Moral support thread for our overlord...ehm dev!

[pier4r]

An advantage with the “ghosts” is that one gets also a feeling how good a previous release would be (if the ghost loses or gains points). For example Miojo has released a ghost that had around 2200+ at the start and now is about 2100.

That is informative.

I still wait for the GFX answer, how the ghost are picked/deployed. It is every deployment a new ghost?

[pier4r]

And from telegram

"Anyway that is the fun. At first I missed the elimination maps, but then I have to say that resource maps are cool because one can play the elimination first and it is quite hard since one has to bring death on the enemy swiftly."

[GFX47]

I still wait for the GFX answer, how the ghost are picked/deployed. It is every deployment a new ghost?

I think I already answered on how they get picked: basically the same rules as before but applied to past voluntary deployments.
Every voluntary deployment is a potential new pickage ghost.

[pier4r]

The fact that bots do not explode anymore at the end of the map is pretty intriguing! As I had many games where I would have been in numerical superiority having pushed the enemy on the border, now the victim lures my bot away and the rest of the enemy team scores a victory!

[pier4r]

Anyway I was not so sure about the ghost thing. It is awesome. Playing against the entire spectrum of the matchmaking range is great for variety and challenge.

Playing with "still valid" (for the resolution engine) strong and inactive players (Miojo, DH, milkhunter and others) is great for benchmarks.

I really like it and I would love to know the details. Namely: how an opponent is chosen and , between the available ghosts of the opponent, how is the choice made for the match.

[pier4r]

The ghosts have also an high psychological value! (especially for the ones that care about points in the ladder)

I am not sure how they are picked, GFX hopefully will explain the new matchmaking soon. Anyway if someone is playing a little but his score is high, the other players can recognize whether is score is well deserved or not, as they play against the ghost that has its own score. If the ghost keeps an high score, then the player at the top didn't have just a lucky streak of wins.

For example Miojo "compiled" the miojo ghost for the alpha 12.x recently but quite "old" in gladiabots terms. On the 10th of February was the last voluntary deployment. Well, after 15 days the ghost is still floating around 2100.

That is great, because the others know how strong is really the last deployment of Miojo and Miojo himself knows that if he would play again, without much changes, his score would be around 2100. So the expectations are set for the player and his opponents.

Awesome.

In this very moment my score is 2184 and I am 1st, the second one has 2153, but I know that my ghost, that hopefully is using my last deployment, is around 2100 and in particular I am already expecting that meeting mumpsimus will be painful (but data! From failures one can get lessons learned).


So a lot of "what if" expectations are well solved that would prevent people to play to keep their score (since they know how is going on) and people not at the top to wonder whether the top deserve their place or not.

Awesome!

[pier4r]

Now since alpha 12.x the "closest enemy at distance exists" , that before ensured that the closest enemy was at distance X and not less, is changed with a simple "enemy at distance exists". So I have to do a lot of work to fix the nodes that otherwise may be active when they should.

In this case the new favourites, where one can really define his favourites, are great. They simplify the work pretty a lot.

Actually I am not sure if it would be better to convert them in "myself" distance from closest enemy.
Tested. Myself closest enemy works. Hopefully this does not get changed.

[pier4r]

The ghosts have also an high psychological value! (especially for the ones that care about points in the ladder)

....

For example Miojo "compiled" the miojo ghost for the alpha 12.x recently but quite "old" in gladiabots terms. On the 10th of February was the last voluntary deployment. Well, after 15 days the ghost is still floating around 2100.

Observing the ghosts behavior it is also true that the ghosts play mostly against those other players that play often. If one does not play that often (say, under 20 games a week), it takes long time before the ghosts played enough random maps to determine a sort of a trend.

With random maps (that are not extremely random, but I find them very well done) one has to play against other players many times before one could say "yes, I mostly dominate/lose". And the counter is almost set to zero again every time one (or the opponent) makes a change.

This means: please play, play, play.
This also mean that "old" ghosts like miojo, floating between 2050-2100 are really amazing.

[pier4r]

The more I play and I analyze how the ghost works, the more I like it, also when the details are not yet clear.

I was really for a ghost expiration: I thought after 10 days the ghosts can go, but with the current algorithm (if I got the gist of it ) it is even more interesting as one can use specific ghosts as benchmarks. At least until they get invalidated by a new resolution engine.

For example in January 2017 Arcueid created a "total harversting" tactic for Set your priorities that was highly effective, with over 80% win ratio in Grandmaster (at that time me, nullpointer and tct were the strongest guy).
Then Arcueid went inactive in February 2017.

When he was back in May 2017, his tactic suprised everyone, even if the game resolution engine did not change (it was always the same alpha 7 version). Except for those that kept in the code the "anti total harvest" strategies, all the others evolved the code and lost the counter tactic.

Instead if the Arcueid ghost had been there, everyone would have been reminded to keep in the AI also the necessity of an anti total harvest strategy.

In this case Vectorpleximus (floating around 1900), Miojo (around 2050), Dh (around 2020) are not playing for real, but their AI is there still able to find refined weakness in the code of the active players.

It is really interesting. So the more (strong) players "compile" their AI for a resolution engine, the more a player, even alone (like me yesterday evening) can enjoy the variety trying to dominate those ghosts. That is far from easy.

[pier4r]

The new random maps are pretty interesting. Especially because there is always the harvesting factor in them. I try mostly to push but after enough "losses" because the enemy defended well, I pushed him but not fast enough, I have to consider harvesting.

it is another layer of complexity in the tactic to keep in mind.

Update: moreover the random maps offers also the "kill the all" mechanics. When the ally bases are deep on the enemy side. So, like a pun, the current random maps covers all the bases.

[pier4r]

The more I play the more I find the combo: random maps + matchmaking with ghosts awesome.

- Before alpha 12, I remember if I wanted to debug a problem I had to debug it right now, even if the situation was not so clear, otherwise I would have needed to wait a lot. Often in slow motion.
Now instead I just continue to deploy until I am "lucky" to find a situation clear enough that I can capture the gist of it even in fast forward. Of course there is still the need of slow motion debugging, but there is also much more data so I do not have to cling to those few games I could play.

- Before alpha 12, some strange tactics that were a pain in the back could have been ignored if the player was not so active. Yes one would lose points, but once in a while (even from top players semi inactive). Instead now the ghosts of every player are there to remember everyone "look, my tactic may be 20 days old, but you should still account for it, otherwise take this burning loss". It is awesome.

- Like the point above, the data from top players semi inactive was scarce, and one could not break their tactic so easily. Now it is plentiful and awesome.

really awesome. I find the AI editor multiple copy/paste/clipboard still the killer feature, but random maps (although not totally random yet) and ghosts are gorgeous.

[Tortuga]

Hello everyone,
I am a teacher and I work with children. I discovered the game, while I was looking for something to teach my students the basics of programming. I usually use Scratch. But immediately get hooked.

I think the game has a lot of potential in didactic, although I see it a bit green to use it with my students. On the one hand, I do not like the aggressiveness and competitiveness that is implicit, but on the other hand I recognize that this is something very common in today's society and that it can serve to make study more attractive. In addition the difficulty of the game, compared with other more typical games, also can discourage to great part of the people, although also it should animate to others.

What I like the most about the game is the creative effort that must be made to move forward, and the fact of having to translate the ideas that come to mind through a visual language that is very accessible for those who have no idea of ​​programming. I think that in its current state it is not the most appropriate method for teaching children programming, although it is a better entertainment than many stupid games that they usually use, and I will recommend to everyone.

For my part, I think I am going to continue playing while my free time allows me.

[pier4r]

Nice to see you here Tortuga! Come on telegram if you can. So we can banter together since at the moment you are farming everyone (acho, Tortuga, zgeneral, streaker).

Link: https://telegram.me/gladiabots_fanbase

[wpisznazwe]

I'd also like to say 'Hello!' to the forum. I was playing the game on Android quite some time ago, then some weeks ago on (broken now) Win7 and now going to start again with a new account on Linux =) My first experience with the game was great, and I see that nothing has changed in that matter - which is obviously good!

Some points:
- Love the game concept and execution so far.
- Love the Trello Roadmap website showing exactly what is going on, voting etc. - customer orientation is so much important.
- Love random map generation - this is a milestone feature.
- Lots of SP matches to debug A.I. - great.
- New interface is much more intuitive - very nice. I would like to see further improvements, maybe some kind of A.I. branches marking (dimming?), which would indicate the condition/action which is impossible to fulfill/act or that there is a preceding action in the tree, which always sets off before the dimmed branch.
- Maybe there could be some more variety in the initial bots positions in the randomly generated maps.
- Far future - maybe more game modes, maybe random quests or something, labirynths with turrets, unlockable gates, more challenges to keep the players busy progressing, idk.
- Account system is a must - waiting for Alpha 13
- I'm not a programmist, but the game 'feels' well optimized and well done overall.

One thing that disappointed me and which I cannot understand for now is how is this possible that the active playerbase is not thousands yet, because the game just deserves more attention.

Anyway I wish you best, curious to see how the game develops in a few years. It is a pleasure to watch the game progress.

[pier4r]

Alpha 13.

- not losing ghosts is pretty neat. Up to 2200 one has 40+ different opponents, awesome.
Although if the resolution engine has massive changes, it would be fairer to avoid keeping "old ghosts". Unless they can be updated. Because buggy ghosts are still a challenge for newcomers but they free tons of points for competitive players (see at this moment lokus and masterplayer). This will happen for every deployment so potentially thousands of points will be injected in the system.
Inflation is expected but it gets ultra quick with some changes (max alpha 12: 2400, max alpha 13: 2700 ). Maybe old competitive ghost could be somehow redeployed and fixed by gfx itself if he has time?

- I love the new editor. Yes it expandes everything at once, but indeed then one knows where things are and can quickly scroll here and there.

- great fix on the sg bug. It is a joy to farm those pesky lame sg user. Although some of them are still quite competitive as they capture things with the right timing.

- the matchmaking is still a tad buggy (private messages incoming) but it got a bit better. The variety is large.

- the skins are great, although I don't use them. That could be a source of income. 8although I still think that tokens to play or test is better).

- For the moment that's it. When I spot other nice changes or introduction I will tell it.

[pier4r]

Oh yes, the ability to edit the AI directly from the replay is awesome.

[ElvenMonk]

Updating ghosts only after victories could help.

[darktheatre]

version 13 most awesome,more multiplayer enjoyable

[pier4r]

I am not sure if I wrote it before, but knowing the "played hours" that are fairly accurate (at least for my tests) is quite nice. I was expecting it was scaring people.

[pier4r]

Not yet dived in alpha 14.2 anyway the ghost matches are cool! Especially to see ghosts that didn't get any update since long time increasing or decreasing their score (a good indicator of redistribution of points. The same AI is valued more in terms of score).

For example Miojo or also others.

Not all the ghosts gets more points (due to redistribution) though. See for example the attachment about zgeneral ghost.