The Extreme AI's Fast Imperial Strategy 

When I shift into second gear in my truck I often find myself thinking “okay, Feudal Age here we go!”

Spirit of the Law drag racer confirmed.

Italians fights Turks hand cannoneers with Genoese Crossbowmen, then switches to skirmishers Italians AI "I think I forgot something?" "If you forgot, then it wasn't important" "Yeah, you are right" [Condottiero under the rain]

A capped ram push of just capped rams.

AI: Clicks up to Imp at 20 mins. Me: Clicks up to feudal at 23 minutes with 35 vils

he did it perfectly. and then he did what AIs always do and lost all his army to static defense completely unnecessarily. some things never change.

Spirit of the Law: uploads Me: Time to stop rewatching Spirit of the Law, and start watching Spirit of the Law

thats savage af, red deleted all his walls to troll teal.

Mole is soooooo out, every serious scientist uses Steppelancer as the new unit of quantity!

17:24 just a villager casually repairing *the map.*

your gear shifting analogy is... so spot on. perfect.

AI always deletes its walls on Arena once it expands beyond the initial base. It also builds really crowded bases so I imagine this is to ease pathfinding.

What I want in the next patch is AIs trash talking to each other.

At 17:12 Teal AI overchopped the woodline and opened a hole in their walls which Red exploited at 17:40 hahahaha

Only spirit of the law would measure the total cost of age advances using (clearly the SI unit for this magnitude) Steppe Lancers

Instructions unclear; got into street racing, now I live my life a quarter mile at a time.

Ah, yes, Fast Imp into throwing away the army to a castle

It's nice to see how an AI acts, used to do it 20 years ago (it could hardly handle a wall). Could you do one on an AI with no gold? Or little starting food or wood around, to see how it manages different resource shortages

Interesting how the AI has improved so much and can easily defeat a new player and yet it still makes mistake not even the newest of players would do, like going for the gold in the middle of the map or building outside your walls when you're being attacked. Goes to show there's still a ton of room for improvement.

Nice video Spirit. Also like how the monastery with relics for the teal player were at the back of his base. Small things but good improvements.

Always nice to see a Spirit of the Law video in the morning!

the current AI wont be able to pull off a successful fast imp. It just struggles way too much against walls/fortifications, ie it doesnt focus down a gate/one section of wall, it hits all the sections at the same time, or gets distracted by a unit inside the wall and runs into towers/castle fire. And speaking of castle fire, the AI just doesnt know how to deal with it. Instead of sending 2/3 BBCs to fire at it from max range, it runs hand cannons into its range, like it forgets that its there. Really missing Barbarian AI right now

Hey Spirit!! I saw your video about what to do with the first 15 vils in a generic map. Watching this video made me thinl about how cool it would be if your made more play-oriented videos, since the one I mentioned was actually top quality content!

Glad to see the ai is learning let’s hope it watches the rec and learns from its mistakes

I love SotL gameplay commentary, whether it's AI, T90 style, or just playing. The combination of micro narratives and describing what's going on on screen makes it incredibly entertaining and professional feeling!

What i miss in age of empires 2 is personalities in age of mythology when you selected a AI in random map you select a personality either RUSH, BOOM , DEFENDER . Rush atacks early Boom atacks mid game has al ot of economy and tries to atack in mid game Defender atacks late game. and usese a lot of towers and defensive buildings.

I know I'm getting annoying, but i hold it, I miss the guitar intro 😢 still keep up the good work m8

If possible can you someday invite Prome to watch an AI game with you. With his insight there could be a ton of fun things to learn about it.

One thing that i started to notice, is that the ai that occupy your spot in the game, lose almost all the time. Even in teams. It´s like the ai´s that are at your side, will always be worst than the ai´s against you and your team. Make sense, but makes the games one sided when you want to try to improve your playing watching the ia´s.

1:20 Incidentally, look at that marvellous DE gen. Red has a spare gold and stone outside his wall, whereas teal's is neatly tucked away behind the wall. 8:38 it gets impatient and buys the food :D 10:40 castle is only well placed if your opponent doesn't do fast imp... or does fast imp but is teal 17:07 if an AI signals in the forest, does it make a sound? 18:29 the farm micro is next-level

portuguese fast imp is also pretty popular for access to arquebus and the feitoria boom allowing you to pop up fast after reaching imp

The gearing analogy of faster age up makes perfect sense

This leads to the question: which civ has the best fast Imp? Turks, Italians, Byzantines, Vikings, Malay?

Thanks for uploading --I too was curious what the AI would do when I read that it is supposedly capable of "Fast Imperial". Dark age farms are pretty awful but in this case, it needed 6-8 of them to get a decent time, 21-22min. When I do this as Byzantines I can generally get 20-21min but then have to research chemistry afterward. A bombard cannon should be queued ASAP as this is what pressures the opponent's buildings, but it didn't do this really at all. As you noted, there is a lot of opportunity to improve this build but it's mostly in the strategy after imperial is reached.

I proposed putting a version of the AI difficulties on ranked on reddit last week :D Maybe not easy, but hard, hardest and extreme would be pretty interesting. Their names should be changing to avoid people abusing tower rushes etc though.

Teal is losing and thinks to himself: "how do I turn this around?" Answer: *USE VILLAGERS TO ATTACK RAMS!"

I'd be interested to see a video expanding on the maths of that point about delaying feudal in order to get faster castle/imperial. Presumably it comes down to getting out villagers earlier rather than researching feudal age. But how long would it take for that effect to get overwhelmed by economy techs or extra town centres?

Oh my god I remember watching this when it came out and you saying "7 on sheep? Thats a little weird" was what alerted me to the fact that I was putting one too many on sheep

That manual transmission analogy is great! Definitely a mindset that I have a hard time fighting.

Thanks for that video

Great analogy with the stick

Going on your most forward gold while you have map control is actually a very clever thing to do and one thing I learned from watching hundreds of viper videos. You can go to your save gold in the back as soon as you're forced to retreat.

The car accelaration metaphor is perfect damn

i stopped paying attention for a sec and then you started talking about engines, thought i accidentally switched to a tavarish video lol

Spirit of the law... Or what helped me cure depression. Cheers for the videos, mate!

what works quite well at my ELO is kind of a semi fast imp with turks. i go for 27+2 castle age with immediate castle drop forward and a monastery with a two monks and the speed upgrade to collect all the relics (i drop them near the castle if the monastery is further away, just so the enemy cant take them until i am ready to carry them back) and then go imp when I can. depending on whether i have to build a few janissaries to kill some enemy troops or not, I reach imp at about 27-30 minutes, but I am already producing janissaries at that point. im sure higher ELO players can beat that quite easily but at my range (currently about 1100ish), I have won 12 out of 13 games in the last few days that way

here law, guys of the spirit Hey

meanwhile, the hard AI rips me apart in less than 30 minutes... btw, i really think that we need to see what is the supposed ELO rating of the different levels of AI. a robot, like Alphastar in Star Craft, would also be fun to play against. The Viper would "enjoy" it quite a lot, I'm sure. PS: How is the Extreme AI working in a team game? what if the Turks are pocket in a 3v3?

getting a few instances of the AI elo rated would be a useful metric, that way elo itself could be compared against a static metric (and all new ais would be on new accounts)

You know what'd be a great update addition? Make it so that "Enable Cheats" gets an Advanced subsection where you can enable only particular cheats, like enabling "going above and beyond" while leaving "how do you turn this on" and "torpedo" off. Maybe even add options to only enable cheats only for certain players, a limited amount of uses for each cheat, and restrictions on them, such as having only one use of "how do you turn this on" and the restrictions that you must be in the Imperial Age and have 5,000 Stone and Gold, which will be consumed upon the use of the cheat.

I believe the AI scouts are trying to explore beyond the trees

RIP the teal monk who died by his own bombard cannon at 13:38.

Hey man are you going to do tatars overview one day ?

Could be an Arena thing, I played vs Extreme Turks yesterday and it did full Feudal archer + spear. It got destroyed with 4 knights as soon as I hit Castle, felt like a 800 elo player.

Did they make the genoese xbow fire as fast in castle as they do in imp?

for the scouts bugging, I think it's just trying to click into the trees, so the scout gets as close as it can to the rally point for a bit and then moves to a new one

Instructions unclear. Revved too much, blew my engine and lost the race

What is a good timing for a fast imperial with a castle? For a goth huscarl flood

I feel the spirit of drew durnil

Extreme AI is way above 1000 ELO. I am sitting at about 920, and Extreme AI demolishes me every time, likely comically bad. The AI is far better at keeping pressure and managing pressure. Basically, I'd say that at 1000 ELO, microing and controlling all troops while also managing economy is really hard. So like the AIs ability to manage eco through a military push puts it a lot ahead of me. The reason I'd say it is far better than 1000 is that me and my friend (who is about 1020) did a 2 people controlling one civ and we still lost really handily to the Extreme AI, and that was mitigated what is probably our biggest weakness compared to AI

Hi Spirit. What do you think of a fast imp with Japanese? Ive only played a handful of online games, but I just tried against Teutons only the other day. Do you think it is a viable strat for katapuruto? If so, how would you do the build? I got to imp in 33 or so mins which is fast for me and ended up winning, but that could be because I was up against another noob player

scouts feel bugged in general including auto scout for the player usually getting stuck in forests

the gears methaphor is pretty good actually

I like how teal started researching elite jannisaries despite not having any of them XD

A 25 vil fc is a little tight. However on a map like bf it is relatively comfortable

16:19 Let me repair this....No, I forget this.... NO, wait, you still need more!

Are you still planning on doing a juressic park video? 100% sure everyone wants to see that 👍🏻

I missed a SoTL approved fast imp build order

In the past, I played a game like Age of Empire that built a Chinese city and at the beginning of the start there was only town hall with a Chinese wall and a game map that was branched out for each player city ‏I forgot the name of the game. Can anyone help me find the game? because I played this game about 10 years ago, and even though I'm looking, I can't find the game.

you should start use the UPSL unit of measurement in your videos all the time! (Units per Steppe Lancer)

What did I learn here... that SOTL street races in his free time 🏁

Going up fast from dark to feudal is good for cumans or?

Okay so what tower is best to build outpost, normal tower, bombard Tower, Castle, or Town Center?

When things goes out of control....I be like How do you turn this on

Spirit, you should do a vídeo comparing Fu SAracen Knights vs fU zealotry saracen camels

Can you make a video about auto-scouting? I was wondering how good it actually is.

At 8:25 does the AI buy 100 food and then do it again moments later? Is that part of the program? Anyone notice this?

What map is this? It looks very cool

next AI strat: fast full maxed out paladins

Hey man, I have a quick question regardless the topic of the video: What program do you use to record?

I lkeep these AI videos alot.

So what is the purpose of going fast imperial? Is it for earlier in the game access to trebuchets and bombard stuff?

Jeez, SoTL, your voice is too buttery smooth. What's your secret?

Is it possible that the AI chose a slightly earlier castle time to make sure it secured the relics?

is there any reason for the fast imp Ai not to use the 200 stone by selling in ton the market?

I'm doing a civ overview booklet (have watched ALOT of SOTL vids) and focusing on Turks at the moment. One of the categories is 2 best teams the Turks sync with well. I'm thinking Berbers and Vietnamese (Potentially Spanish?). Turks get their Teams 20% gunpowder production faster (which Berbers and Vietnamese have a lot of) they in turn give a great trash unit and gold bonuses to the Turks, which is essentially for Turks. Any more suggestions? Thanks. Loving the vids still SOTL!

At 11:11 in the video the red monk picks up a relic and the same instant the blue monk (which should not have vision on the relic as far as it seems to be away from that relic and is walking towards that relic fo far) immediately changes direction as if blue could see the relic being taken by red. Is the line of sight for monks really that far? Or do players, too, see if a relic is picked up by an enemy monk without having any unit for vision near that relic? Or is that a cheating feature of the AI?

Fast imperial is like robbing a bank. You can profit massively but you only have about 3 minutes before the cops show up and wreck your shit

This strategy has the same problem as every other strategy the AI uses - throwing units at a castle. Played against it a couple of times - on both attempts the AI lost 30+ units, mostly hand cannons to castle fire, while I sat behind - boomed and overwhelmed him with units.

3:00 There is one universal truth about game AI: It can not plan ahead! They think entirely in "this moment". But this particular misbehavior should be solveable, by having it advance to the next "step" of the sequence only after it has the right amount of stored resources and/or income and/or villager count.

what about cobra car racing?

Why was the AI garrisoning the Hand Cannoneers in the Archery Range?

I love how pro players can always find minor imperfections in the extreme AI. In reality this AI wipes the floor with most average players

17:27 AI deletes all his walls

It would be interesting to see how Khmer would take advantage of being able to tech up without buildings of that Age. It would save a lot of wood that could be used for more farms

A top 5 of best civ' musics ?

I miss the guitar intro! 😢

My biggest problem with fast imp is how vulnerable you are in castle...this has been a problem for me and I also caught the AI trying it yesterday (as Turks). Eagle rushed him (as Incas) and won easily.

I think the bug occurs when the AI attempts to send the scout into the opponents base but had a pathing error due to the walls. A theory, a shitty one and I don’t know why it would have changed but I believe there were pathing updates.

This leads to the question which AI Civ is better?