Ship Best Target Selection

[Wanderer]

Is there anywhere that ship target selection is detailed?  The reason I'm asking is because of the bonus selection and occassionally I see my ships pounding on silly things, like fortresses wailing away on bombers when there's a stack of fighters in range.

With hull types being so important, I'd like to read some more about the under the hood algorithm that drives target selection.  This gets even more wonky when you start considering things like the Impulse Reaction Emitter or the Zenith Polarizer, needing odd-ball choices for their selectivity (energy consumption and invert the armor selection, respectively).

I'm currently trying to determine DPS/materialCost and DPS/Cap for ship selections to optimize my choices for the AI 10 campaigns I've been goofing off with.  The RAW numbers are astounding.  I'm trying to figure out how much weight I need to apply to the bonuses, from what I've seen I should only weight positives at 25-50% and weight negatives at 300%, since ships seem to get 'stubborn' about killing something it should switch off of unless it's the last thing in range.

As an Example, two similar ships, RAW values:
Sniper: 3,333.33 DPS.  0.92 DPS/MaterialCost.  63,333.33 DPS/Cap
Sentinel Frigate: 4,666.68 DPS, 1.16 DPS/MC, 104,644.44 DPS/Cap

However, that RAW ignores the 6x multiplier the Sniper gets against four different hulls, most importantly the Polycrystal one for dropping bombers.  However, the sniper levels that off by being 0.1 against Command-Grade.  It also 'insta-hits', so you don't waste shots with it (I've seen sentinel shots disappear half-way to target simply because said target died).

So, weighting here is incredibly important, thus my curiousity.

[x4000]

It's been discussed, but it's really really complex (and tends to change), so I don't think there's anything comprehensive.

One thing that is a factor is that once ships have chosen a target, they don't switch targets for another 5-10 seconds.  So if the bombers come into range first, the fortress would attack them first, then not notice the fighters for a bit.  That's simply due to computational limitations (it's tens of millions of calculations per second to run a big battle as it is).

[Hearteater]

This gist from previous threads on this topic is, the targeting considers basically all the math properly.  However, ships don't select optimal targets at infinite frequency.  There is a lot that goes in with building target lists, sorting them by best targets, and spreading out damage to maximize effectiveness.  You'll always be able to micro better than auto-targeting, but you probably can't micro fast enough to improve the entire fight.  But yes, hull types, multipliers, armor, armor penetration, immunities, munition boosters and such are all considered.

There is a lot of info on the forum, but it is a little hard to search on.  Try "target list" and "target selection" in quotes.  Here is a good thread on Focus Fire.

Edit: ninja'd.  I can't help but feel this is retaliation for the export thread

[x4000]

Edit: ninja'd.  I can't help but feel this is retaliation for the export thread

  I didn't even start responding to that other one until I'd already posted my one here.

[keith.lamothe]

fortresses wailing away on bombers when there's a stack of fighters in range.

Yea, as above that's almost certainly a case of the fortress shooting at what's in range and not getting a new target list until sometime after the fighters come in range.  If you're able to reproduce a case where they _keep_ shooting at the bombers (aside from you having explicitly told it to attack bombers, thus setting the preferred-target flags) for a long period of time while fighters are in range I would be interested in seeing that.

With hull types being so important, I'd like to read some more about the under the hood algorithm that drives target selection.  This gets even more wonky when you start considering things like the Impulse Reaction Emitter or the Zenith Polarizer, needing odd-ball choices for their selectivity (energy consumption and invert the armor selection, respectively).

That part's really easy, actually.  For each eligible target it knows how much damage it would do (using the same calculation as would be used if the shot hit the thing, which of course includes all those considerations) and one piece of the eligible-target-list sorting comparison prefers targets against which significantly more damage would be done.

More complex is stuff like the Zenith Paralyzer, Ion Cannons, engine-damage stuff, etc.  Ships for whom damage (i.e. attack power) is not the priority.  A lot of the rules in the sorting method are for handling those.

I'm currently trying to determine DPS/materialCost and DPS/Cap for ship selections to optimize my choices for the AI 10 campaigns I've been goofing off with.  The RAW numbers are astounding.  I'm trying to figure out how much weight I need to apply to the bonuses, from what I've seen I should only weight positives at 25-50% and weight negatives at 300%, since ships seem to get 'stubborn' about killing something it should switch off of unless it's the last thing in range.

Have you tried the reference tab export?  I believe it gives some at-cap stats.  It's what I used as a reference for the big rebalancing about 13 months ago.  Though I probably added some columns to the spreadsheet manually in that process so what you're seeing may not be as useful.