Log in

No account? Create an account
Ramblings of Purest Sanity [entries|archive|friends|userinfo]

[ website | Read this from time to time. It's good for you. ]
[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

Verb me! [May. 10th, 2011|01:40 pm]
I've been trying to work out how to word something and haven't quite found the correct way to phrase it. Here's the situation:

I'm running a simulation in which all of a particular type of agent starts in one area:


Over the course of the simulation agents might move around or change type or whatever. So there might be more or less Xs than at the start and they might be in different places. I'm looking to describe the situation where the Xs end up outside of the intial area (this only happens under some combinations of assumptions and parameters) For example:


(I hope you're all reading in a fixed width font btw)

How do I describe this? Currently I think I've said things like "The group expands out of the inital area" or "The group spreads more widely through the social network" The problem is I don't want to be ambiguous with the behaviour where the group simply gets larger. For example:


In fact if you look at the "outside the inital area" example above it's got smaller. So what's the verb for what a social group does when it occurs in different places in a social network, putting it into contact with new outgroup members, but sometimes reducing the overall size of the group. Spreading? Expanding? Evangelising? Distributing? Scattering?
link3 comments|post comment

Morning [Apr. 26th, 2011|07:43 am]
I woke up this morning and I'd had a good nights sleep for the first time in a while. I got up and dressed and wondered downstairs to make myself some breakfast. I ran into Siz getting her own breakfast and we started talking about things, it became quite a serious conversation so we moved back upstairs, as we tended to do when things got like that. She was sorry that she'd treated me so badly and wanted to have another go at making our relationship work if I did. Then I heard something under the bed and reached down to find it was Mattisse, who made a half hearted effort to bite my hand off, before rolling over, demanding snuzzles and starting purring nicely at us, which made us laugh.

Then I woke up and realised four things in quick succession:
(1) Something's wrong with my spine and I'm in blinding pain.
(2) I'm still broken up with Siz.
(3) The cat I grew up with for 20 years is still dead and I still wasn't there when it happened.
(4) I have really only have five hours sleep and have lots of work that needs doing.

This is not the greatest start to a day ever.
linkpost comment

House [Apr. 23rd, 2011|01:27 am]
I desperately need either two or (preferably) four people to take over this house next year. It is a truely epic house and has brought me a lot of happiness. It's significantly improved since we moved and the smallest room is bigger than some of the largest rooms in other houses I've been in. Seriously, this place is a steal, I'd have stayed here if I could've found the group of three to let me. Get in contact if you're interested.
link5 comments|post comment

Predictive Charity [Apr. 13th, 2011|11:52 am]
I've mentioned this paper before (Sato, 2010) but I've come across it again because it's going into my thesis under "Reasons that building predictive computer models can be pretty handy"

I'm reading a paper published August 2010. It discusses a big earthquake predicted to hit Japan that was apparently predicted back in 2004. Specifically it looks at the level of prefabricated housing avaiable to deal with displaced residents who no longer have homes and analyses the predictions of an expert panel set up by the Japanese govornment 2006-2008 to look at the evacuation problem following such an earthquake.

The recent quake itself really wasn't a surprise, it seems that the magnitude of the problem was anticipated to be greater than before:
The anticipated Tokyo Metropolitan Earthquake is expected to be approximately eight times more destructive than the Kobe Earthquake (1995) in terms of damage to housing; it has already been termed the "Super Urban Earthquake Disaster."

Now the response to this disaster has been tremendous. I haven't been able to turn around without seeing some individual or institution giving generously to help those who've suffered through the crisis. What I cannot help but wonder is how much more effective could the money have been if it were spent in advance of the disaster to ensure that the extra resources (temporary homes, medics, aid workers etc.) were already on the ground when it happened?

Or does that happen? When a big disaster is anticipated does someone (govornments? big charities?) spend big in advance to help prevent the worst of the problem knowing that charitable donations after the fact will cover it? Is there room for a predictive charity, where people give to cover disasters that haven't happened yet, knowing their their money can be more effectively spent in advance than after the fact? Does such a thing already exist?

I just can't shake the feeling that given some things can be predicted so far ahead of time we could be better at dealing with them.

But then maybe I shouldn't be surprised, look at how the environments going.
link1 comment|post comment

Self trolling [Apr. 2nd, 2011|12:37 am]
So I went back to the notes for that LARP thing, since I figured if it'll ever happen it'll be in the summer term. Since I (desperately) need something to distract me this evening I figured I could finish writing it. I have discovered what I can only assume to be a deliberate provocation from my past self:

"To solve this plot hole I could go back to the bar encounter but add an acolyte who is interested in"

That's the very end of the document. Yeah.
linkpost comment

Musical Interlude [Mar. 28th, 2011|07:21 pm]
Earlier today Charlie got me to thinking about series that, for some reason, feel the need to include a single musical episode. The Buffy musical episode comes to mind most readily, but I can think of a few others. I'm told that these episodes are always bad, which they most certainly are, but I also enjoy them a lot.

So...what sort of examples of this can you think of and would you come to an evening where we decided to pull a bunch of them togehter?

Also I've not forgotten to reply to the comments on my last entry, I'll get around to it :P
link7 comments|post comment

Demons [Mar. 26th, 2011|04:28 pm]
Did a quick simulation of my demons final battle, assuming that the players do nothing and skip town before the battle starts.

It ended with King Marcus with his unit 90% killed surrounded by 600 zombies, 250 skeletons, 250 infantry, 150 knights and 50 elite undead. There were 40 elite undead outside of the city and that was pretty much it. 163 friendly individuals and 100 enemies left the battlefield before the end.

That means everone else from the 1,273 friendly and 5,211 enemy units died.
linkpost comment

I beg you to differ [Mar. 23rd, 2011|10:20 am]
I just came across a paper discussing how similar friends are in terms of their opinions. Essentially it concluded that friends are more similar to each other than they are to random strangers, but are less similar to each other than they think they are. Apparently the reason for this is that when you infer your friends opinions discussions you have with them are part of the process, you also use a lot of sterotypes and project your own opinions to fill in any gaps (or even over their real opinions).

I'm now curious about where I've been doing this with my friends. With what frequency do I think you hold an opinion that you'd disagree with? I can think of no proper way to test this (or more accurately I can, but I have enough work of that type on) so I'm just going to throw it open. What do you believe that I might not know? What do you think I believe that I might not?

Gimme a few on each point and I'll try to respond in kind.
link5 comments|post comment

Pokemon II [Mar. 21st, 2011|01:59 pm]
So last time we got up to speed on the basics of the game, a few folks seemed interested so I figured I'd ramble on the subject a bit more. I'm also interested to hear what you fine folks have to say about the game. A few of my favouriate things:

Ditto is a great idea, it only ever learns one move, which is to turn into the opposing pokemon and inherit all of its moves. I love stuff like this, since a lot of pokemon are resistant to their own moves (especially if they have STAB) so it can end up slowing down the game a bit. This is wonderful if your opponent is poisoned and you aren't.

Smeargol is a beagle that paints using its tail. It also only learns one move, but it learns it a bunch of times. It's called Sketch and when used it copies your opponents last move and replaces itself forever, so Smeargol can learn literally any combination of moves, including things the designers didn't put together for fear that they'd be too broken. This isn't my favouriate use for it though, my best use is to get it to copy dittos move and then use the other three slots for moves you otherwise wouldn't use, but are handy outside of combat or that you'd only want to use once per battle and wouldn't was a slot on. A prime example is...

This is a great move that goes underused. It swaps your opponents item for your own. Most people use items like "Do more damage" or "Heal every round" or similar. However there are some items that rarely see any use such as "The owner goes last" or "The owner is badly poisoned"

Kecleon is an awful pokemon that loses every battle but I can't help but enjoy it. Whenever it's hit by something it turns into the type of the move that hit it. So if it gets hit by a lightning attack, it becomes lightning type and therefore is resistant to further lightning attacks. It would be great if it's stats didn't suck so bad and that it's impossible to get STAB reliably and that it doesn't learn any good moves. Also dragons kill it. A lot. (Dragon moves do 2x damage to dragons) I still love it and played through a lot of the last game with one. As a bonus it has a stripe that makes it look like Charlie Brown.

I've only ever seen this once, in the hands of an opponent and I don't think I'll ever get one since you need to have been to a special event (Grumble grumble). It's a shame because it has a great schtick, which is worthless in the main game but would be fun for PvP - when you switch it in, it takes the appearance of one of the other pokemon in your party. The AI doesn't care, but a human opponent will almost certainly use the wrong type of move on it, which is all fun and games :) Especially if you teach it U-turn, so it can make an attack and switch out and your opponent won't know when it comes back later. I think I'd disguise it as something your opponent is forced to attack a particular way, like...

Much love for this one. It has a maximum of 1hp and can never get any more. In the event of taking damage it ends up really really dead. To make up for this it is immune to any move that isn't supereffective against it. So if your opponent happens to not have the right set of moves they can't possibly beat it - a great one to keep back for your last pokemon if you're feeling lucky. On the downside as a Bug/Ghost pokemon it is vulnerable to a lot of things. And you can always set fire to it.

This is an ability that some pokemon might have. It's effect is that if you are at full hitpoints, no move may reduce you below one hitpoint. I love the idea of fighting with a level one pokemon against much higher levels and being unkillable because you heal to full health (which would be 4/5 Hps each turn). The only way I see to do this is to use shell bell (An item that heals you for 1/8 the damage you deal) and find some attack that dealt a consistant amount of damage when faced with a relatively high defence stat. Such as...

...Pain Split
This move adds your health to your opponents health, divides by two and sets both of your life totals to that. Having it on a level 1 sturdy pokemon would let you halve your opponents life each turn until you brought them down to your level, while keeping you invincible. Sadly this combination is currently impossible, but it's still a great move in lots of situations.

U-turn/Volt switch
These moves let you do an attack and then swap out to put something else in your place. It's great fun if you've got them on something fast and can afford to slap something and swap to something else that's resistant to them. Also there are lots of abilities that trigger on you joining or leaving a fight, if you set it up right you could heal half of your life each time you do it or some similar dodge. Someday I'll make an entire team that use just these moves to bounce back and forwards.

This pokemon changes to match the weather. It's normal type, but if it's funny it becomes fire type, raining makes it water type, blizzards make it ice type and sandstorms...just hurt it a lot. I don't know why it can't become rock type. Anyway what's great is that it gets a move called 'weather ball' which does damage matching the type the current weather gives it. So it can voluntarily change its type as it wants and then hit with STAB in that type. In reality it doesn't work this well because your opponent just switches out when you change the weather, but I love the idea of it enough that I used it for most of the last game. With Kecleon it made up my duo of being cool but ineffective.

These moves change the weather for a few turns, which I think is a great idea. They all change how effective certain types of move are. Some of them change specific moves (e.g. thunder becomes more accurate in a rainstorm). Some abilities trigger off them (e.g. a dry skin pokemon heals every turn in the rain). You can do all sorts of cool things if you build around making the weather go a particular way. Of course if your opponents done the same expect to spend half the game fighting over how the weather ought to be ;)

There are lots of more effective things that those outlined here. Earthquake has the best damage of an accuracy 100 move. Spore has the best hit rate for making things sleep. Various dragons are the best pokemon in terms of having great moves and lots of hitpoints. Baton pass, Screening, Spiking...there are lots of commmonly used practices that work well. The thing is I play for fun and I find the cool or interesting things a lot more fun than having buckets of raw power.

So if you play, which things do you find fun / think get underused?
link2 comments|post comment

Pokemon [Mar. 18th, 2011|11:13 am]
Pokemon is one of my guilty little pleasures in life. I'm not sure what it says about me that I consider fetishism a regular pleasure and playing a childrens computer game to be a guilty pleasure, I try not to think about questions like that too much. Before I get distracted writing about what I intend to write about, I know this is a something that a lot of you (guilty or otherwise) engage in, so perhaps we should get a group together and play some multiplayer sometime next week?

Anyway, I got to thinking about the level of strategy in the game, as I inevitably have done with every game since snakes and ladders (Dominant strategy: Get the first turn by any means available, other than explicit cheating this is the only strategy that makes a difference, dominant strategy offers approximately a 50.001% chance of winning). Partly this is because the AI is so bad, I can't decide if that's to make the game easier for its target audience or to let you always be the underdog (in terms of raw power) and have a good chance of winning anyway.

Ultimately you win by removing all of the opposing pokemons HP, so sooner or later it comes down to dealing damage. It turns out the formula for this is more complicated than you'd expect:

Damage ~= (Attack x Movepower / Defence) x Typebonus x STAB

Attack is more complicated than it seems, because you have a regular and special attack rating, each move is either regular or special. So you should choose your move based on which of your attack ratings is higher.

Defence is more complicated than it seems, because it is also broken into regular and special. So you should decide based on the ratio of your attack rating to their defence rating. But you cannot explitally check your opponents defence rating, you can get an idea from what pokemon it is (All of their stats fall within a specific range) but you could also actively train your pokemon to have high ratings where they normally have low ones to trick your opponent. You could always experiment in the battle, but that means potentially wasting a turn on a suboptimal attack.

Typebonus is a multiplier based on the type of the move and the type of the defending pokemon. This multipler can be big enough to blow all other factors out of the water, ranging from x0 to x4. However the defender can have two types, in which case their resistances are determined by multiplying the bonuses together. Also you cannot explicitally check your opponents type, but this at least is fixed by pokemon, so if you just memorise all 600odd of them then you're set.

Also the type chart, I'll get you started but I don't want to record the whole thing here: Grass does double damage to water which does double damage to fire which does double damage to ice which does double damage to ground which does double damage to rock which does double damage to fire which does double damage to bug which does double damage to psychic which does double damage to poison which does double damage to grass which does double damage to...There are 17 types and in about a third of cases something will happen if one type hits another, you get the idea. Also it's not reflexive or in a consistant pattern, for example fire deals half damage to water and water deals double damage to fire. However electric deals half damage to grass, but grass only deals regular damage to electric. Also if fire hits fire it does half damage, but if normal hits normal it does full damage and if dragon hits dragon it does double damage.

So you need to choose the move with the best typebonus that is of the type (physical or special) that best maches yours and your opponents defence.

Then there's STAB, which stands for "Same type attack bonus". You add 50% to your damage if your move is the same type as the attacking pokemon. So sometimes a nominally supereffective move performs less well than you'd expect.

Finally there's movepower, all of these multiplers mean nothing if you don't know what you're multiplying. A power 120 move is probably better than a power 10 move with good stats, STAB and typebonus. This is often context dependant for some types of move, so some moves deal more damage if you're at full HP, or if your opponent is injured, or if it's the first turn etc.

So that's it. It's a little complicated, but the right strategy is to take your movepowers in context, multiply by your appropriate power and your opponents appropriate defence, apply the byzantine type system, account for STAB and pick the one with the biggest number.

Only moves can also only be used a limited number of times per battle, so you need to account for that and make sure you don't waste a move that you'll need later.

They also have accuracy, so you need to determine that moves individual chance of missing and factor that into your decision. Is it better to maybe kill your opponent this turn or definately get him next turn? What will his moves do to you (No you can't check your opponents moves either, so it might be worth having a scout whose job isn't to kill but to survive getting hit by everything and learn your oppnents capabilities)

Also a lot of moves have special effects, like swapping your pokemon around or putting various sorts of upgrades and downgrads on your or your opponents pokemon. Sometimes a mix of the two. Some types of change will persist if you change pokemon, others will be gone. There's even a move to let you take your nonpersistant bonuses with you when you change - but do you want to deal 0 damage for a turn just to get that?

But aside from the PP, accuracy, special effects, attack/defence, movedamage, contextdependantdamage, typebonus, STAB there's no much to consider.

Well there is one other thing, you or your opponent could give up making a move in order to switch pokemon. This switch happens before anything else. So if you can anticipate your opponents move you can swap to something that will resist most of the damage (through appropriate defence or typebonus). You also need to work out whether they are likely to switch and if so what to, so you can pick a move with the appropirate bonuses for taking that thing out.

Of course there are moves that affect switching, making it impossible, making the new pokemon take damage when it occurs, forcing the pokemon to switch in, switching out while making an attack. Come to think of it there are moves that affect everything mentioned so far, moves that change pokemons type, moves that change their defences, moves that change the odds of your move going off first, moves that change how many PP other moves cost, moves that heal damage, moves that alter accuracy, moves that influence other moves for context dependant damage.

Also there are two things I've not mentioned yet. Each pokemon has an ability, which does something special and each can be holding an item, which either offers a permentant bonus or is automatically used for a one off bonus. There are various combinations and synergies between these, for example holding a "wake up if asleep" item and using the "Heal to full health but fall asleep" option.

Really though, that's just the simple stuff. The really complicated part of the game is how you set up the pokemon in the first place. Since you get to choose which ones you have and can influence which are their high stats, which element they are, which ability they have, which four moves they have and what other pokemon are on the squad that might benefit them. That's where the really deep strategy comes in.


Who the hell makes a game like this for children?
link12 comments|post comment

[ viewing | 10 entries back ]
[ go | earlier/later ]