Fractal Softworks Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

Starsector 0.97a is out! (02/02/24); New blog post: Simulator Enhancements (03/13/24)

Pages: [1] 2 3

Author Topic: Sensor Normalization  (Read 13759 times)

Wyvern

  • Admiral
  • *****
  • Posts: 3784
    • View Profile
Sensor Normalization
« on: December 21, 2015, 01:10:38 PM »

Edit: This thread has been co-opted for general discussion of current sensor mechanics, starting around here.

So, the current sensor system is neat - but it's also highly non-intuitive, and to (what seems to me to be) relatively little purpose.

A larger fleet can see farther - but it can also be seen from farther, to essentially equal degree.  And, due to the non-linear response of detection range to changes in sensor strength / profile, it barely matters what ships you're actually using - a large fleet with some warships using Augmented Engines doesn't really have much of a disadvantage compared to a mid-sized fleet using phase ships almost exclusively.

So here's my suggestion: normalize the whole thing.  Make it so you can detect fleets at fixed ranges regardless of fleet size, and reduce the system to three circles: range to detect fleets with transponders, range to detect fleets without transponders, and range to detect fleets that are running silent.

Then, for things that really should modify sensor ranges, you can give some special case rules:
  • Phase Ships might get "Reduces the range at which you can be detected by up to 25% (current value X%), depending on the proportion of your fleet's deployment points that are phase ships."
  • Augmented Engines might get "Increases the range at which you can be detected by up to 20% (current value X%), depending on the proportion of your fleet's deployment points that have augmented engines equipped."
  • Hulls with advanced sensors might get "Increases the range at which you can detected enemy fleets by 10%.  This effect does not stack from multiple ships."
  • And (D) grade hulls might get "Increases the range at which you can be detected by up to 25% (current value X%), depending on the proportion of your fleet's deployment points that are (D) grade."

This is definitely a bit game-y; you'd have to invent some rather twisty bits of lore to explain why it's harder to detect a Wolf class frigate that's in a fleet with an Afflictor than it is to detect that same Wolf on its own.  But I think it'd have similar practical effects to the current system, plus it's much easier to understand and predict, plus it naturally deals with the early-game low-sensor-range problem without needing a special "low engine interference" bonus.

Note that I've done away with the civilian hull penalty entirely - it's neat from a realism standpoint, but it's just not fun from a gameplay standpoint to have ships with what amounts to a "You probably want to avoid this" marker on them.

_____
As a bit of a tangent: I wouldn't remove the "civilian grade" hull mod, though - I'd just repurpose it.  Say, 20% reduction in maintenance costs but -10% max CR & increased CR drain from hazards & abilities?  Easy to keep running under normal circumstances, but not as good at dealing with hyperspace storms or emergency burn.
« Last Edit: December 22, 2015, 03:54:43 PM by Wyvern »
Logged
Wyvern is 100% correct about the math.

Dabor

  • Lieutenant
  • **
  • Posts: 92
    • View Profile
Re: Sensor Normalization
« Reply #1 on: December 21, 2015, 02:42:07 PM »

I don't think this would help, and it has massive gaps on its own (your gamey example of a non-stealth ship being stealthier in a fleet with stealth ships than on its own.)

The current system also has one massive, natural, intuitive, brilliant result: fleets tend to see each other at the same time. Unless one fleet has a bunch of civilian or phase ships, a group of 5 frigates or a 50 ship siege fleet see each other at the same time. 2 50 ship fleets see each other at roughly the same time. 2 5 frigate groups see each other at roughly the same time.

While you get advantages to narrow degrees in certain contexts, I think the solution is to make sensors LESS normalized. Make phase ships much stealthier, and make sensor ships much better at detecting. The same way as a fleet may run a cargo ship or a fuel tanker, it should add in dedicated sensor craft. Maybe silent fleets will even have some sort of "stealth generator" ship that reduces the fleet's sensor signature (maybe the effect in battle is as an anti-missile ECM field).

It's not that your system is outright bad or anything, it's just that if you're going to make a gamey system you may as well just go back to the "everyone sees everyone" system. I don't think it was really outright toxic to the game or anything.
Logged

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 23986
    • View Profile
Re: Sensor Normalization
« Reply #2 on: December 21, 2015, 02:44:53 PM »

So, the current sensor system is neat - but it's also highly non-intuitive, and to (what seems to me to be) relatively little purpose.

A larger fleet can see farther - but it can also be seen from farther, to essentially equal degree.  And, due to the non-linear response of detection range to changes in sensor strength / profile, it barely matters what ships you're actually using - a large fleet with some warships using Augmented Engines doesn't really have much of a disadvantage compared to a mid-sized fleet using phase ships almost exclusively.

Hmm. I find it hard to agree with this premise. A medium-sized fleet of phase ships will have a huge advantage over a large fleet, let alone one where some ships are using augmented engines. The advantage lies in breaking the symmetry - fleets can see each other at the same range *until* you add ships with bonuses or penalties into the mix.

Given that the profile reduction for phase ships is very hefty, the hypothetical phase-ship fleet would likely see our other hypothetical fleet at something like double the range it would itself be seen at.


Then, for things that really should modify sensor ranges, you can give some special case rules:
  • Phase Ships might get "Reduces the range at which you can be detected by up to 25% (current value X%), depending on the proportion of your fleet's deployment points that are phase ships."
  • Augmented Engines might get "Increases the range at which you can be detected by up to 20% (current value X%), depending on the proportion of your fleet's deployment points that have augmented engines equipped."
  • Hulls with advanced sensors might get "Increases the range at which you can detected enemy fleets by 10%.  This effect does not stack from multiple ships."
  • And (D) grade hulls might get "Increases the range at which you can be detected by up to 25% (current value X%), depending on the proportion of your fleet's deployment points that are (D) grade."

This is definitely a bit game-y; you'd have to invent some rather twisty bits of lore to explain why it's harder to detect a Wolf class frigate that's in a fleet with an Afflictor than it is to detect that same Wolf on its own.  But I think it'd have similar practical effects to the current system, plus it's much easier to understand and predict, plus it naturally deals with the early-game low-sensor-range problem without needing a special "low engine interference" bonus.

I'm not opposed to the idea of "simple rules with exceptions". I think that often produces easier-to-understand mechanics than "complex base ruleset". Thing is, it feels like the baseline sensor rules are already very simple - more ships, easier to see and be seen. And then you have the "exceptions", marked with hullmods - phase ships, civ ships, sensor ships.

The part people are having trouble with - gauging the exact impact of these exceptions - I don't think gets particularly easier from having a different set of exceptions.
Logged

Wyvern

  • Admiral
  • *****
  • Posts: 3784
    • View Profile
Re: Sensor Normalization
« Reply #3 on: December 21, 2015, 03:17:33 PM »

Given that the profile reduction for phase ships is very hefty, the hypothetical phase-ship fleet would likely see our other hypothetical fleet at something like double the range it would itself be seen at.
Doesn't match my in-game experience; no matter what I do, I don't seem to end up with more than 100 points of difference in the display between range-to-detect versus range-to-be-detected.  Admittedly, I haven't done anything super-extreme like run around with a single phase frigate, or installing augmented engines on a bunch of civilian vessels, or even just completely ditch my support train of (non-civilian-grade) cargo haulers; but in general, once you're looking at the numbers displayed in the 800+ range, nothing really seems to matter much anymore.

Thing is, it feels like the baseline sensor rules are already very simple - more ships, easier to see and be seen. And then you have the "exceptions", marked with hullmods - phase ships, civ ships, sensor ships.

The part people are having trouble with - gauging the exact impact of these exceptions - I don't think gets particularly easier from having a different set of exceptions.
Simple to state, but complicated to display / communicate the effects.  And while I agree that people (myself included) have trouble guaging the impact of the exceptions, I don't think that's the only source of trouble; witness the number of people asking "Why can't I just have a circle showing how far I can see?"

And the 'different set of exceptions' - I explicitly chose the examples I did for clarity of effect.  "Up to -25% detection range (current value -7%)" instantly communicates what it does (and how much of an impact the rest of your fleet has on the effect), while "reduces sensor profile by 75%" only communicates "This ship is stealthier by some amount that may or may not make any real difference given the rest of your fleet".
Logged
Wyvern is 100% correct about the math.

Dabor

  • Lieutenant
  • **
  • Posts: 92
    • View Profile
Re: Sensor Normalization
« Reply #4 on: December 21, 2015, 03:29:18 PM »

I have to second the note on the experience front. Developing a normal fleet pretty much always leaves the sensor strength/signature within something like 10% of each other. Throwing in a couple of phase frigates or an apogee doesn't really feel very different.

What's your opinion on dedicated stealth or sensor ships? Rather than something like a Doom or Apogee which gives this as a side perk, these would be high-maintenance ships where 80% of their benefit is reducing signature/increasing sensors on the campaign map.
Logged

Cik

  • Admiral
  • *****
  • Posts: 607
    • View Profile
Re: Sensor Normalization
« Reply #5 on: December 21, 2015, 03:33:00 PM »

there should probably be a mod in the techtree called stealth plating that acts like lightweight armor but has a sensor bonus instead.

the reason 'stealth' ships aren't as valuable as they probably should be on the strat map is that fleet detection is all or nothing, it's not on a per-ship basis which means that a fleet of even as much as 50% phase ships will still be detected (phase ships included) at a pretty good range. if it was per-ship you could have a bunch of freighters escorted by omens and shades that get jumped by pirates and well

surprise!

it also doesn't help that phase ships are not exactly easy to find, so making a fleet entirely of phase ships (while strategically kind of desirable?) is pretty difficult. likewise, not having any shields in a fleet is probably a pretty egregious weakness.
Logged

Wyvern

  • Admiral
  • *****
  • Posts: 3784
    • View Profile
Re: Sensor Normalization
« Reply #6 on: December 21, 2015, 03:40:28 PM »

Just as an addendum, I want to say: I, personally, do actually like the current sensor mechanics* - but I like complex fiddly things with lots of math, and have many times found that I need to back off from that in order to make game bits that I'm assembling approachable to other people.

So, personally, I'll be just as happy if this suggestion doesn't get implemented; I'm making it to offer thoughts about how to address issues that other people seem to have.

_____
* Aside from the same-y feel that Dabor seconds, where I can't really seem to do much to manipulate sensor range versus detection range.
Logged
Wyvern is 100% correct about the math.

ANGRYABOUTELVES

  • Admiral
  • *****
  • Posts: 592
  • AE ALTADOON GHARTOK PADHOME
    • View Profile
Re: Sensor Normalization
« Reply #7 on: December 21, 2015, 03:43:13 PM »

There are a lot of people on the forums confused about sensors, but you have to remember that most of the time, only the people who are having a problem will complain. I haven't said squat about the sensor system because I got the basic concept almost immediately. Like Dabor said; fleets see each other at the same time, unless special modifiers come into play like transponders, running dark, or phase/civillian/sensor ships. The bonuses from sensor/phase ships could stand to be quite a bit larger, but otherwise it feels pretty much fine.

Maybe there should be a sensor tutorial that goes over the basics of running dark, transponders, phase ships, that sort of thing. I like the system and found it fairly intuitive, but from the amount of forum noise about it, it could stand to be explained better.
« Last Edit: December 21, 2015, 04:07:12 PM by ANGRYABOUTELVES »
Logged

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 23986
    • View Profile
Re: Sensor Normalization
« Reply #8 on: December 21, 2015, 04:10:26 PM »

Hmm. (Detected-at range for 5x Doom: 780. 5x Onslaught: 1080. 5x Onslaught w/ augmented engines: 1160, but the sensor range drops from 1080 to 950, so the practical difference for installing augmented engines on all of those is something like 20%.)

I think the 800ish range is just a sweet spot for player fleet size. The stats aren't going to fluctuate much if you're not making a concerted effort to build your fleet for stealth (or cargo-hauling, which isn't exactly relevant right now), but that's very much the point. A civ ship here and a phase ship there aren't going to make a big difference. A lone phaseship has a detected-at range of 370, though... but it still might make sense to tweak the formula to spread it out a little more uniformly on the lower end; perhaps it jumps up too quickly.


Simple to state, but complicated to display / communicate the effects.  And while I agree that people (myself included) have trouble guaging the impact of the exceptions, I don't think that's the only source of trouble; witness the number of people asking "Why can't I just have a circle showing how far I can see?"

Fair enough - but if you can draw a circle, but that circle is then lying to the player because of all the exceptions, is that any better? :)

And the 'different set of exceptions' - I explicitly chose the examples I did for clarity of effect.  "Up to -25% detection range (current value -7%)" instantly communicates what it does (and how much of an impact the rest of your fleet has on the effect), while "reduces sensor profile by 75%" only communicates "This ship is stealthier by some amount that may or may not make any real difference given the rest of your fleet".

This is true, but the same information about the effect of a given phase ship in your fleet could already be derived given your fleet makeup (which was Gothars' suggestion in the release thread), and the effect is fleet-composition-dependent in either case.

It'd be good to see, though. "This phase ship is contributing 3% to my detected-at range, where this Wolf is contributing 7%" would be useful information.

I appreciate the suggestion, btw. This is definitely something that bears thinking/talking about, and - evidently - could be explained better. I do feel like it's more a question of that, though (and perhaps tweaking the curve), than revamping the system entirely.

If you're going to have stealth, you need detection ranges to be variable, right? Otherwise you're not giving the player an opportunity to build for stealth. And if you have variable detection ranges (which holds both for the current implementation and your idea in the OP), then you can't have a circle. You see 'em when you see 'em. It's more important to be informed when they're about to see you/when you're about to lose them - that's what you care about, right? In terms of gameplay - e.g. sneaking into a port right now - I think the current UI does a good job for that particular task.

In terms of conveying everything that goes into it, though... that's where it gets complicated, and part of the question is just how much to explain. I mean, talking about "logarithmic dropoff" is probably not a good idea.

... it'd be a lot simpler if ships could just contribute a linear amount. Too bad this leads to ranges that just don't work out gameplay-wise - either super-tiny on the low end, or super-huge on the high end. And if you add a minimum (a la "low engine interference"), then you'd still have an enormous "dead zone" where your fleet composition doesn't matter at all, until it starts to exceed the minimum. And really, if the formula is "linear, but..." then it might just as well be a black box.


There are a lot of people on the forums confused about sensors, but you have to remember that most of the time, only the people who are having a problem will complain. I haven't said squat about the sensor system because I got the basic concept almost immediately. Like Dabor said; fleets see each other at the same time, unless special modifiers come into play like transponders, running dark, or phase/civillian/sensor ships. The bonuses from sensor ships could stand to be quite a bit larger, but otherwise it feels pretty much fine.

Maybe there should be a sensor tutorial that goes over the basics of running dark, transponders, phase ships, that sort of thing. I like the system and found it fairly intuitive, but from the amount of forum noise about it, it could stand to be explained better.

Thanks for chiming in! Honestly, I do feel like it's in pretty good shape. Some amount of misunderstanding is unavoidable, but there's always room to improve. And yeah, a campaign tutorial would be good (but later). Hmm. Perhaps some of those help popups.


What's your opinion on dedicated stealth or sensor ships? Rather than something like a Doom or Apogee which gives this as a side perk, these would be high-maintenance ships where 80% of their benefit is reducing signature/increasing sensors on the campaign map.

Talked about it a bit internally; we'll see!
Logged

Wyvern

  • Admiral
  • *****
  • Posts: 3784
    • View Profile
Re: Sensor Normalization
« Reply #9 on: December 21, 2015, 05:03:03 PM »

Here's an example of what I mean by same-y:

My fleet in .7 consisted of an Onslaught and a Dominator, both with Augmented Engines, three destroyers & three frigates (no sensor relevant mods), two Omens, and a civilian-grade Ox-class tug.
My detection range is listed as 1090.  My to-be-detected range is listed as 1110 - a difference of 20 points.  If I put those Omens in storage, those numbers go to 1000/1100 - still a pretty trivial distinction for a fleet that's really not concerned about stealth mechanics at all.

Another sample fleet, now from .7.1: 1x Doom (w/ augmented engines), 4x phase frigate, 3x destroyer, 1x frigate - this would be the aforementioned 'mostly phase ship medium size fleet' - gets ranges of 990/960; this is a fleet that feels like it "ought" to be stealthy, and in fact is stealthier than average... to an essentially trivial degree.  If I replace Augmented Engines on the Doom with Unstable Injector and add an Ox to keep its speed up, the numbers go to 1020/970; slightly better, but not to a degree that makes any practical difference.

If I take the above fleet and drop two of those destroyers, paring it down to just the combat components (and that one Ox), my numbers are 960/880.  Better, but still not enough to feel like it really matters.  It's not until I ditch everything that's not a phase ship that I get any appreciable stealth - a Doom and five Afflictors gives 940/650, which is finally starting to feel like maybe it's a little bit sneaky, though getting there costs me a point of burn speed.

In other words, unless you go for a really extreme example (like augmented engines on everything, or a fleet consisting only of phase ships), the numbers don't really change much; you can detect fleets about as far away as you can be detected, plus or minus only a very small margin.  The log scaling of sensor ranges takes what feels like they ought to be significant differences and squishes them all into "nope, doesn't matter".
Logged
Wyvern is 100% correct about the math.

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 23986
    • View Profile
Re: Sensor Normalization
« Reply #10 on: December 21, 2015, 06:32:08 PM »

My fleet in .7 consisted of an Onslaught and a Dominator, both with Augmented Engines, three destroyers & three frigates (no sensor relevant mods), two Omens, and a civilian-grade Ox-class tug.
My detection range is listed as 1090.  My to-be-detected range is listed as 1110 - a difference of 20 points.  If I put those Omens in storage, those numbers go to 1000/1100 - still a pretty trivial distinction for a fleet that's really not concerned about stealth mechanics at all.

Hmm. I wouldn't say 10% is trivial, although I won't argue that sensor ships couldn't contribute more than they do. You're not going to make a task force entirely out of sensor ships, after all. On the other hand, it's a question of whether "sensor ship" is an Omen's role or a side perk of having one. In the latter case, a 10% difference here doesn't feel too out of place.

Another sample fleet, now from .7.1: 1x Doom (w/ augmented engines), 4x phase frigate, 3x destroyer, 1x frigate - this would be the aforementioned 'mostly phase ship medium size fleet' - gets ranges of 990/960; this is a fleet that feels like it "ought" to be stealthy, and in fact is stealthier than average... to an essentially trivial degree.  If I replace Augmented Engines on the Doom with Unstable Injector and add an Ox to keep its speed up, the numbers go to 1020/970; slightly better, but not to a degree that makes any practical difference.

That, on the other hand, doesn't feel right. Ah, I see why - the "Phase Field" hullmod really ought to be a multiplier. It isn't, so Augmented Engines does a number on phase ships - it's almost as if the Doom wasn't a phase ship at all, and had poor sensors to boot. Made a note to adjust that.

If you remove augmented engines (without adding an Ox), we're again back to a 10% difference. For a fleet that's only slightly more than half-phase, does that feel too far off?

Again, though, it might be nice to make those numbers a little higher. 20-25% for this case would feel more right, I think. It's just a question of fitting that in without producing extreme numbers on the far end of the scale.
Logged

Wyvern

  • Admiral
  • *****
  • Posts: 3784
    • View Profile
Re: Sensor Normalization
« Reply #11 on: December 21, 2015, 07:36:20 PM »

For the record, I'm not looking at percentages, I'm looking at the absolute values - in all of the above examples save the last extreme phase-ship-only one, there's no more than a 100 point difference between sensor range vs detection range; that's a pretty small margin to play with for trying to see things before you're seen.

(Come to think of it, the only reason we can actually be sneaky to get into ports is because patrols run with their transponder on, and transponder on vs. off completely overpowers any other differences in fleet stealthiness.  Hm.  I wonder what'd happen if the diminishing returns on sensor profile stuff were a bit less harsh, and transponder added only +500?)
Logged
Wyvern is 100% correct about the math.

Dabor

  • Lieutenant
  • **
  • Posts: 92
    • View Profile
Re: Sensor Normalization
« Reply #12 on: December 21, 2015, 07:41:25 PM »

Again, though, it might be nice to make those numbers a little higher. 20-25% for this case would feel more right, I think. It's just a question of fitting that in without producing extreme numbers on the far end of the scale.
One think that comes to mind is that in many games of various genres, it's absolutely reasonable to expect yourself to have a few times the detection range of "roaming mobs" (i.e. pirates and bandits) and for military assault groups, patrol groups and transport to have rather significantly different detection ranges.

Frankly, 10% or 25%, while important in terms of being able to abort your decision to approach before your opponent even knows about it, feel kind of trivial and anti-climactic, and seeing as you often get detected from a distance where avoiding that enemy would be a multi-system detour, or upon warping into/out of a system, it's rare to feel the difference.

One way to make this more noticeable might be to make systems impact active abilities. For example, ships with Phase Field lose 75% of their signature when Going Dark rather than 50%. Ships with Augmented Engines gain 6 burn instead of 5 on emergency burn. Ships with Improved Sensors increase your total sensor range during a Sensor Burst by, say, 5% each (on top of their naturally good sensors). Maybe Phase Ships don't get a sensor penalty when Emergency Burning due to their better stealth-while-moving design. Just a lot of little things that players can feel the difference of.
Logged

ANGRYABOUTELVES

  • Admiral
  • *****
  • Posts: 592
  • AE ALTADOON GHARTOK PADHOME
    • View Profile
Re: Sensor Normalization
« Reply #13 on: December 21, 2015, 08:08:03 PM »

If you remove augmented engines (without adding an Ox), we're again back to a 10% difference. For a fleet that's only slightly more than half-phase, does that feel too far off?

Again, though, it might be nice to make those numbers a little higher. 20-25% for this case would feel more right, I think. It's just a question of fitting that in without producing extreme numbers on the far end of the scale.
Phase Field says it reduces sensor profile by 75%. That number gives the impression that a fleet that's full-phase will have a sensor profile that's 25% of its sensor range, a fleet that's half-phase will have a sensor profile 62.5% of its sensor range, etc. Just going by what the game says it does, a fleet that's more than half-phase should have a sensor range more than 50% higher than its sensor profile; e.g. if its sensor range is 1000, its sensor profile should be 625 or lower.
Logged

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 23986
    • View Profile
Re: Sensor Normalization
« Reply #14 on: December 22, 2015, 12:57:44 PM »

(Co-opting this thread to carry on discussion from release thread.)

I wonder if it'd raise different questions (i.e. why do the numbers for each ship keep changing as you add/remove ships?)
Hmm. Maybe it would make sense to show the "range" values as a percentage instead. ... And it wouldn't confuse by showing a range number that's not useful aside from self-comparison purposes. Wrote this down.

Maybe a combination of percentage and total range for the tooltip, like this:



______
Your total sensor range is 1500 SU against a typical medium fleet, 1000 SU against a typical small fleet and 2000 SU against a typical big fleet.

Your fleet has 80% of the sensor range of a typical fleet of its size.

Contribution by ship:

Dominator 30%
Omen 30%
Tarsus 10%
Buffalo 10%
______



Another example:
Spoiler
______
Your total sensor range is 1000 SU against a typical medium fleet, 700 SU against a typical small fleet and 1300 SU against a typical big fleet.

Your fleet has 130% of the sensor range of a typical fleet of its size.

Contribution by ship:

Apogee 100%
Wolf 15%
Wolf 15%
______

[close]


Percentage might be clearer than specific numbers, as it better indicates that the contribution of each ship can only be seen in the fleet context. Then again, maybe you meant that and I just misunderstood.

The thing here is that it's not showing the "sensor profile" and "sensor strength" values anywhere. Those seem important to show somewhere, since the hullmods talk about them.

Either that, or the hullmods need to phrase things differently, but how? Maybe not mention the values at all, hmm. This almost seems like a case of giving too much information, causing confusion, where the information really isn't useful in the first place. Sensor profile 4 for ship A? Doesn't tell you much, nor does that ship now having sensor profile 1 due to a bonus.

I'm also wondering if it's necessary to show the actual range number at all - maybe a sentence saying something like "the range at which your fleet will detect other fleets depends on the fleets' relative sensor profiles and strengths. Your fleet's sensor strength is X% of normal, and the profile is Y% of normal for its size". ("Profile"/"strength" being used here to refer to the ranges, not the actual per-ship profiles/strengths, which - not being informative - would be completely hidden.) Hmm.
Logged
Pages: [1] 2 3