Yeah that would be nice for the weaker player, to be able to practice in normal resource conditions. It would be easy to implement lower health vills, lower collection rates, less building work rate and HP. Also throw off the build order for the more experienced player to level the playing field more.
The tricky part is the bonus damage multiplier. I really like the way this was implemented in the handicap system, extra rewarding to making counter units. It doesn’t really work in reverse though. Would a player with 10% handicap even bother with counter units then? Halberdiers would do 3 bonus damage to cavalry, and Elite Skirms 0.4 bonus damage to archers. I don’t see an easy answer to this problem.
I recently tried 15 % handicap for the hardest (ie extreme) AI against me.
My attempt was to use Poles greedyly and defend with towers in feudal. Then to use their insane knight spam and raid the opponent to death.
My opponent was magyars (randomly chosen).
I made several mistakes like letting the first waves of enemy scouts in cause i didn’t scouted my woodline was open to the edge of the map, but saved all my vills and forced them back with spears. But that play heavily delayed my castle uptime, I needed to sell a lot of my stone in the end.
Also the knight raids didn’t worked as the opponent made a lot of pikes and TCs very close to each other and all over the place. In the end obuch + bbc was the comp working as the opponent made the right switch to cav archers + magyar hussar but never got the numbers up needed so I slowly pushed his base and got the V at almost exactly 1 h in the game. 10-15 minutes later and I would have been out of gold (yes, with poles +3 relics).
That was really challenging I must admit. I think the AI made a lot of strategic misplays, but it was much more difficult in the strategic part than I expected. I don’t tink I could have managed this with any other civ than poles and their insane eco bonusses.
I don’t think I could do 20 % handicap even if I would play flawlessly. That one was already tough enough for me.
But then I would need to taunt them to do this.
My test against 4 ais was without taunts.
I also chose persians to maybe see war ele + trashbow combo but it didn’t even make the techs for that. Nevertheless persian was a good choice cause of the faster working TCs - it took out 2 opponents fast enough so it could deal easier with the remaining 2.
I think that’s the key to have a civ with an insane early eco bonus that alows to kill several opponents fast enough while also maintain a strong enough eco to don’t fall behind the remaining civs.
And I don’t think bohemaians fit that category.
Another Idea could be a civ like malians with a high unit variety and several units with high bonus damage Also an option could also be vietnamese so the ai doesn’t wastes any time with it’s attacks looking for the enemy.
Vietnamese have always been a good civ for the AI to play. Bohemians could actually be ok, with the fast Imp times due to the age up and gunpowder. Also good archers, and even more bonus damage. Goths could be really good, because it can reach the lategame a lot faster, and then just spam infantry really, really fast for ages. Turks might also be ok because of the fast Imp gunpowder, or the scout pressure.
All lost in about 55 minutes. Usually thy beat 2 opponents and then get slaughtered by the remaining 3. The best shot I saw so far was Incas which almost beat 3 opponents but then lost in a basetrade.
A little bit disappointing was the cumans game as the AI didn’t made the 2nd TC i hoped for, Maybe I will give it another shot until the AI goes for that attempt.
I will tell the stories of the games at some point, so we can get an Idea what happend there and probably get an Idea how the AI works and can probably chose a civ that can do it.
Cause I think the AI makes a lot of silly mistakes, if it wouldn’t make those it could easily win with any civ imo. But maybe a good civ choice would lead to the AI making less of these mistakes and then winning. The key is that the AI needs to take out all but 2 opponents before they make a coordinated attack.
Goths were an interesting one. They didn’t even killed 2 opponents but they holded the longest: 59 minutes.
I didn’t pay much attention to the handicap system when it was introduced, but after seeing T90’s recent video of him beating the Extreme AI with 160% handicap, I realised it could allow me to play 1v1 against a friend who has barely played the game.
I tried playing against Extreme AI first, with me having a 200% “handicap” i.e. boost, just to get a feel for it, and 200% is very very strong, as you can have a mass of fully upgraded crossbows before the opponent can possibly have reached castle age, which obviously is very hard to counter.
Playing against my friend, with him boosted to 200% (we both feel it should be called boost not handicap, as it’s the opposite of a handicap), he couldn’t beat me, but we played 4 games, and he did better each game. In the 4th game, he reached castle age before me, and was ahead of me in score, and I was only able to win because I knew what he was going to do, and countered it. Even then, the timeline showed he had enough military early enough that he could have killed me if he’d attacked with it. I think a few more games of practice for him and I’d have no chance. The 200% player can build castles at a rate that means you can’t do much without imp, but by the time you get to imp he’s going to be miles ahead, even with very suboptimal play.
Overall, I feel this is a great feature for allowing enthusiasts of the game to play against much less skilled players. It’s certainly an issue that an optimal build order differs so significantly, and I did explain to my friend that the numbers he’d need on each resource at each point in time was very different to what he’d need playing without the handicap. One way to address this would be to make it from 75% to 150% instead of 100% to 200% (150% still being double 75%), so both players have their build orders messed up. So if a 1400 player were playing against TheViper, for example, they could try 95% and 110%.
Well this is a problem caused by the nature of rating systems. high level players are always “underrated” and low elos “overrated” the more you go to the edge of the scale.
Important is that we have something reliable for mid elo, cause that’s the majority of the playerbase. Something that holds in like 800-1600 elo is good enough imo. At high elo players will figure out their state individually anyways.