Rocket scatter calculations

[Alketi]

Are we sure that rocket scatter (either the calculation or the tooltip) are working correctly?

As the floating tooltip reads +/- X tiles, I take this to mean there is X tile deviation in any direction.

In the video below, the rocket has an approximately 8 tile diameter. It's roughly centered on the Mec, yet with a 2.36 tile deviation it misses the Mec entirely.

Images:
http://imgur.com/a/ejAsY

Video:
https://youtu.be/FgpnvQNKd9Y?t=7m7s

[chrisb]

Rocket scatter is up to 4 tiles, 2 with FITH. The deviation is just telling you in a simple way that your roughly 70% likely to hit within that distance. Doing some rough calculations it seems like he had an 8% chance to scatter 4 tiles, and he scattered 4 tiles.

[Alketi]

Rocket scatter is up to 4 tiles, 2 with FITH. The deviation is just telling you in a simple way that your roughly 70% likely to hit within that distance. Doing some rough calculations it seems like he had an 8% chance to scatter 4 tiles, and he scattered 4 tiles.

So the number presented to the player is not the max scatter, but the one sigma scatter?

Now, I suppose this all makes sense, since the max scatter is still the original 4 tiles, but how is the player supposed to know what this number represents? As you video shows, the player thought it implied max scatter.

I would suggest perhaps putting a percentage in parenthesis after the +/- figure. (68%) or whatever is used in the calculation.

[chrisb]

The max scatter is listed on the bottom UI when you click Launch Rocket.

I think showing the deviation is not necessarily that useful. It's fine if you understand non-linear functions and can work out what the number means, but it's not really intuitive for most people. It might be simplest to just display the % chance to scatter at each tile distance. It has the most direct relation to what is actually happening, and most people can look at a list of numbers and figure out if it's worth doing or not, or at least learn what the numbers mean after a few bad rockets. Had he known he had almost a 10% chance to scatter 4 tiles, and a 36% chance to scatter at least 3, he might not have taken the rocket, or at least know the gamble he was taking up front.

• 0 - 4.9%
• 1 - 22.0%
• 2 - 37.2%
• 3 - 28.0%
• 4 - 7.9%

This gives the most direct information. These numbers are from a 4 Scatter, 80 Aim, 20 Range rocket. Compare that with a 9 range rocket.

• 0 - 41.0%
• 1 - 41.0%
• 2 - 15.4%
• 3 - 2.6%
• 4 - 0.2%

It's easy with this information to see just how massive the scatter falloff becomes with the extra distance, which was his biggest issue. The better play he could have made if he wanted a good rocket would have been to move much closer. He easily could have moved half a dozen tiles closer and had much lower chance to scatter.

He also seems to have the LW1 habit of steadying his rocket, which has no effect on scatter at all.

[Tuhalu]

It's easy with this information to see just how massive the scatter falloff becomes with the extra distance, which was his biggest issue. The better play he could have made if he wanted a good rocket would have been to move much closer. He easily could have moved half a dozen tiles closer and had much lower chance to scatter.

But moving first adds 2 tiles of max scatter to his rocket. Do you mean he should have moved closer and used the rocket on his next turn?

[chrisb]

It's easy with this information to see just how massive the scatter falloff becomes with the extra distance, which was his biggest issue. The better play he could have made if he wanted a good rocket would have been to move much closer. He easily could have moved half a dozen tiles closer and had much lower chance to scatter.

But moving first adds 2 tiles of max scatter to his rocket. Do you mean he should have moved closer and used the rocket on his next turn?

No, on the turn before he launched the rocket he used steady weapon on his blue move. He could have moved forward a bunch of tiles and had a vastly better rocket. That looks like a LW1 habit that completely backfires in LW2 because rocket scatter is vastly different. Personally I don't like it, it's very clunky and unreliable at anything beyond 10 tiles range. I think it would be much more intuitive if it wasn't using such a discrete calculation but rather used something a lot smoother than jumping whole tiles. Or at least show the actual chance to jump instead of showing the outcome of some non-linear function that most people don't understand the ramifications of.

[Alketi]

• 0 - 4.9%
• 1 - 22.0%
• 2 - 37.2%
• 3 - 28.0%
• 4 - 7.9%

I agree that a list of tiles and hit percentages would be the most useful way to display the information. Then the player would have no confusion.

The problem with the current "+/-" is that it implies RANGE or max variance, when it's actually an average that can be exceeded, which frustrates the player as the video demonstrates. I've also been fooled into thinking that the HTML text meant "max scatter for a rocket fired at this location" and then watched in horror as the rocket flew even further off course.

Also, the "Max 4" footnote isn't always accurate on a per-shot basis, as sometimes the max scatter may only be 3 tiles (or fewer) depending on range and aim.

So, yes, a short table would be ideal. Alternatively, the word "AVG" would at least serve as a disclaimer and prevent the game from being blamed.

[chrisb]

• 0 - 4.9%
• 1 - 22.0%
• 2 - 37.2%
• 3 - 28.0%
• 4 - 7.9%

I agree that a list of tiles and hit percentages would be the most useful way to display the information. Then the player would have no confusion.

The problem with the current "+/-" is that it implies RANGE or max variance, when it's actually an average that can be exceeded, which frustrates the player as the video demonstrates. I've also been fooled into thinking that the HTML text meant "max scatter for a rocket fired at this location" and then watched in horror as the rocket flew even further off course.

Also, the "Max 4" footnote isn't always accurate on a per-shot basis, as sometimes the max scatter may only be 3 tiles (or fewer) depending on range and aim.

So, yes, a short table would be ideal. Alternatively, the word "AVG" would at least serve as a disclaimer and prevent the game from being blamed.

Max scatter is never 3 and doesn't change based on range or aim. There are only two modifiers to max scatter. FITH is -2, moving is +2. So you can only have 2, 4 or 6 tiles of max scatter. Also moving adds a -30 aim malus to the shot, so moving and shooting a rocket, even with FITH and 100 aim, is almost always a bad idea.

[Alketi]

Max scatter is never 3 and doesn't change based on range or aim. There are only two modifiers to max scatter. FITH is -2, moving is +2. So you can only have 2, 4 or 6 tiles of max scatter. Also moving adds a -30 aim malus to the shot, so moving and shooting a rocket, even with FITH and 100 aim, is almost always a bad idea.

Is joinrbs scatter calculator wrong then? Max (theoretical) scatter never changes, but max scatter on a per-shot basis depends on aim/range/movement.

Aim: 100
Range: 10
Max scatter: 4 (no FITH)

0: 88.5%
1: 11.0%
2: 0.5%
3: 0%
4: 0%

Max scatter in this case is 2 tiles, not 4. That was the point I was making - that max scatter can change on a per-shot basis. 4 tiles is impossible under some scenarios, so a single hardcoded tooltip isn't perfectly representative of individual cases.

[chrisb]

Max scatter is never 3 and doesn't change based on range or aim. There are only two modifiers to max scatter. FITH is -2, moving is +2. So you can only have 2, 4 or 6 tiles of max scatter. Also moving adds a -30 aim malus to the shot, so moving and shooting a rocket, even with FITH and 100 aim, is almost always a bad idea.

Is joinrbs scatter calculator wrong then? Max (theoretical) scatter never changes, but max scatter on a per-shot basis depends on aim/range/movement.

Aim: 100
Range: 10
Max scatter: 4 (no FITH)

0: 88.5%
1: 11.0%
2: 0.5%
3: 0%
4: 0%

Max scatter in this case is 2 tiles, not 4. That was the point I was making - that max scatter can change on a per-shot basis. 4 tiles is impossible under some scenarios, so a single hardcoded tooltip isn't perfectly representative of individual cases.

It's not really impossible, it is just very very unlikely. With that rocket it has a 97% on each roll of not scattering. So in order to scatter 4 tiles it would need to roll 3% 4 times, giving it a 0.03^4 chance, which is a 0.000081% chance of it happening. He is using a 1 decimal place formatting which rounds it off to 0 for the number display.

If you make the "I" column about 3 times wider and make the "I8" cell use more decimal places you'll eventually see that it's not actually zero. The add decimal places button is right below "Format" in the menu bar.

[Alketi]

Gotcha! Precision, of course!

Well, you'd better add 10 decimal places to your list as well. Because shooting wild under the guise of "0%" chance is probably worse than the current text.

I do understand the issues faced in presenting the data properly. However, I still maintain that "+/-" is more confusing than intended.

I presume they used the ASCII symbol for "+/-" so that no translations were needed, where something like "AVG" would need to be translated. Ah well. Pay up front or pay later when someone curses the game on YouTube. )