get the league redone atm!
-
Guys.. Viruses here are underpowered..
First. I downloaded new DarwinBots 2.42.9 and noticed at least one thing.
Viruses shots goes much closer that before.. All my test SIMs w/
viruses shows much less infection rate.. So now they are quite uneffective.
I wouldnt touch mkvirus thing, its slow.. but let it be.. but virus should go much
far away that now. Many viruses can go throught significant distances to infect someone ;)
Thats why they are till here.. after billions of years.
Im voting for fixing it again :>
-
Guys.. Viruses here are underpowered..
First. I downloaded new DarwinBots 2.42.9 and noticed at least one thing.
Viruses shots goes much closer that before.. All my test SIMs w/
viruses shows much less infection rate.. So now they are quite uneffective.
I wouldnt touch mkvirus thing, its slow.. but let it be.. but virus should go much
far away that now. Many viruses can go throught significant distances to infect someone
Thats why they are till here.. after billions of years.
Im voting for fixing it again :>
I beleive this was fixed in a more recent version. Try the latest buddy drop.