Node Reactivity Depth Discussion(+10ms per node)

User avatar
Revenge
Neural Network
Neural Network
Posts: 289
Contact:

Node Reactivity Depth Discussion(+10ms per node)

Post#1 » 16 Jan 2018, 18:49

From Trello topics:
https://trello.com/c/YFkL1zvp/31-increa ... des-is-low
and
https://trello.com/c/xwUZh8fQ/820-bot-c ... -frequency

The change to the game effectively limits ai node branches from becoming too long, and forces people to build ai a specific way(horizontally). I do not agree with this change at all, as it limits creativity, while punishing good ai building.

This change feels unnecessary, considering node leagues would essentially do the same thing, but would allow players to keep their creativity in designing the ai the way they choose.

Feel free to discuss whether you agree or disagree with this change, and what you think the cap should be if it is implemented. (Maximum delay that can be added to a branch(50ms? 70ms? 100ms?))
Last edited by Revenge on 16 Jan 2018, 18:52, edited 2 times in total.

MrChris
Automaton
Automaton
Posts: 149

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#2 » 16 Jan 2018, 18:51

+1 totally agree.
------------------------------------------------
My in-game name is MrChris

Creator of the unofficial Gladiabots stats page: https://gladiabots-stats.info.tm/mrchris
And the Gladiabots retreatment simulator: https://cmrichards.github.io/glad_simulation

MrChris
Automaton
Automaton
Posts: 149

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#3 » 16 Jan 2018, 19:09

Code optimization should be done to make your AI simpler to make you AI more agile. This is something a player should decide to do by themselves when they think the time is right and they've learnt enough.

It should definitely not be enforced by the developer of the game onto the player by punishing them.

Terrible idea. I can't think of a single reason why this makes sense.
------------------------------------------------
My in-game name is MrChris

Creator of the unofficial Gladiabots stats page: https://gladiabots-stats.info.tm/mrchris
And the Gladiabots retreatment simulator: https://cmrichards.github.io/glad_simulation

mcompany
Autonomous Entity
Autonomous Entity
Posts: 857

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#4 » 16 Jan 2018, 19:25

I have more to say than just this question (I'll wait until I set up my laptop), but one of those trello cards implies a cap to the amount of lag you can get. What is that cap?

User avatar
Ritter Runkel
Neural Network
Neural Network
Posts: 482

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#5 » 16 Jan 2018, 20:10

Please don't add these kind of restrictions. We had the discussion some weeks ago:

viewtopic.php?f=5&t=1180&hilit=Size+limit&start=40

Jeffrey Harpen
Algorithm
Algorithm
Posts: 91

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#6 » 16 Jan 2018, 20:28

I don't see any advantage in adding such a feature either. The more restrictions, the more disappointed players get (especially newbies).

mcompany
Autonomous Entity
Autonomous Entity
Posts: 857

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#7 » 17 Jan 2018, 17:41

Gfx?
mcompany wrote:I have more to say than just this question (I'll wait until I set up my laptop), but one of those trello cards implies a cap to the amount of lag you can get. What is that cap?

pier4r
Skynet
Skynet
Posts: 3053

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#8 » 17 Jan 2018, 22:13

As I wrote on telegram (well, I am mostly inactive in this period, snort) I see it as a way to reward small AIs of newcomers/creative people vs large AIs of veterans. This because a large AI by itself is not inherently better, indeed it takes a lot of games to polish it.

Anyway exactly for what I wrote above I don't really see the need of rewarding smaller AI. Having a large AI is already a chore (a partial reason why I am not yet back) and so who has a small AI is rewarded on the long run to be more flexible. To keep a large AI flexible - at least until we have a really powerful editor with multiple nodes copy/paste/clipboard - one needs quite some work, especially to avoid conflicts.

TL;DR: I understand that is an attempt to make things "fair", but it misses the point, as large AI are already maintenance intensive by themselves.
http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

User avatar
GFX47
Dev Team
Posts: 2416

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#9 » 18 Jan 2018, 11:33

That's a way to incite new players to keep their AI size reasonable and reduce the chance they eventually realize they went overboard and quit the game by lack of motivation to optimize or restart from scratch.
That's also, I think, an interesting constraint for advanced players that could see it as an additional layer of difficulty and a new opportunity to be creative with optimizations.

Also, I'm currently working on AI edition/debugging improvements.

sollniss
Automaton
Automaton
Posts: 158

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#10 » 18 Jan 2018, 12:35

Honestly, this will have the reverse effect.

In programming we learn to be modular. You have one class that does one thing, and one class that does another thing. Separation of concerns.

What this change is promoting is the opposit. Why would I have a Defense module and an Attack module when I can get to way less nodes if I mash them together?

Same way that specialized modules now become bad design. Before this change I would have a 1v1 module, I would have a module for specific situations.

Now I'm forced to remove these and throw everything into one pot. Basically you are telling us "please use spaghetti code, but make it short".

Node count isn't the only way to judge if an AI is good. I've seen some peoples AIs and they look like a pile of garbage. People don't use modules and put everything into one AI, thats why their AIs get messy and they quit. Not because they used 200 node instead of 100.

Also, no, it's not an "interesting constraint for advanced players". It's a handycap, you are forcing us the make our AIs worse.

User avatar
GFX47
Dev Team
Posts: 2416

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#11 » 18 Jan 2018, 12:49

Only condition and action nodes are counted.

sollniss
Automaton
Automaton
Posts: 158

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#12 » 18 Jan 2018, 18:17

My modules contain other modules which again contain other modules. This adds up to a lot of nodes inside the main module.

Each module itself doesn't contain more than 20 nodes.

mcompany
Autonomous Entity
Autonomous Entity
Posts: 857

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#13 » 18 Jan 2018, 18:38

GFX47 wrote:That's a way to incite new players to keep their AI size reasonable and reduce the chance they eventually realize they went overboard and quit the game by lack of motivation to optimize or restart from scratch.

Hmm... I'm not exactly sure that you will quite get results like this. Simply put, I feel like this will put a level of optimization that will have a few ridiculously dumb solutions that makes a new rock-paper-scissors simply because the fastest actions can't understand why it makes those actions. Also, I believe there will be a massive loss in the amount of difference between one strategy to the next

mcompany
Autonomous Entity
Autonomous Entity
Posts: 857

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#14 » 18 Jan 2018, 18:40

Also, this is still a question
mcompany wrote:One of those trello cards implies a cap to the amount of lag you can get. What is that cap?

User avatar
GFX47
Dev Team
Posts: 2416

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#15 » 18 Jan 2018, 18:51

There's no cap for now.

Jotunn
Script
Script
Posts: 21

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#16 » 21 Jan 2018, 01:18

I'm against this change. It'll kill my main ai. I guess simplebot will return to being everyone's main ai.

User avatar
LuBeNo
Autonomous Entity
Autonomous Entity
Posts: 531

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#17 » 21 Jan 2018, 08:14

I also don't know about this feature. This behavior would prefer writing an AI for each bot, because each bot specific node is validated false for the other bots (and therefore would delay execution). Perhaps bot specific code should not count to the number of nodes, because it could be precompiled.

If implemented the delay has to be visualized well.

Also a good-idea-AI will probably loose against an fast-executed-dumb-AI, because of this. So you would loose players that are depressed by the fact their AI is smarter, but killed easier.

IMO these values are way to high:

Code: Select all

Assault:     250ms -> 200ms + 10ms per checked node
Shotgun:     250ms -> 150ms + 10ms per checked node
Machine Gun: 250ms -> 250ms + 10ms per checked node
Sniper:      250ms -> 250ms + 10ms per checked node
Image
My algorithm of life: if(self.tired) sleep(); else if(self.hungry) eat(); else follow(Jesus);

User avatar
Kanishka
Skynet
Skynet
Posts: 1417
Contact:

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#18 » 24 Jan 2018, 17:41

GFX47 wrote:That's a way to incite new players to keep their AI size reasonable and reduce the chance they eventually realize they went overboard and quit the game by lack of motivation to optimize or restart from scratch.


It is not. Don't.
As a separate mode though, why not? I mean hey we could test a prototype, see how terrible it is, and then decide what to do.
Fixes break an AI more than bugs do. :ugeek:

Gladiabots Off-Topic Chat


My Stats: Kanishka_RN3;Kanishka_MiPad

User avatar
GFX47
Dev Team
Posts: 2416

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#19 » 24 Jan 2018, 18:15

I postponed it for now.

odomobo
Hello World
Hello World
Posts: 6

Re: Node Reactivity Depth Discussion(+10ms per node)

Post#20 » 19 Feb 2018, 17:44

I just got the game a few days ago, so take this for what it is, but...

Please don't implement this. I think it can only be harmful from a gameplay perspective. I have many thoughts related to this, but the main reason is that it's important to strictly reward players for making a more intelligent AI. If people want to hear my other thoughts, I'd gladly post them.

From a technical perspective, I think the best solution would be to place a node cap on each bot for ranked, unranked, and tournaments. This should be as high as as possible -- any gameplay-related node restrictions should be separate from this. To determine what the node cap should be, take the slowest system you plan on supporting and see how many nodes it can reasonably run. If you do this, make sure you have features which allow players to work around this limitation. For example, a total node count in the AI editor, which can be enabled in the settings, and a warning anytime a user submits a bot with anywhere near the node cap.

Return to “General Discussions”

Who is online

Users browsing this forum: No registered users and 2 guests