Creep Boost discussion


  • Culture

    The increased carry capacity means you can be at a certain location longer, meaning you can get more value of other boosts.

    Example:

    You want to build up a wall, lets say you can drain your CARRY in 20 ticks

    Your trip from storage->wall is 20 ticks

    if equally balanced with move it will take out 40 ticks per energy-pickup trip.

    If you've boosted WORK you waste about 1/3'd of the time traveling back and forth. If you boosted your CARRY to 2x this is reduced to 1/6th.

     

    I personally think this is the kind of stuff you should be able to use your carry boost for. You can also use it to jump-start your new rooms. Now it also requires a shitton of move boost which makes it to expensive to even use (more expensive than XGH2O).



  • That's right Dissi, I did not consider that scenario. With the weight penalty removed from CARRY that might be enough to make it worthwhile for those situations today. But with the added weight we have today I don't think it is.

     

    I'm not sure what to do with MOVE but I agree with Atavus that we should be careful about buffing any combat boosts, and MOVE can be both a combat and economic boost. We might have to accept that MOVE will not normally be useful in economic situations.

     

    2x HARVEST might just work as well... Since a WORK part is twice as expensive as a CARRY part you gain 2x as much by boosting its efficiency. You're also getting more WORK for less MOVE with that one.

     

    Artems suggestion might just work, sounds like a good start at least. (Perhaps with the exception of the MOVE buff)


  • Culture

    > 2x HARVEST might just work as well... Since a WORK part is twice as expensive as a CARRY part you gain 2x as much by boosting its efficiency. You're also getting more WORK for less MOVE with that one.

     

    Currently you have a 300% increase. With this you still need ~3 mining parts for mining out 3000 capacity sources. With this x2 you can come by with 2x harvest we should have a 600% increase in mining. This allows for a single WORK to mine out sources with up to 3600 capacity.

    This should make it worthwhile to get this going 24/7. You can effectively spare out 5 WORK with minerals this way. Saving 500 energy per  generation per source.

     

    The move boost seems fine as it is now, with the removal of extra weight by CARRY it might have more uses now.



  • I am now at home and I have decided to add some more maths for the T1 CARRY boost. This time I'm just showing the formulas I came up with as it might be a good idea to discuss them and what numbers are most reasonable to calculate with.

     

    Lets compare everything with the current credit cost on the market. If we want to we can easily convert credits back to energy again at the end:

    COST_OF_ENERGY = MARKET_PRICE_OF_ENERGY = ~.2 (At the time of posting)
    COST_OF_MINERAL = MARKET_PRICE_OF_MINERAL = ~.5 (At the time of posting) <-- You can easily expand the formulas for T2-T3 boosts by being more specific with mineral prices. I think Catalyst is still more expensive than this for example.

     

    I think you need 2 base minerals to create a boost but please correct me if I'm wrong:

    COST_OF_T1_BOOST = 2? * COST_OF_MINERAL + LAB_REACTION_TIME = 1 + LAB_REACTION_TIME

     

    COST_OF_T1_BOOSTING = 20 * COST_OF_ENERGY + 30 * COST_OF_T1_BOOST + EXTRA_TRAVEL_TIME_TO_AND_FROM_LAB = 4 + 30 * (1 + LAB_REACTION_TIME) = 34 + 30 * LAB_REACTION_TIME + EXTRA_TRAVEL_TIME_TO_AND_FROM_LAB = 34 credits and ? amount of ticks = 170 energy and ? amount of ticks

    I won't try and guess the amount of ticks at all here, I realize the formula above is slightly wrong in that regard also because LAB_REACTION_TIME and EXTRA_TRAVEL_TIME_TO_AND_FROM_LAB can kind-of happen in parallel. To make this formula more correct you'd probably have to replace LAB_REACTION_TIME with some kind of opportunity cost of what you consider using lab time is worth (you might have to consider that this time can be spent on something more worthwhile). Although if your labs would just be idling otherwise that cost is 0 I think.

     

    The value of a carry boost can certainly be discussed... But I think that at least in the case where you have a choice to add more CARRY parts (without having to sacrifice anything else, because of 50 body part limit) the valuation below should hold true. Please correct me if you think otherwise.
    VALUE_OF_T1_CARRY_BOOST = CARRY_PART_COST + BODYPART_SPAWN_TIME = 50 * COST_OF_ENERGY + 3 ticks = 10 credits and 3 ticks = 50 energy and 3 ticks

    Edit: This is probably the absolute worst case, and maybe it shouldn't be a good idea to use boosts in this scenario. But I'd like to see some actual formulas for best case values. Such as FAT repairers or remote builders that can have more WORK and MOVE instead of extra CARRY parts because of the boost, therefore being able to WORK more and spend less time carrying energy back and forth. I might try to add some later but that formula will depend on task (repairing/building and distance between workplace and storage)

     

    Conclusion:

    COST_OF_T1_BOOSTING = 34 credits + ? ticks = 170 energy + ? ticks

    VALUE_OF_T1_CARRY_BOOST = 10 credits and 3 ticks = 50 energy and 3 ticks

    Even if we double the value of a T1 carry boost it's still a net loss. (At least for this worst-case valuation of the boost)

     

    Please correct me where you think I'm wrong.

    My opinion is that there should be economical boosts available that are more viable to use than directly selling the minerals on the market for credits/energy.



  • I have started doing some really deep calculations on the cases of fat (50 part) builders/upgraders/repairers, their optimal body layouts with and without boosts depending on what distance they need to work at. And what the cost benefit/loss of using boosts will be...

     

    I might post it here later on once I've finished it and cleaned up my thoughts more. Anyway, I just realized that it's very inefficent anyways if these guys have to carry anything at all. It will be much more efficent to program pure carriers that supply them. It will always be the most efficent if the worker has just 1 CARRY part with one or several suppliers with purely CARRY and MOVE for it.

     

    Therefore boosts on those units is NOT the most efficent use-case at all. It might seem convenient. But it's more efficent to program specialized carriers for them and then *MAYBE* boost the specialized carriers. Basically the best economical use-case for the CARRY boost has to be pure carrier units. If you use the boost on some other unit it should be a suboptimal economical strategy compared to a proper supply-fleet. Boosting your sub-optimal "remote" workers with CARRY boost might still be useful. But it should be more useful to program a supply fleet for them.

     

    But... in the case of specialized carrier units.. the value of the boost is just what I said. The "worst case" value. A T1 boost is just equal to buying 1 more CARRY part. The only gain is less CPU and path-blockage because of potentially fewer units.


  • Culture

    Less needed creeps would be a welcome change 🙂

    http://i.imgur.com/CwjR6Rn.png



  • Hehe yeah definately for you big players. But I feel there should be more incentive for us new players to get into the "boosting business". Edit: There is actually incentive already with the upgradeController and repair boost. But HARVEST and CARRY seem vastly underpowered in comparison.

     

    bonzaiferroni has done some excellent calculations here on the optimal use-case of the T3 upgrade boost: http://screeps.com/forum/topic/860/When-is-boosting-useful- . Although he has not tried comparing cost against market prices but instead calculated what it costs to produce purely yourself. I'm pretty sure that is what most high level players do though and that it must be less efficent to buy those same minerals from the market. Otherwise I have no idea what drives up the market prices on all minerals and who's buying, because there are no gigantic wars going on right now as far as I know?

     

    It seems like the first answer, "always", in that thread is wrong. Boosting is only useful during attack/defense and when using the upgrade, repairing or building boost. (Building boost has very short-lived uses though).

     

    It doesn't feel right that the HARVEST and CARRY boost seem so underpowered in comparison. I think the CARRY boost might need to have both the fatigue increase removed AND be increased 2x in size or something in order to even be comparable. Maybe it's something else that should be adjusted too. Maybe slightly lower lab boosting costs? Maybe they should depend on wether you're applying a T1,T2 or T3 boost?

     

    The major problem with these boosts seem to be that there's a HUGE value in increasing energy effiency. Basically if your creep can spend it''s whole life WORKing with the boost you will profit ticksToLive*boostEfficency. This profit is HUGE compared to the relatively small gain of being able to use less body parts.

     

    I might have to compare what happens with efficency too if I go straight for T2 or T3 boosts instead of T1 but I think I've gotten tired of math for today... 😛



  • Very quick estimation of profit summary for economic boosts:

    Legend:

    CREEP_WORKING_RATIO = The amount of time a creep spends working. Not idling. Hopefully you can get it as close to 1 as possible.

     

    T1 upgrade,repair,build: CREEP_WORKING_RATIO * ticksToLive * 0.3 = ~450 energy

    T1 carry: CARRY_PART_COST = 50 energy

    T1 harvest: WORK_PART_COST = 150 energy (You need one less MOVE)

    T2 upgrade,repair,build: CREEP_WORKING_RATIO * ticksToLive * 0.65 = ~975 energy

    T2 carry: CARRY_PART_COST * 2 = 100 energy

    T2 harvest: WORK_PART_COST = 300 energy (You need two less MOVE)

    T3 upgrade,repair,build: CREEP_WORKING_RATIO * ticksToLive * 1 = ~1500 energy

    T3 carry: CARRY_PART_COST * 3 = 150 energy

    T3 harvest: WORK_PART_COST = 450 energy

     

    Those are a basic estimation of current boost profit values. It's not factoring in costs, but they differ way less than profits.

    If we wanted to buff everything to the same effectiveness value, we would have to buff harvest by 3x and carry by 10x. But buffing harvest more then 2x seems silly... But I think carry might need  removing both fatigue and buffing extra carry value by 2x. Definately remove fatigue increase on it.



  • I think these points seem reasonable:

    • Remove weight penalty from CARRY boosts
    • Increase MOVE boosts effectiveness 2 times
    • Increase harvest boosts effectiveness 2 times 
    Artem from Screeps

     

     

     

     

     

    I like these. Simple, and effective. 



  • I can never make formatting work.