Replays, why are they extremly inaccurate

[Funken_A]

I watched a replay of one of my attacks and it was very different then actual attack.. I watched a defense replay and although I lost 500K gold 600K food and 3k oil the replay showed me holding this player to 45% and the player took less then half that loot in the replay.

It is very bizarre

 

[Exiliado]

A replay is not a video recording, just a reproduction based on recorded user inputs. One possibility is that there's a lack of precision in recording user actions at the level of split-seconds that causes a domino effect. It might be easy to rectify (and they just haven't bothered) or it might come at the price of a noticeable delay that would upset many people. Like if you clicked Rally or deployed a plane and there was a split-second delay to ensure that the replay was going to line up perfectly. It would be infuriating if that happened. The replay inaccuracies are also infuriating but not quite so much as would be a delay during the actual battle.

 

[Tsamu]

I have noticed this too. Replays are not complete recordings of a battle. Only the time and place of troop deployments and rallies are recorded. Then the game engine is used to recreate where the troops would have gone, what they would have destroyed, and how the defenses would have reacted. There is a little randomness in the engine, so replays will not be completely faithful.

A major factor affecting the accuracy is incorrect stats. If a bonus is not taken into account for the replay, the attacking troops won't do as much damage as they did during the actual attack. This often happens after an update, and the devs can fix it fairly quickly. Another factor is the hardware you are using versus the hardware of the attacker. If your tablet/phone is slower, the troops will take a fraction of a second longer to make a decision or react to an attack, and over the course of a battle it might make a significant difference.

My feeling is that replays have gotten worse recently, but I haven't seen widespread reports, so it is probably just me.

 

[Funken_A]

definitely good reasoning's to why...

They are pretty useless in the form they are now though. You can't really use them to change your base around because they don't actually depict the weakness exploited. The same for your attacking style. You can't change it because you can't actually see exactly where you went wrong.

 

[Manifesto]

Personally, l usually know exactly where l went wrong.
Those mistakes don't need a replay, my own embarrassment keeps them fresh in my memory.

 

[Manifesto]

I can accept a star's difference in the replay or even a few percentage points, it's the nature of the replay, or should l say re-enactment.
However, showing half the loot stolen highlights a real need to improve things.

 

[LordJames]

I think one of the biggest frustrations with the discrepancy between the replay and the reported battle results for the attack is that it seems to mask the incidence of cheating. About 25% of the time I actually watch a replay happen, it reveals that the attacker did something like use 10 aircraft to attack, and frequently the discrepancy isn't just a star or a bit of loot, it's an entirely different outcome.

I get from a technical perspective that capturing every single bit of detail about an attack and storing it for a replay is a hurdle. However, even the idea that the game engine incorporates a bit of randomness in the attacks doesn't necessarily excuse this level of disconnect. Even if there's a degree of randomness, this sort of thing is implemented using random number generating algorithms that start with seed values that could be saved. Those "random" numbers would then be a reproducible sequence that could be regenerated during the replay to produce consistent results.

Really there's not a very strong case for not being able to provide much more accurate replays, but in my mind the bigger issues involve getting the rampant and frankly blatant cheating out of the system.