Tiger spotting

[poesel]

Noob started a thread (http://www.battlefront.com/community/showthread.php?t=105850) in the CMBN forum about spotting between stationary and moving tanks. I wanted to apply the new knowledge to this problem.

The setup is a bit different from noob as I have the same tank on both sides. It looks like this:

Distance is 80m. I used the spotter team to time the 'tick' of the tanks. What I found is that the tanks would spot OUTSIDE their 7 second cycles! So I was wrong about the 7 seconds always being fixed. If there's another cycle or some special treatment for this situation I don't know. At this very short range spotting was nearly always instantaneous.

So I changed the distance to 520m. I did not try to time the ticks this time. I know that the moving Tiger comes into sight at the 33s mark and I measured the time from there. I ran the test only 10 times - its a bit time consuming. If someone has good reasons I'll do more. Or better someone else can do it!

https://dl.dropbox.com/u/8811801/Tank%20Spotting%20Test%20002.bts

The average for the static Tiger was 4,5s, for the moving 8,7s. The average deviation 2,6s/4,7s and the range 10/15 (https://docs.google.com/spreadsheet/ccc?key=0ApF6zBjLENkUdFNSRFRPZzY3Y1c0VGRoV01DOER5cUE).

The static Tiger has a spotting advantage of 4s seconds in this scenario. But because of the high deviation there are instances where the moving tank spots faster than the static one.

About the polling cycle: I really don't know what kind of mechanisms are at work here. It looks that there is spotting outside the cycle but sometimes is rigid inside the cycle.

Steiner - you see its not always fixed. Also I'd like to see you using the exploits you described here. An AAR would be nice...

[Steiner14]

That's really good news, if the spotting system is somehow more complex as previously feared.

Did you make a run without the spotters, too?

[poesel]

Back to what I originally wanted to test: Tiger spotting. This time its about probabilities.

Same setup as before: one unit spotting 80 others on a flat field. Tiger vs Infantry and Tiger vs Shermans. Each test run 10 times. The Tiger is buttoned.

I measured the number of IDed units after one minute.

Sample result:

Raw numbers:

Tank vs Inf - units spotted after 1 min:

53, 49, 55, 63, 60, 61, 63, 66, 38, 57

Tank vs Inf - units spotted after 1 min:

43, 43, 53, 49, 44, 45, 46, 44, 47, 54

That is an average of 46,8 Shermans and 56,5 rifle squads spotted after one minute. This looks wrong to me. Tanks should be spotted easier than infantry.

One might object that the tanks are hiding each other. But I tested this and that's not the case. IIRC Steve said so long ago, too.

A thing to note is that generally the tanks to the front of the Tiger were spotted more often than to the back. This effect was much less for the infantry. This is a subjective observation, I didn't count.

A cause for this may be that a rifle squad occupies three AS and thus gets checked three times per unit while the Sherman only gets one check. But this seems not to be reflected in the probabilities.

[SlowMotion]

Very interesting stuff in this thread.

[dieseltaylor]

In a sense these excellent series of tests reveal a rather more human problem. The ability of humans to count/double count under stress. In this sense I doubt it being a Tiger is relevant as if it were a Sherman counting Shermans I would expet the same result [though it would interesting if possible.!]

Seriously if you have a number of guys spotting in a car park and give them a minute would you be able to handle the information to give an accurate response? And more importantly placement of the identified enemy? In practical terms most crews could barely handle ten targets and locations in a minute would be my guess - and even that would require relative spotting to an identified point.

Spotting too easy ........

I realise that whilst we flounder to understand what is going on with the spotting we are unearthing some unlikely events and behaviours in these unlikely events.

One of the chief ones being the ability to apparently spot forwards and backwards with equal [mainly] accuracy. I am interested in the diagram which apparently shows that viewing to the starboard [right side] is not as good - is there a pattern there also?

[Vanir Ausf B]

Raw numbers:

Tank vs Inf - units spotted after 1 min:

53, 49, 55, 63, 60, 61, 63, 66, 38, 57

Tank vs Inf - units spotted after 1 min:

43, 43, 53, 49, 44, 45, 46, 44, 47, 54

I assume the second set should be Tank vs. Tank.

Great work on these tests. I agree that this seems to be the opposite of what it should be. It may partially explain why we often see complaints about infantry being spotted too easily, but not so much with vehicles.

[c3k]

That is a great test... The pictures are very helpful.

Side/rear visibility in tanks: I don't have my Tiger references at hand, but the driver and radio operator had sidewise vision (I forget to what degree). As well, the TC had 360^ cupola vision blocks. Some versions of the Tiger had pistol ports around the back and sides of the turret with vision capability.

Having said that, the question remains, "should a tank have such good spotting capability for 360^?"

[poesel]

I assume the second set should be Tank vs. Tank.

Uuups, yes, sorry. The data is here: https://docs.google.com/spreadsheet/ccc?key=0ApF6zBjLENkUdFNSRFRPZzY3Y1c0VGRoV01DOER5cUE

@dieseltaylor: you are of course correct that handling that amount of information in one minute is next to impossible. But this is a VERY artificial scenario. If you spot 80 enemy units around you in one minute then you have other problems anyway...

[dieseltaylor]

I was musing about what to do with multiple spottings in the event I were designing the game. Essentially I would say that a unit commander can only handle so much and whereas with a squad your men may take on immediate infantry threats with a tank it is very much a single brain has to calculate and deal with the situation and I think there is a maximum amount certainly 10 or less that a tanker can deal with.

In my design the closest forward faced targets get priority as the crew report in but in any event definite identification and range stops at 10 everything after that cannot be fully registered.

I am sure somewhere someone has researched this type of problem!

[Offshoot]

A cause for this may be that a rifle squad occupies three AS and thus gets checked three times per unit while the Sherman only gets one check. But this seems not to be reflected in the probabilities.

Heh, so my maths is really rusty, but surely if you have three trials to spot one unit of infantry (because the squads cover three ASs) compared to only one trial to spot each tank, that is going to affect the number of spotting events in the 1 minute time frame in this test? Regardless, to remove this possible variable, it would probably be best to do this test using three-man scout teams, which occupy only one AS, rather than full squads.

[JonS]

In my design the closest forward faced targets get priority as the crew report in but in any event definite identification and range stops at 10 everything after that cannot be fully registered.

Interesting idea.

It does raise the question about the 'not fully registered' targets above 10 though. Do you ignore them? Pretend they don't exist? Present them as <?> spots?

What if the first 10 targets are dismounted crews, but the next 10 is a fleet of Fireflies - is it reasonable that my tank become target fixated on irrelevancies while mortal peril awaits just beyond their artificially imposed event horizon?

Wwhile the player can be presented with fuzzy information, the program has to have definite, specific, and explicit instructions to follow.

[Steiner14]

Excellent test again.

Regardless, to remove this possible variable, it would probably be best to do this test using three-man scout teams, which occupy only one AS, rather than full squads.

Great idea.

Poesel, do you have some more time to spend?

[dieseltaylor]

What if the first 10 targets are dismounted crews, but the next 10 is a fleet of Fireflies

JonS

BF did model Fireflies as being less important than vanilla Shermans in CMx1 if the Sherman was 10 metres closer! - this was not good IMHO. I would suggest that the target overload is of units that can badly hurt you first to be assimilated and that is of course a function of proximity and type of enemy.

And yes if you were presented with two teams with anti-tank weapons with-in their range you would certainly give them a very high priority compared to a tank 1500 metres away. You can discard dismounted tank crews, armoured half-tracks etc form the highly dangerous and not count them high in the ten total.

Apparently the natural number that people can naturally handle items is in fours I learned today. Not say for societies were numbering stops at one and at three.

* Has anyone checked whether the vanilla Sherman /Firefly problem occurs in CMx2?

[JonS]

And yes if you were presented with two teams with anti-tank weapons with-in their range you would certainly give them a very high priority compared to a tank 1500 metres away. You can discard dismounted tank crews, armoured half-tracks etc form the highly dangerous and not count them high in the ten total.

Sure ... but that means the tank (gun, whatever) has to assess and classify EVERY target it can see, not just the first 10. And then re-assess and re-classify every time it sees a new target. Which is pretty much the opposite of what you started with.

[dieseltaylor]

Sure ... but that means the tank (gun, whatever) has to assess and classify EVERY target it can see, not just the first 10. And then re-assess and re-classify every time it sees a new target. Which is pretty much the opposite of what you started with.

JS

Fair point. Target overload is a problem and from poesel's test we can see the game has unrealistic ability to plot large numbers pretty accurately inside a minute.

I have two ideas how perhaps to handle this:

- number off from the gun positions so for instance if we say realistically the maximum number that can be spotted and placed accurately is ten then the first decision tree comes with what is currently with the direct line of sight of the gun. If you have three potential targets the next decision is which is the most dangerous and this can a fairly basic AT team head on-on is less dangerous than an AT team on your flank or rear. In the event your gun was facing broadside and you saw an AT crew close to they would be high on the threat scale. So effectively you build up a threat scale that encompasses :

proximity

size of target

chance of them hitting you

chance of you hitting them

chance of serious damage

In the broadside example you might have in line from the gun barrel a AT team, a 6pdr ATG, and a Sherman and on the basis of the formula you then go for the highest threat. This could be slightly randomised and moreso if their threat values were equal.

It could be tweaked on the basis if facing forward and the tank has a hull MG then that alters the equation. If the threat value falls below a certain danger threshold then the turret turns to another of the noted threats to left or right and the same process is run.

My theory is that a tank crew can only handle so much information in a minute and that normally this spotting will be to the front and particularly where the gunner can wreak some damage. SO counting off targets from the front quadrant comes first and if you hit ten thats is your effective limit and having thirty enemy around you becomes an irrelevancy.

The second method is actually to give a maximum number of accurate spots for each crew member with visibility in the required direction. This actually could work in conjunction with the first option and is perhaps a more sophisticated model because it does allow vehicles/tanks with better visibility to make use of it. Particularly relevant to armoured cars with two drivers!

It is surprising how quickly targets all-around the Tiger are spotted despite only the tank commander having all-round spotting ability. Assuming the commander can only take so much information as to where and how far enemy formations are in each minute then one would expect more detail over minutes rather than such a rapid assimilation.