Forums

It's a design. If you can't win you are too slow, no cheating involved.

L2speak for yourself, buddy.

I love trials

same here

Pretty sure they’ve dumbed the standard of the drivatars down. Feels more like expert than unbeatable. In single player at Ambleside, I generally catch the lead drivatar in the last 20%. In the trial, even if I start last, I catch them all very quickly. Basically, you need to be running 1.55 in the Hoonigan to win in single player. 2.00 is easily good enough for the trial.

I feel the same. Unbeatable is sometimes unbeatable in SP but it's easier in trial. It could be because of all the network stuff.

Drivatar on Expert is not as difficult as in Trial. With proper tune, Expert is a breeze. With Trial it's only easy if you arrive in 6 so you have cooperative teammates (no infighting).

But most people doing Cross Country can't even handle Expert, and Cross Country is notorious for being more difficult than other disciplines. If AI gets a good lead it's very hard to catch because they're unfazed by obstacles. Even with best car and tune the AI simply has more speed on the fastest CC tracks.

Either way, this game has similar rubberband boost to competitor games like Need for Speed, so the difficulty is artificial even at lower levels. On Expert the lead sometimes behaves as an Unbeatable CPU.

Edited by user Saturday, May 16, 2020 2:37:24 PM(UTC)  | Reason: Not specified

Every season, both the Trial and the Playground events should be just to participate, not to win.

This weeks rally trial I am certain has a extremely low win percentage. Every week this is the always the most tedious and some time brutal event to win e.g. Spring time Easter bunny A700 Buggie race.

This weeks trial is especially brutal. Why?

1. People show up with the rental option instead of a tuned car
2. The drivatars are like M1 tanks and if they hit you (like the enjoy doing) go flying across the track
3. Even if you are a rockstar player with the best tune in existence, one mistake and there is no way to make up the time or position.
4. The number of people that drop of of the races I am sure are high

As for the playground event?

I think this one is obvious. Most teams when they are down two games to zero do not want to continue playing so they drop.

I cannot think why winning this every week makes the game in any more enjoyable or challenging. The monthly online racing is something you have to participate in for the season to complete it, but you do not have to win. Besides, if someone finds these online events enjoyable, they can continue with their online career.

Every season, both the Trial and the Playground events should be just to participate, not to win.

This weeks rally trial I am certain has a extremely low win percentage. Every week this is the always the most tedious and some time brutal event to win e.g. Spring time Easter bunny A700 Buggie race.

This weeks trial is especially brutal. Why?

1. People show up with the rental option instead of a tuned car
2. The drivatars are like M1 tanks and if they hit you (like the enjoy doing) go flying across the track
3. Even if you are a rockstar player with the best tune in existence, one mistake and there is no way to make up the time or position.
4. The number of people that drop of of the races I am sure are high

As for the playground event?

I think this one is obvious. Most teams when they are down two games to zero do not want to continue playing so they drop.

I cannot think why winning this every week makes the game in any more enjoyable or challenging. The monthly online racing is something you have to participate in for the season to complete it, but you do not have to win. Besides, if someone finds these online events enjoyable, they can continue with their online career.

Again, the issue is not if I can win against them or not. The issue is that I've played the trial 4 times, gotten a podium position in all but one of the races, and lost every time because the other 5 drivers ALL finished in last.

Also I hate situations where I have to bring a certain car. I bought into the Horizon series for the freedom. While I enjoy the seasonal playlists gently encouraging me out of my comfort zone, when there is only one or two viable options to complete an event (especially an event with an open theme like this one) it grinds my gears. Meta cars are for rivals and ranked adventure. I should be able to bring any S1 900 built rally monster to the trial and be competitive.

I’d say you’ve been unlucky there as a lot of the random teams I’ve been on have had a couple of players who can beat a few drivatars and, if one or two of the others quit, that’s usually enough. I’ve done the trial 20-25 times and my teams have probably won about half.

This week, the prize is a car which can be acquired for 50k in the AH. Basically, there is no reason for anyone to do it if they don’t want to. In those circumstances, what’s wrong with making it a bit harder for those who want it to be a challenge?

Also, if you do want to get a win and are struggling for teams, just ask on the main weekly thread. Ziperr and others have been offering to convoy regularly for anyone struggling or get this done.

Edited by user Sunday, May 17, 2020 12:27:16 PM(UTC)  | Reason: Not specified

I disagree with just about all the above, sorry.

Don't show up in a tuned car, or at least, make sure you keep the PI down. The drivatars are much more beatable at 850-880 than they are at 900. The other advantage is you might be going slower, which is key in this one. Too many people are going fast and not getting around the corners. It's easy to make up time and positions after mistakes, but just take it easy. Finally, dropping is good - you won't count towards the drivatar's team points if you drop.

That's generous of them, but I'm more commenting as a criticism of game design, not out of frustration of not completing the event. I don't need it, like you say.

I like getting challenged, that's fine, but I hate getting cheated with a passion. The AI isn't winning due to superior driving or better tuning. They're winning because the game gives them boosts and invisible advantages. I'd rather that the AI get a PI advantage to raise the difficulty then to see them outgrip me in a heavier build with skinnier tires or blow away a hay bail without any slowdown. Add that frustration on top of seeing all of my teammates come in last 4 trials in a row, and you can see why I think this week's decision about the trial was a poor one.

It depends what you mean by cheated though. I agree with you that the fact that the AI phases through objects if they are a certain distance in front or behind you is very poor game design. That is a real flaw as it can, particularly for lightweight cars like the buggies, mean that they do impossible times. It’s not really an issue on Courses 1 and 2 for this trial though.

I’m slightly less fussed if they are doing relatively realistic times in an unrealistic manner. Generally, in this trial, assuming I can catch them near the start of the last race to stop them getting a big advantage on the straight section with loads of debris, they seem to be setting tough times but ones which don’t seem impossible for someone of moderate skill to manage with a little practice.

I have been very careful to exclude Cross Country from my argument that Drivatars don't "cheat" as defined having total unfair advantage over player for very reason, that CC difficulty isn't consistent with Dirt, Road or Street racing events.

Passing through objects, or flying, ignoring all physical limitations (didn't see that one on this CC Trial though) is definitively having unfair advantage over player. There's also rubber band issue in single player (has also been relevant on some Trials) where Drivatars get additional speed boost after passing outside of player vision range. That is most notable on circuit type races.



It's poor game design and it's very difficult to build a case to see that as not cheating especially when rubber band doesn't compensate for player advantage at all.

Definition is important here. Horizon AI cheats all the time, but excluding some isolated cases that can be classified as bugs, it cheats for players favour as much tries to pose challenge. Rubber band compensates for player car PI (but some other features as well, I don't know formula for that) but also things that Drivatars AI can't do, weight transfer, power gliding, alternate racing line, etc. except for CC where it appears to have advantages but very little of compensation excluding poor cornering (that probably isn't intended as compensation). It appears that haybales can really slow Drivatars down, unlike other objects, but in CC, they have really weird ability to catch up even from that.

While it's a all abstraction, Drivatars trying to attaining whatever time targets, there's limit where abstraction can get too far and just doesn't contribute to great Horizon experience, I'd dare to say for what most players are looking for, CC is broken. There's useful window of opportunity at the moment to gather statistics by comparing data, say if there's overlap with group that invest Horizon points to wheelspins and those who invest lot's of time for this sort of Trial. Perhaps to make something that works for next game, I hope but...

Silliest thing is that they could just add another optional playlist challenge, maybe once a moth. Dirt racing, or Street Scene Road wet, actually Unbeatable Drivatars and Drivatar vs. Player team PI matching set to Off. Would contribute a lot of players getting better with finding right braking points, better corner exits, becoming better tuners and in general better players. None of this CC garbage is needed to make that happen.

Turning up in a stock car at these types of races is just plain stupid. If you can handle an uncontrollable car for several laps because you wanted lower PI for the drivatars then good luck to you (and i mean lots of good luck).

Rally cars may do well on the dirt but put some grass in between and it's like going to the ice skating ring (not to mention obstacles and water crossings).

Generally for most players if they miss a checkpoint it is not easy to make up time as the drivatars just leave them in the dust and you generally need more than 3 laps to catch them from last. (Experienced players can catch them but they are usually in decent well tuned cars).

Dropping is never good in my books. You can have 3 players and be leading but that matters for nothing if the other 2 are taking up the last 2 positions. Anyway, it' supposed to be a team event, not single player.

This weeks Trial was a bit frustrating but doable. I won't be doing for kicks though.

Although I didn't enjoy this week's trial I will say one good thing about it, it does a great job of illustrating who drives and who just holds RT or LT all the way down with no in-between.

In terms of them doing realistic times in an unrealistic manner, yes, it doesn't really matter all that much in some ways, but people do prefer to feel that AI isn't cheating at all. A true non-cheating AI runs through the same input layer as the user, and is only permitted to provide inputs at the same rate as a user can. So, for example, a non-cheating AI that acts via the controller input layer shouldn't be able to flick the stick from one extreme to another in less time than a person can physically do that. They use this principle in competitions for AIs in RTS games, where the AIs are only permitted to perform actions at a certain rate that matches what top humans can do.

The AoE II AI has always been very good, even back in the late 90s when the game was originally released, but it did cheat and people complained about it. With the recent release of the definitive edition, they overhauled the AI, and they were able to make it play at a higher standard, while not using any information a human wouldn't have access to, and only performing actions a human can perform. I'd argue it still cheats in that it plays in a way that is impossible for a human to copy, micro-managing different units at the same time in a way that no human can.

Anyway, I just thought it was an interesting and relevant example of how it is seen as important in games in general to have AI that doesn't cheat, and that plays in the same manner as a human. It's certainly an area where Forza has room for improvement in future games.

Oh absolutely. I think I agree with all of that post Bree. The drivatars leave quite a bit to be desired generally in Forza. I’m all for more realism but that isn’t going to change in this iteration. I’m therefore looking at the secondary issue of whether it’s a fair test and, in my view, it is this week.

Anyway, I think I’m in a small minority of enjoying the trial this week 😄. B class dirt racing in Winter from Thursday, so that should be a little easier.

Stock cars are often very controllable. For this on, try the Honda Civic Coupe GRC - it's really stable and I'm usually winning or finishing in the two two or three in this trial with it.

True, it is very slippy - hence take it easy. I am almost never on full throttle. It's frustrating seeing other players come past me full bore only to go skidding off at the corners/bumps while I'm tracking the racing line fine.

Even in a stock car it's possible to come back to a top 4 or so finish if you miss a checkpoint - as long as there aren't too many 900 rated cars. That's really the key.

[/quote]Dropping is never good in my books. You can have 3 players and be leading but that matters for nothing if the other 2 are taking up the last 2 positions. [/quote] Yet if those two dropped you'd end up scoring only the top three and would be fine.

For all of the frustrations (hell is other players etc.), that's very true.

lol!!!!! I'll pay that.

OK, it's less consistent but still very same.

I don't see any reason why AI cars should care about physics or objects if even in road racing they can go much faster than possible. It's the same principle. It's not cheating, just design how to look like your competition. AI always starts with faster cars and ends with slower cars. It's not important what happens in between. CC could have bad time settings for AI so they are faster than usual and sometimes completely broken. This trial was fine, you need to be fast, but not unrealistically fast if even I can do it.

Consoles have very cheap CPU and it's very funny to think that game can simulate 12 AI cars with everything real. Maybe in 20 years, not now. But it's pretty good for playing if you are used to... and it's not cheating. It's complete dream world with different rules.

Edited by user Tuesday, May 19, 2020 1:13:36 AM(UTC)  | Reason: Not specified

I'm not sure if your argument supports your point and conclusion supports neither. So Drivatars don't cheat and even if they pass through objects it's not cheating because Xbox fairy told me so and in the end it doesn't matter because I liked the event. LOL

What we are really talking about is time bracket set for Drivatars in CC, let's see about that:

While first race was easy, take Ambleside Rush, time needed to win is about 2.0 - 2.05 that is about there with top 10% of Ambleside Rush rivals.

With all 2.951 times there are set really even isn't a total ghost town. It's notable that only Rally Monster in top 100 is Jezza's Ford Capri FE. Then there's no rain in Rivals. That's the reality of it. You may perhaps feeling modest about your skill or being delusional, it doesn't matter, 10% isn't the same as average. I gave it this event 11+ goes and we won only 2 times, that's way below how these usually goes.

It's all right to like the event, I just don't understand why there's need to make excuse for bad design choices or unfinished feature at the same time. Frankly I really liked to see a bit more challenging events just in other race types. I just don't see it working with what most players expect from Trial.

I don't mean as car class thing limiting reaction window like S1+, but less restricted AI like we had on Trial couple of months or so ago:
Rally event, 1. Derwen **** er Trail, 2. Mortimer's Garden's Scramble and 3rd race was Glen Rannoch Trail. Drivatar PI matching vs Player team was set of Off.

It got a lot of hate but I think it was one of the better events ever. Drivatars in that race were really good at braking and clean exists that of course reflected to overall speed, no cheating or super computer needed. When races were lost, it was because player team within racing and trying tricks like wall bouncing, and stuff, that usually work, but are indeed slower techniques vs Drivatars that were racing properly.



I used to say that thing for moths in other topic, but I really don't know what limiting time bracket with ~3 seconds has to do with computing power. Unless you meant what Breeminator wrote about. I don't know, that ~3 seconds might help with to maintain suspension of belief too.

This being Horizon, doesn't make much sense to call for more realism. The biggest issue with the AI is the implementation of routines made to force you to lose. Sometimes you're in the lead, the cars between 3rd-12th are far behind, but 2nd place tails you as if a magnet was strapped to its front bumper. Reminds me of NFS and these are bad memories.

The game also tries to force you to move up in difficulty because you're "winning too much". Perhaps not coincidentally, magnetic Drivatar gains a stronger pull after you decline. Who cares if I want to play on an easier difficulty? It's not like I need to play against harder AI to hone my driving skills, as these are linked to the clock, not to opponent difficulty. Maybe my racecraft, but still.

Speaking of Trial, I've played Rally Monster S1900 CC in the past (both FH3 and 4) and didn't enjoy it very much. It's very imbalanced, due to different tire quality between cars. The Hoonigan RS200 is a better car because it's born better, but the Capri as an old chassis can only hang with the big boys because of its offroad tires. Wet mud is a hindrance to the cars with worse tires.

Unfortunately this category lacks the Delta S4 which was demoted to Retro Rally but can rise to S1900 with bodykit and bigger turbos. It's a tricky car, but its high acceleration could make it competitive in theory.

Edited by user Tuesday, May 19, 2020 1:17:12 PM(UTC)  | Reason: Not specified

issue now is i notice same bug and such over and over. for over 3 months now all easily replaceable. their not fixing it at all.

Drivatars don't care about obstacles because of "fixed" pace. If they crash they will be behind you later even if it's not possible. The problem with CC looks like bad pace configurator. It's probably dynamic because of custom tracks and it's harder than road racing. But still the same and no cheating for me.

It's not about my skill, many people here are better and I can't do CC too much. And trials are easier than SP so it could play a role too.

I meant it's not AI what many ppl think. Computing of physics is very limited and mostly not counted at all. Still, it looks like good AI to play with if you agree with the rules.