AI size limits

[Revenge]

Look what I found
viewtopic.php?f=2&t=1093

Yeup! 25, 50, and infinity were my suggestions. You said 20 is fine with constraints, but I don't think any constraints would happen and would be difficult to decide as the game developed. So 25 seems fine. Also, I might as week announce it now...

My next node league tournament will be with 25 nodes. So I'll give players a week to prepare their ais.

[BenJV4]

I voted "No..."

Simply because it will limit the possibilities. And the very last goal of this game is to find one totally universal AI, which is an endless goal.
It does not fit well with the random map as well because random implies you have to take a lot of possibilities into accounts. I would be very unpleased if I were limited in nodes count and then seeing my bot not doing what I want to in some special case and not being able to do anythings cause my AI is "full" and already optimized for the most common cases.
In fact, making the AI limited make the game like "Make the most optimized AI for the most common cases", there is only one so once you reach it, "End".

The unlimited count implies that there is an unlimited count of AI, you can always find a better AI taking into account a new case. And as we have random and thus unlimited count of Map, we have as well an unlimited count of cases, and there you go for an infinite game lifetime.

I understand the remark and/or needs for easier gameplay for beginner so it is fine to create a special category or some tournament if it does not replace totally the real challenging category "unlimited".

You can't please everyone. If you set only unlimited category, you may lose the "lazy" players.

If you remove the unlimited category or make it marginal, you may lose the "brainstorming" players. You will lose me for instance. This is not a matter of intelligence, but really lazy/brainstorm way/willing to play.


What is the targeted community?
I think the game is made for the "brainstorming" category of player and whatever you try to do to ease the game, you will most probably still lose the lazy one for another millions reasons.

[pier4r]

I still believes that showing the node count / node evaluation at the end of the match is psychologically enough.

Who loses to someone with much more nodes will accept the loss better , who loses to someone with much less nodes will think about optimizing his ai.

[GFX47]

I still believes that showing the node count / node evaluation at the end of the match is psychologically enough.

Who loses to someone with much more nodes will accept the loss better , who loses to someone with much less nodes will think about optimizing his ai.

Nice idea. I'll try something like that in the next iteration. Shouldn't be too complex.
Beta 3 should include new game modes/rules. Hopefully I'll find a way to support modding on this via steam workshop, allowing players to make any combination between modes and constraints like AI size. Can't wait to work on this!

[ElvenMonk]

Sending our best prays to you on this, GFX!

[pier4r]

Wait finally we get the node count that we discussed for like 4 months? Yay!

Love gfx!

I guess I will end playing a little and dividing my donation budget between you and Wikipedia.

[ElvenMonk]

Forget Wikipedia plenty of people already support it, give it all to gfx)

[pier4r]

Off topic: I don't like this way of thinking. That's the tragedy of the commons.
In short the tragedy of the commons says "oh there is a free resource I use it as much as I can without caring for it" and then the resource gets exhausted.


If something is nice I don't have to think "oh lol others pay for it" to take it for free.

Otherwise also piracy is justified. "Oh lol there are enough people that pay for this, I can take it for free even if it is valuable for me".

No. What is valuable for you deserves your contributions. You cannot assume that the others pay for you. Otherwise it is freeriding or in any case it is not sustainable.

[GFX47]

AMEN TO THAT

[ElvenMonk]

That's not what I meant. I said you can safely put Gladiabots way above Wiki in your priority list. I have a lot to say about Wikipedia specifically, and I'd argue it deserves anything but judgement, but let's not start flame on this here

[ElvenMonk]

Good catch GFX did in some another topic it's actually lines not nodes, that matter.

With one line you can connect only one node, even if you have more in AI.
And if you have multiple lines going to single node, it can make a huge difference.
.

[pier4r]

Today I had an idea (discussed also on discord a bit) about a solution to make leagues for a certain amount of nodes.
It is a bit like some competitions that let everyone compete and then show the best placements according to categories (for example in a marathon).

Simply showing rating points / node count (branches count would be more appropiate but let's start small).
Surely one can "cheat" reaching a nice rating and then deploying micro ai, but the ghosts will show the trick.

In this way everyone can see how efficient one is. One problem would be the psychological effet for newcomers but I don't think this is a problem. This because, as in many competitive sports, the more competitive one get, the larger the amount of effort required for small gains. Therefore towards the top the number of nodes needed to implement various ideas should be quite substantial.

So a newcomer having, say, 40 nodes, and a rating of 1300 won't be much worse than someone at 2300 with 300 nodes. The former player will have 32.5 rating points per node, the latter 7.6 rating points per node.

So actually it should encourage newcomers and can provide an hint if a player has, for his rating compared to the ghosts he is fighting, too many nodes. Maybe in that case it is time to rework the ai.

The main point being: players should realize that they have to consider the rating/nodes ratio only of those players with similar rating, as each rating range has a certain difficulty to overcome.