Anti Anti Virus

[Numsgil]

Here's my thoughts on overcoming some of the new anti virus techniques.

Most of the new ant virus techniques work with *.thisgene and *.genes.

I haven't tried this yet, but what if you make your virus do:

.delgene dec and .genes dec.

.delgene dec will make any delgenes miss their mark, probably screwing up the bot.  This only works if the virus comes at the end of the genome.  If the bot isn't trying to delgene anything then delgene will be < 0, and so fizzle.

.genes dec makes the bot think that it doesn't have too many genes.  This only works at the beginning of the genome.

Most of the anti viral stuff I've seen are shortbots.  Which makes sense when you don't want to mess with isolating a virus in 50+ genes.

I'll probably work out a SC2 at somepoint that tries to overcome anti viral techniques.  It's amazing how the SB are taking over the leagues  :D That's what you get with the new math operators.  Such power.

[Light]

Im not sure I want to give you pointers on this topic :bigginangel:

 I know the weaknesses in Furiae's antivirus code and what you are suggesting is heading in the wrong direction :rolleyes:  

Antivirus code Ironically would do better in longer bots, it would be harder to mess it up

[Numsgil]

Tried it out against Shen's The One, and it works.  Sorta.

The bot deletes itself instead of the virus.  Then other The One's take it out because they don't recognize it as a conspec anymore.

Problem with viruses is that you have to keep the slim as possible so they copy to a shot faster.  I'm working on streamlining SC's virus right now.

[shvarz]

The bot deletes itself instead of the virus. Then other The One's take it out because they don't recognize it as a conspec anymore.

This just shows how nicely we simulate the viruses :)  It is one of the most typical reactions to viral infection!

[Numsgil]

Well I'm gald you're happy.   :ph43r:  I'm up the proverbial creek without a paddle.

[Shen]

Seems like we have a virus arms race on our hands  :ph43r:

On a side note, you're right about the new maths operators being powerful. Perhaps too powerful, especially since there free. Have you thought about making the cost equal to a store? I dislike the free actions because they area a weakness in the balance. Same goes for inc/dec and to some extent 'free' rotation.

Also the new math ops arnt included in any mutations. Just a niggle :P

[PurpleYouko]

Are you talking about stuff like floor and ceil when you mention new maths ops.
I am not sure whether the mutations have been extended to these yet but I can say for sure that the newer comparrison functions like <= and angle, have definitely been added. I did that myself and I have seen a bunch of mutants with these things in them.

[Numsgil]

I've been thinking of adding costs to all things you can do thats comparitive to its relative complexity.

For instance, add takes X many basic commands (basic being OR, NOT, XOR, AND), so it should get charged X * basic command cost.

That allows things to get more complicated, but get charged the same amount.

Just haven't gotten around to it.

[abyaly]

If you make storing/incing/dec'ing to an invalid location cost the same amount as evaluating a COND then the cost differences between SG bots and normal bots should be indistinguishable. In fact, SG bots will be slightly more expensive.

[Numsgil]

That's true.  Maybe a new cost field should be added which lets you control the cost of invalid stores.

Nice thinking