CPU and Specialization

Discussion in 'FAQ & Feedback' started by Slywraith, Nov 14, 2019.

  1. Slywraith

    Slywraith Ensign

    Joined:
    Nov 17, 2016
    Messages:
    11
    Likes Received:
    11
    This is just my casual observation: most people don't like to deal with efficiency curves, algorithms, or even arithmetic with large numbers. Especially when making decisions in complex environments or building multi-faceted constructs. Most of the time, they wont even bother looking at the mundane statistics and instead 'feel it out' as these things progress. I confess to a mix of both, since both appeal to me.

    Going straight to the point, CPU feels like a complex solution to a simple problem. If you wanted to specialize a ship, you only really needed a set of blocks that accomplishes just that - a type of specialization. For example, a "Maneuverability Core" that increases the efficiency of thrusters and RCS, but limits the amount of other devices like weapons, constructors, and worsens fuel efficiency. Make it an interesting shape to set it apart. Still want those sweet high end upgrade blocks? Sure, just attach it to the ship and it improves its strengths and further increases its disadvantages, up to however many tiers you want.

    There aren't really even that many categories of parameters to improve on ships, so it's not as though it would be terribly difficult.

    + Top Speed and Manueverability
    + Weapon Damage and Fire Rate
    + Fuel and Warp Efficiency
    + Shield and Armor Strength
    + Mass and Volume Efficiency for Cargo
    + Improved Mining Rate and Returns
    + Improved Constructor Speed, Limit, and Energy Efficiency

    There are probably more exotic specializations others can think of, but those are the major ones. For example, an anti-personnel variant on weapons that specifically bumps up damage against NPCs and Players.

    I acknowledge that there is less flexibility in specializing a ship in one particular way, but I think giving ships and/or bases one specific advantage gives them more character and gives Players a reason to have a variety of craft.

    Frankly, it's also easier to understand and more likely to be properly taken advantage of by even the most casual players.

    Keep it simple.
     
    #1
    Liang, thedevilfrog, Ronewird and 2 others like this.
  2. xelthor

    xelthor Commander

    Joined:
    Oct 30, 2019
    Messages:
    81
    Likes Received:
    147
    ^ Thank you. This is what many have been preaching vs CPU. Same end result, but easier for players especially new ones to digest.

    I like that it gives ships identity and allows for greater flexibility and variety in prop design. There is a lot more that can be done and much of it can help hold the players hand as they progress. For example a mining core/ship could even come with a predefined color palette to help convey that ships design goal. I use the last example as an example, not a strict idea to implement.
     
    #2
  3. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    114
    Likes Received:
    54
    Politely I'd have to disagree that differing cores would be a simple or effective solution. Providing one 'currency' (CPU) that you get to 'spend' on the various things you want to do is very intuitive, simple and elegant. Ideally it allows you to remove all the arbitrary and counter-intuitive limits ("that's enough missile turrets for you, but feel free to add more plasma for some unknown reason"). Adding cores with bonuses is too conservative - all ships will still have to have weird arbitrary limits on guns etc.

    A system like CPU means limits are softer, so instead of everyone having the same number of guns with little bonuses creating fairly bland differences, people rock up to fights with radically different designs that emerge naturally from the single understandable limitation. It sucks when it's first added because at the moment it's IN ADDITION to all the other mechanics and limitations it should replace, and also it is woefully unbalanced and shakily implemented because it was dropped into a complex game only 5 minutes ago. If they do lots of experimentation and listen to feedback my hope is it will get simplified and more elegantly implemented, which I agree it isn't at the moment.

    My main problem is with the way they've added multiple CPU levels etc. That's needlessly complex and doesn't scale well, and its not intuitive at all :(
     
    #3
  4. zztong

    zztong Rear Admiral

    Joined:
    Apr 12, 2016
    Messages:
    1,424
    Likes Received:
    1,505
    I don't mind the notion of having different cores, but using them to create predetermined classes is usually something I find frustrating. I still chafe at having a split between CV and BA. What you suggest above, to me, sounds more like how you might want the AI of a Core to allocate power, kind of like how you manipulate power in Star Trek online.

    The game already kind of supports modularity. If I want my HV to mine, I put the miners on it. If I want to harvest trees, I take off the miners and put on the harvester. If I want to harvest XP and resources from Zirax patrols, I take off the miners/harvester, and put on guns. If I want to haul more Cargo, I take off the harvester and add some cargo extenders... but honestly, HVs suck at cargo. A tiny bare-bones CV is much better in the early game, after a POI raid and some established trade with the Polaris faction.

    This to me this was the CPU challenge to engineering. If I just want to use a Tier 1 Core for everything, how can I make that happen? I was already using modest designs, so I was close. So it turns out I can have 4 different HVs or I can try to make something modular. If I make one modular HV, I can get away with a smaller CV to handle it, making it more likely I can use a Tier 1 Core for the CV. Or maybe I just dismantle the HV and make a new one each time I need one. It also saves trying to repair them.

    So in a sense, the CPU system is forcing me to engineer, and in that regard its cool. What troubles me is that the CPU system doesn't seem to represent computing.
     
    #4
    Kassonnade likes this.
  5. Arrclyde

    Arrclyde Rear Admiral

    Joined:
    Mar 6, 2015
    Messages:
    1,737
    Likes Received:
    2,366
    It ain't such a simple solution. In a game with fixed sizes, you can have a balanced fixed number of modules. So your idea to exchange X for more power to Y works.
    But in empyrion you have all kind of different sizes. Basicallly keeping everything like block limits the same. You still can only have X numbers of turret types because of the specialization blocks you use. And bigger ships have the same limited choice of specialisation blocks. Meaning it doesn't matter, your 50 meter long CV has 20 turrets, as well as the 500 meter long ship.

    A CPU system could make things grow dynamically in size while still providing the fact that you have to choose if you want to trade in acceleration for more firepower, or more storage for less production capacity.

    But the way it is currently implemented it is "not the best way". Imo, almost the worst way. It is to static to linear and doesn't account for the games very dynamic build system. They will only differ in details, but probably ending up to have a similar size and just enough of everything. Some might have a constructor more, others have less of a certain type of weapons, while others again have mor thrusters. Building bigger makes no sence at a certain size so tier 4 ships will all ne the same.... more or less.

    Not a fan of the current CPU system. But additional specialization blocks.... i am not a fan of either.
     
    #5
  6. piddlefoot

    piddlefoot Rear Admiral

    Joined:
    Mar 4, 2015
    Messages:
    4,470
    Likes Received:
    3,572
    Whats the point of a space game that has war in it, if you have no distinct ship classes.

    CPU does NOT bring distinct ship classes.

    Instead of this CPU, if we had been given a dozen new starter blocks with attributes tied to them and now you can build a proper drop ship or space fighter or atmo fighter, man we could have had a really unique game, where fighters were actually fighters and not just a mash up of all the same garbage which, CPU, will never ever give us distinct ship classes.

    CPU could and SHOULD have been used to booster that news set of core's, giving things like wings an aerodynamic bonus etc, RCS units would not be a total waste of polygons, 60,000 odd builds on workshop would not need repairing re-working, the forums would not be full of abuse and anger over this poorly implemented CPU system, builders would not be leaving, bad reviews just a matter of time, how often have you ever seen me complain in the public feedback section.........

    CPU was / is a bad system, poorly implemented, that limits creativity and shrinks in general the size of our builds.

    Missed our mark in Empyrion by the look of it.
    Coulda woulda shoulda.
    Some other game will do it you know, sooner or later, because its the right way to do it...What you ask, flight controls and how its actually done in Empyrion since 2013.

    https://empyriononline.com/threads/proper-flight-controls-new-cores.37118/
     
    #6
    Last edited: Nov 20, 2019
    Liang, thedevilfrog and ECubed like this.
  7. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    114
    Likes Received:
    54
    Everyone agrees diverse ships, like missile destroyers and scouts and battleships, would be good.

    You can either specifically lay out the classes, defining the number (max will always get used) of guns etc each class gets to use.

    Or, you can give ships a generalised resource, and tell them "spend this resource on what you want the ship to be good at". Missiles, speed, defence, constructors, whatever. CPU is Eleon's attempt to move in this direction.

    I think predefined ship classes limit creativity a lot more than the second approach. If there isn't a predefined class that suits your idea, you can't build an effective ship of that type.

    You obviously need some kind of limitation to prevent people putting 1 million turrets on a small ship, but the first approach limits all sorts of specific parts and modules of your ship, while the second only limits stuff overall (to prevent it being overpowered).

    You're right that CPU doesn't smoothly achieve specialisation, yet. That's because the generalised resource approach, while being (imho) easily superior, is WAY harder to implement. There's all sorts of crazy tweaks and testing needed to balance things. It will be a while before it feels more liberating than restrictive, but if Eleon refine it carefully it will get there, and its worth a few months of design awkwardness to get there.
     
    #7
    Last edited: Nov 21, 2019
    Arrclyde, Vermillion and geostar1024 like this.
  8. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    6,291
    Likes Received:
    9,081
    Eleon has made it unnecessarily difficult on themselves with the way they architected the CPU system, though. I'm honestly not sure it's possible to simply tweak values into place under the current implementation.
     
    #8
    SacredGlade and stanley bourdon like this.
  9. sillyrobot

    sillyrobot Captain

    Joined:
    Aug 21, 2016
    Messages:
    462
    Likes Received:
    687
    No, people will show up in the 2-3 designs that show the most promise with respect to effectiveness in combat. The CPU limit won't enhance creativity: it will enhance efficiency of design. The game doesn't have substantial difference in combat choice: weapons have similar ranges, firing characteristics, and damage output. Defences are simple hit point sinks and evasion.
     
    #9
    Kassonnade likes this.
  10. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    6,291
    Likes Received:
    9,081
    True, which means that weapon stats will need another look (they need it anyway).
     
    #10
    Kassonnade and Arrclyde like this.
  11. sillyrobot

    sillyrobot Captain

    Joined:
    Aug 21, 2016
    Messages:
    462
    Likes Received:
    687
    If you differentiate the current weapon types, it won't help. At best, you'll develop a few different dimensions of effectiveness and will get people picking one over the other based on their own priorities (range, burst damage, limits the need for manual aiming, resource cost per shot, etc.) You'll still end up with a handful of effective designs because that's what CPU really does. It enforces efficient building and removes a few dimensions of effectiveness (multiple-layer armor for example) by making them insanely expensive compared to the alternatives.
     
    #11
  12. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    114
    Likes Received:
    54
    In EVE Online there's dozens of ships, and most have an effective role in some kind of fleet, even 'late game'. It's impossible to find a single best ship - none dominate the game. You also have CPU/power which means there's a whole range of ways to set up those ships. And it's based on only a handful of factors, like circling speed vs turret aiming, range, DPS, and a small number of modules that drain energy or slow the enemy down. There's not really stopping a similar set of factors being implemented in Empyrion, except Emp has an actual building system that could allow for much more creativity.
     
    #12
  13. sillyrobot

    sillyrobot Captain

    Joined:
    Aug 21, 2016
    Messages:
    462
    Likes Received:
    687
    It's also based on the fact that, unlike Empyrion, you are generally part of a much larger group and specialization means the battle group will work better. A group of complementary specialists perform better than a group of generalists much like a pike square fares better on the battlefield than a mob. In solo action, the reverse is true. The specialist is much more likely to be faced with situations where his capability is negated but the generalist has enough all-round effectiveness to muddle through.

    Specialisation is promoted by being a member of a large group where the specialties are seen to enhance the performance of the group. It is not promoted by design efficiency.
     
    #13
  14. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    114
    Likes Received:
    54
    Specialisation is promoted by challenges that demand 'the right tool for the job' - that could include PvE stuff in Emp. Also, a lot of gang skirmishes in Eve are for example 3v3 sized, and there's specialisation in those for sure.
     
    #14
  15. sillyrobot

    sillyrobot Captain

    Joined:
    Aug 21, 2016
    Messages:
    462
    Likes Received:
    687
    Two people working in tandem should often take on some level of specialty roles. Having a strength paid for by a weakness elsewhere is quite tolerable when one or more allies complement your design.

    I agree having a correct tool to accomplish a job is another driving factor for specialisation.

    Empyrion as it stands today really only promotes three active tool types to single player and small cooperative groups: combat hammer, resource collector, and salvage/cargo carriage. If the devs want to promote specialty, they need to provide specialty circumstance sufficiently frequently that a need is perceived. If you want turning mode to be a dimension that defines specialty then there should be opponents it is better to dogfight than go tot-to-toe with or where player evasion is both possible and better than damage mitigation. If the devs want range to be a strong consideration then situations must exist where the player can outrange his opponent and manage to get substantive effect compared to tanking the same danger at close range.

    These situations only exist accidently right now. Adding a new constraint that promotes effectiveness of design will simply provide the player base with more consistently designed combat hammers, resource collectors, and salvage/cargo carriage.
     
    #15
  16. Arrclyde

    Arrclyde Rear Admiral

    Joined:
    Mar 6, 2015
    Messages:
    1,737
    Likes Received:
    2,366
    I think the idea of having a dynamic pool of points that you can spend on specific devices (defining its roll) is a much better than a fixed class system. As already said: specialisation with cores need limits of devices in a specific roll vessel. So you have a limited pool of devices for each class making ships differ very little within a class. That would be an even more restrictive way for ship building.
    While with points system you can choose if you really want that 12 rocket launchers or if you sacrifice 4 off them to fit in another setvof thrusters or a constructor to rearm your weapons. Or have only 6 and add another shield generator (not possible now but it could be with a point system) doubling the defensive capability. You can specialize in one roll, or change focus to multiple rolls without being the best in a specific roll.

    With a fixed core class system pretty much every vessel will end up being the same. Only looks a bit different, bit size, weight, acceleration, speed and all that jazz is defined by class and is not flexible.
     
    #16
    geostar1024 likes this.
  17. piddlefoot

    piddlefoot Rear Admiral

    Joined:
    Mar 4, 2015
    Messages:
    4,470
    Likes Received:
    3,572
    Not true at all, its a matter of mathematics.

    Right now you have for example 24 modifications via CPU, with ship classes via the core, like its been from day one, ie, HV, SV, CV, you get all new cores with new attributes PLUS the 24 extra modification via CPU if its applied to things like weight and wings for lift, and things like that, it literally gives you well over 5 times more content.

    Its new content, its content that inspires new building, rather than trying to inspire people to go back and once again rapair every build they have ever made.

    People keep saying new cores limits creativity, but that a lie or a complete misunderstanding, its a huge expansion on what we have, 10 times more flexible and it would not have dropped Empyrions dedicated server numbers by, what Steam is reporting as 100% and a 10% drop in SP...
    CPU is a poor feature, poorly implemented.

    There are no real logical argument against doing it through the cores.

    This is how Eleon literally built the game when they started in 2013.

    Ship class is absolutely determined by the ships core, the starting block you layed when building it, that starter block IS what contains all of your flight control attributes, you have a whole 3 in the game for over 5 years now, 3 that are locked, and you all seem to love it, thats all Im proposing, more of that, more of what you have loved from day one, its time for proper new flight controls, proper new classes of ship and the ONLY way it can be done properly in Empyrion is how they have done it from day one, if you are at all serious about doing it right.

    I cant put it any blunter than that.

    Its very simple maths.

    CPU brings a couple of dozen modifications, the same for every ship ever made.
    New cores bring literally as many new flight models and ship classes as the devs want or that we can think up for them PLUS the two dozen odd CPU modification points, its a no brainer what system is better for the game.

    If you think CPU is a better way to do this, WHY hasnt CPU been used to define ALL classes, HV SV and CV from day one, and when you realise the answer to that, you will realise new cores really is the only way to do this properly.

    I dont know why Steam is reporting 100% drop in MP I checked servers and some are populated, but them numbers are definately down.

    Name Current Players 30-Day Avg. 30-Day Gain 30-Day % Gain
    [​IMG] Empyrion - Galactic Survival 1,789 1,338.2 -148.8 -10.01%
    [​IMG] Empyrion - Galactic Survival Dedicated Server 0 0.0 +0.0 -100.00%
     
    #17
  18. Slywraith

    Slywraith Ensign

    Joined:
    Nov 17, 2016
    Messages:
    11
    Likes Received:
    11
    This definitely got a few more replies than I expected it to. I admit I wrote off my suggestion post after CPU dropped and it became evident that I was way behind the idea to implementation curve. That being said, I felt I should weigh in one more time out of respect to those who offered their opinions.

    I was somewhat skeptical about CPU after launch, and while I still have some reservations about it, I feel that it's an adequate way to impose design limitations and invite specialization. I didn't really notice the impact of CPU very severely on my designs, and in some ways it made me a better builder. For some perspective, most of my designs only needed to be upgraded to T2, much fewer to T3, and none to T4. I don't build especially large - usually - so that may be why.

    Some of you thought that rigid classes defined by a core would negatively impact your freedom as builders, and to that I agree. It's completely static and has little in the way of advancement outside of their particular specialization. However, I did mention in my post (albeit briefly) that each specialization would be able to expand as the CPU system currently does to meet the need of larger ships. Ideally, this would allow them to support more devices as they're bumped up through tiers. If you desire to build dynamically onto a ship throughout it's life, I can see why this wouldn't be your cup of tea. I like to upgrade some of my designs throughout play and "future-proof" them accordingly, so the appeal isn't lost on me. It's fun!

    I still prefer hard limits and 'Class Cores' to the CPU approach. It's more efficient for smaller builds since it improves the effectiveness of every device in class as opposed to simply adding more devices to achieve a greater effect. Though when I say simply, I would also note that the complexity of making each new device efficient in the structure of a build is no small task, and it's certainly an intriguing challenge in its own right. Depending on device, of course. Still - it's possible to have that complexity and hard limits, as well.

    I think that my ideal would be a mix of CPU and 'Class Cores'. Soft limits on devices ala CPU and hard limits on device specialization. The best of both worlds, I think. Creativity and choice efficiency rolled into one.

    It is definitely true that the soft limit approach Eleon is taking is more difficult. It's here to stay, I think, so I wish them all the best so that we can all benefit from the end result.
     
    #18
  19. Germanicus

    Germanicus Rear Admiral

    Joined:
    Jan 22, 2018
    Messages:
    3,135
    Likes Received:
    6,104
    Only the slightest hint of restriction or rule is what drives EGS-Testers nuts:D AND leads to lengthy discussions.
    They don't like to be ruled but to rule themselves! If they only would act that way in RL Mars would be already be colonized:rolleyes:.
     
    #19
    Last edited: Nov 28, 2019
    Slywraith likes this.
  20. piddlefoot

    piddlefoot Rear Admiral

    Joined:
    Mar 4, 2015
    Messages:
    4,470
    Likes Received:
    3,572
    Imagine the sunsets on Mars , how awesome would that be to see !

    Alot of the testers are passionate about the game they love so much and want to see it get the best features possible.

    Unfortunately dude, its not up to us testers to get to Mars !
    And that is probably more linked to the lack of NASA budget ever since the Shuttle program problems.
    All the cash was sucked out of NASA vs the Cold war era.
    When you see how far ahead NASA put the USA, its utterly astonishing they dont have a BETTER budget than the cold war era right now.
    And a whole new shuttle program....
    Those shuttles put things in space, that to this day, we still cant put into space those sized objects, the ISS would not exist if it was not for that shuttle program.
    Hubble, Kepler, the list goes on and on, its brought so much knowledge and advancement , like I said, its astonishing the US dont have a similar or better program running now.
     
    #20
    Slywraith and Germanicus like this.

Share This Page