What's the formula for HURRY cost?

[Enigma]

I normally upgrade my rover or supply crawlers to synthemetal trance in the very early game. It really does not matter what you upgrade them too so long as they are very expensive. Generally it is more efficient to pack on the expensive abilities than to give it higher armor. In the very late game drop wave nuetronium is one of my favorites.

[Adam_Smith]

Frankie,

If a function increases EXPONENTIALY then the rate of increase increases as x gets larger. Your function does than. It has nothing to do with what the exponent is.
[This message has been edited by Adam_Smith (edited April 20, 2000).]
[This message has been edited by Adam_Smith (edited April 20, 2000).]

[Ned]

Enigma, Early in this thread, you stated you could update a 30 mineral crawler to an 80 mineral crawler for an upgrade cost of 90. This means the extra 50 minerals cost just less than 2 minerals per turn.

I have never experienced this upgrade cost. Can you give us exact details on the upgrade crawler configuration - armor, special abilities and the like?

I assume you would keep this same configuration througout the game even as reactors improve - b/c improved reactors lower the mineral cost of a unit, and this is exactly the opposite of what one wants in a crawler that will contribute its minerals in full to an SP or prototype.

[Enigma]

I am talking upgrading a crawler from a vanilla 0-1-1 crawler to a 0-2t-1 trance synthemetal crawler in the early game. In the later game I generally switch to rover supplies to do this sort of thing.

Unit upgrade cost is the one thing I am sure about

Cost = 20 + (newarmor-oldarmor) + (newweapon - oldweapon) + (newcost - oldcost)

For that crawler it would be
Cost = 20 + 20 + 50
Or cost = 90.

I think that the game considers moving from no armor to any armor an extra 10 energy cost.

Later in the game like I said I build high armor rover supplies with whatever expensive abilities I want. It doesnt matter- Wave/clean is completely useless but it increases the cost.

I always keep it at fission for my crawlers. Remember when you upgrade a unit it is possible to downgrade that unit's reactor. Sometimes I just use an existing crawler rather than build a new one and I often downgrade the reactor.

[Ned]

Enigma, I did some experimenting last night and discovered that the upgrade costs for the from a fission crawler to an armored crawler seems to double if the armored crawler has a fusion reactor. However, the "mineral" cost of that fusion reactor is "lower" by a considerable amount than a comparable fission reactor model. So - even though the e/c upgrade cost increases dramatically if the crawler has a fusion reactor, the mineral "benefit" (for SP's) is dramatically less.

Unfortunately for me, this is the first time I have understood this. When I research the Fusion Technology, the game automatically upgrades my armored crawlers to a fusion reactor. I have never, at least to now, downgraded its reactor.

For any who really want to play this game at its highest level, this information is critical.

[Adam_Smith]

I'm trying this strategy now. I built a regular destroyer supplier. Next turn I will upgrade it to a destroyer crawler with 3-res armor and trance enhancement.

What is the difference with the destroyer chassis to the rover with this strategy. Since destroyers are even more expensive than rovers I imagine that I will get quite a bit of minerals this way.

[Ned]

Adam, I haven't tried destroyer suppliers - but watch out. The trick with the regular convoys with a fission reactor is that their cost is high but their upgrade cost is low. Once you have this figured out - the game becomes even more remarkably easy.

In my current game, I have created fission, trance, ecm, pulse armored crawlers. These cost 144 minerals. Their upgrade cost from conventional fission crawlers is 160 e/c. This is just over - and think about it - one credit per mineral!

I was able to complete a 300 mineral SP in one turn using only two of these for a total cost of 320 e/c. I was similarly able to complete a 400 mineral SP with only three of these upgraded crawlers.

This is almost cheating!

Ned

[Enigma]

Ned-
I can not imagine having to build SP's the "normal" way after I started using this tactic. It gets EVEN BETTER!! Throw in the Nano factory. Although it is pretty late in the game this can be a pretty significant difference. Imagine buying 1000 minerals for only 500 energy!! Very lucrative. You can disband these crawlers to rush buy units at a drastically cheaper cost.

[Adam_Smith]

Ok, here is the results from my experiment. First of all the destroyer supplier is 42 minerals. It cost me 170 EC to upgrade it to a trance 3res.

By the time I got it over to HQ the project was over half done, so maybe I should try it out with a supplier already there to add on to production at the beginning. Anyway, it was only 372 EC to hurry the project. When I added the supplier I only had 15 minerals left and I build it next turn. So add it up and it's 42mineral+170EC compared to 372 EC. That seems financially sound when you consider that my HQ is now free to build something else.

One thing that I realized is that it is better to add the crawler at the beginning, especially if you have +4 Industry and good mineral production. Then when the project is cheaper you can rush it or just wait to build it.

[Ned]

Adam, I was right. The destroyer crawler was not as efficient at this as what I described. Given the numbers, you had 93 minerals to complete the SP. Your destroyer added all but 15, meaning 88. Since it originally started with 42, the upgrade to the destroyer added 46 minerals at a cost of 170 e/c. This is still roughly 4 e/c per mineral, nearly the same as the SP mineral cost of 4.

Now compare my example. Crawler at 30. Add 114 minerals at a cost of $160. Crawler at 144 minerals. The e/c cost is about 1.5 e/c per mineral.

Enigma's example is 2 minerals for each e/c.

Enigma, how did you do that?

[MariOne]

Gentlemen, I bump this thread up, as I was addressed here when I raised the issue of efficient hurrying techniques with a pactmate (in a pbem which was started back in april99 at alpha.owo/multiplaying!).

I sent him a couple of advices and hints, and he said that "my formula" was not so different from the result that you got to after a long long discussion (!!!).

My first comment after browsing thru thius thread (I really couldn't read into all the incredibly misleading details) is:

I'M TRULY APPALLED

What amazes me most, is that 5 minutes (10 at most) of very simple testing with a Scenario Editor would have been enough to avoid many very superficial and sometimes grossly wrong statements.

First, credit to Frankie for having found the most approximate algebric formula for it. With a little twitch it would have been perfect.

Second, blame on all of you for not haveing once verified the actual completion costs you were reporting.

Helium Pond reported a pattern he saw posted somewhere. Didn't it dawn to verify it in a sistematic way?

Many times you all wrote, descending down from this original quote, that "the last 5 minerals cost 2ec each"

This PROVES that NONE of you actually ever completed a unit paying for his last 5 minerals missing to complete it.
Had anyone bothered to test what he wrote, he'd have easily seen that hurrying 5 minerals out of 5 missing, that is "completing" the last 5 minerals for a unit, costs 11ec and NOT 10ec.

What's with your to such length discussed formula now?

Let's get some things straight.

We must avoid confusing complete payment with partial payment.
We should concentrate first on complete payment, then wipe any misleading assumption you made for partial payments.

So, for now, the MAIN variable here is the cost you have to pay for M minerals, when those M minerals are those needed to complete the unit under production, disregarding the current mineral production from that base.
That is, if you have a *unit* costing 30M, and you have accumulated 10M, we want to know how much will cost to pay for the whole missing 20M.
BTW, some of you accidentally got there, that cost is 60ec.

Frankie's formula is correct, only that it yields decimal values. Maybe he's got it for implicit. Actually, SMAC sometimes round up, sometimes rounds down, sometimes rounds to the nearest, so it would be useful to assume that those values have to be rounded down. That is, you can use the operator INT.

So, Frankie's revised formula is:

CH = INT(¹/₂₀M²+2M)

where CH = cost for Complete Hurry

If you substitute M=5, you'll get INT(11.25), that is 11 indeed. Note that on some values of M the CH decimal value will be much nearer to the upper integer, so it could have been natural thinking that the cost would have been rounded up (after all you could expect that you'd have to "cover" a price). Which instead must NOT be done.

It's easy to determine at least the true costs. All you have to do is start a simple game, press Ctrl+K, eventually add some tech for expensive units (but already the initial colony pod will get you up to 30 missing minerals), get in the base, "Edit Base", "Edit Minerals", change the amount of accumulated minerals one by one, hit Hurry and write down the costs the game demands you to pay.

That was all that was required for a simple, scientific and exhaustive approach.
It would have spared to many of you the embarassment of making wrong statements.

---

A couple of considerations, as there's a math teacher reading me, while I'm just an amateur.
Reporting the observed values of a function, I don't find a formula.
But, if the function has a finite number of input values (can't recall whether it's the domain or co-domain...), enumerating the eoutput values, that is writing down the input-output table, I should be DEFINING the function as well as if I had written the formula.
If the function has not finite input values, but those are discrete values, for instance the positive integers, the the "principle of induction" (is that it's english name?) will allow to state the rule for the initial value, and the rule to derive the subsequent values from the previous, and I'd have defined the function as well.
In short, if I can show a pattern in the values, I won't have to chart the whole infinite table.
Now, I'll list the first values DETECTED, to let you have an idea.

M CH

1 2
2 4
3 6
4 8
5 11
6 13
7 16
8 19
9 22
10 25
11 28
12 31
13 34
14 37
15 41
16 44
17 48
18 52
19 56
20 60
21 64
22 68
23 72
24 76
25 81
26 85
27 90
28 95
29 100
30 105


What you reported, about the last 5 minerals costing 2ec each, the next 10 3ec each and so on, was CLOSE to the reality, but nevertheless it was NOT CORRECT.

You'll observe that approximately the cost to pay for all the minerals you miss, increases almost like you said, for each more mineral you need to complete.

Firaxis added tho a strange twitch.
While paying to complete 4 missing minerals costs 8, paying to complete when you miss one more costs 3ec more, NOT 2. Then missin one more, i.e. 6, you have to add just 2ec again. the up to 14 minerla missing, the cost to hurry each time the WHOLE minerals you miss goes up by 3 ec indeed. There you have that for 15 you have to add 4ec to the 14M value, then again 3 ec, then again 4, and so on (that's the induction pattern...).

The other major misleading thing you write, is for instance that "the next 10 minerals cost Xec each".
That's WRONG. What you should say is that the COMPLETION cost for the WHOLE missing minerals increases by X for each more mineral you miss.
Frankie could confirm you that it's not the same thing.

This finally leads us into the issue of partial payment.

Imagine we miss 20 minerals to complete a UNIT. We know now, and we all agree, that to hurry all those 20 minerals out of 20, you have to pay 60ec.

Suppose now that (for any reason) you want to pay only for 5 of those 20 minerals.
How much will these 5 M cost?
From what you say, someone actually interpreted that as they were tha last 5 minerals you needed apart that base production, they costed 2 each = 10ec. Other players actually said to me that the 5M were the "first", and after them you still missed other 15, so you had to count out the "last" 5, then the "next" 10, and there were the 5 you were paying to, at the cost of 4ec each = 20ec. That is, like you took away the HC for 15M form the HC for 20M.

Of course many of you already know that this is a wrong approach. But what pisses me of, is that the way you presented the matter here is misleading, and that it indeed mislead an experienced palyer to the abovementioned wrong interpretation.
That's what drove me to this long rant/lecture, which otherwise I wouldn't have dared to write

How does it work?
Once you know that you're missing M minerals, for that hurry the cost per mineral is FIXED, and each of those M minerals costs HC/M ec!!!
In our example, when you miss 20M, the CpM is FIXED at 3ec! For every mineral of those 20!. Thus, if you eant to pay for 5M OUT OF THE WHOLE 20 you miss, they'll cost 5*3ec=15ec. If you want to hurry 15M out of 20, they'll cost 15*3ec=45ec.
Can you see it? If you have to hurry 15M OUT OF 15, they instead cost 41ec!

So, the CH is determined by the M missing minerals, and within this M minerals the CH is equally distributed. That is, the # of M determines at the same time the CH and the CpM.

Indeed you can factor out one M from Frankie's polynomial, as the known term is missing.


CH = INT(M(^M/₂₀+2))


Frankie will be able to explain you better why you can't take out the factorded M from the INT operator, as it's multiplied with a decimal value...

So, true CpM is only approximated by 2 + M/20.
When M is a multiple of 10 it yields correct values.
That is, when you miss 10M each costs 2.5ec, when you miss 20 they cost 3ec each, when you miss 30M they cost 3.5ec each.

You can see that it's the Cost per Mineral which is approximately linear dependent from the total M missing to completion.

FY convenience I paste here this table too

M - CpM
1 2.
2 2.
3 2.
4 2.
5 2.2
6 2.17
7 2.29
8 2.38
9 2.44
10 2.5
11 2.55
12 2.58
13 2.62
14 2.64
15 2.73
16 2.75
17 2.82
18 2.89
19 2.95
20 3.
21 3.05
22 3.09
23 3.13
24 3.17
25 3.24
26 3.27
27 3.33
28 3.39
29 3.45
30 3.5
31 3.55
32 3.59
33 3.64
34 3.68
35 3.74
36 3.78
37 3.84
38 3.89
39 3.95
40 4.


you see that Hurrying a unit when you miss more than 40M, gets even more expensive than hurrying a project.

For the rest, it's very simple, and you already wrote it.
If you have less than 10 minerals accumulated, the cost for Hurrying gets doubled, and that's true for every item.
Ther's no other condition.
That is there no additional penalty if you have 0M compared to 1M or 9M.
your factions & SE setting have no influence, not if you'r prototyping the unit.
Yes, your Industry setting has NO DIRECT influence. Of course, it will change your unit's cost, and thus if you flip it back and forth in the same turn (shame!!!) the # of M you miss will change.
But as the table is related to the # of M, we don't care here how you caused its value once it's fixed.

---

A word to Ned about his (smart indeed) facility overpayment technique.
You pay facilities minerals 2ec each. you pay Project minerals 4ec each. If you switch from facilities to Project, your accumulated minerals (above 10) get halved. That is, it's as if you'0d payed double for each of them, that is 4ec each indeed, like for the project!!!
Taking your example, you payed 1120ec to go from 10M to 290M, you added thus 280M, and if you do the math you payed exactly 4ec each.
One thing to credit to your technique, is that this way you can pay Project M 4ec each right from the 10M limit on, avoiding the double cost belt concerning Projects.
Who said that it's at 8ec/M for the lower 10% M, was wrong. The 8ec/M limit for projects, it's related to the first 4 M ROWS accumulated, and is NOT RELATED to the overall project cost.
This is a peculiar condition, as M in mineral ROWS ARE related to your SE Industry setting. I leave to you to furhter investigate....
Only to be noted that under the first 10M it's the 8ec/M cost which gets doubled, bringing it to an impossible 16ec/M cost for a project with less than 10M accumulated.

---

About Upgrading costs.
Enigma, I learned the exact formula here in apolyton from another thread, I verified it for very different kinds of units, and it alway worked.
Offhand I'd say that your formula is correct, but it's a Specific formulation which applicability is limited to crawlers costing 30M.

The Upgrade costs formula is:

UC = (WeaponIncrease+ArmorIncrease+NewRowsCost) * 10ec

- it is NOT related to your SE Industry setting, you alway pay 10ec for each row you new unit costs, regardless of the rows length
- it is NOT related to the cost of your old unit, you ALWAYS have to pay for ALL the rows your new units costs

remember that when you upgrade you can't change:
- chassis
- artillery ability
- equipment (between them or with weapon & vice-versa)
and that you CAN'T downgrade weapon or armor, but you CAN downgrade reactors.

As when upgrading there is an additional cost when you increase weapon/armor, in principle it should be convenient to prepare units to upgrade without abilities and with the highest possible weapon/armor already, but that's to be finetuned case by case.
And, of course, multiple upgrades of the same individual unit it's the worst way to waste your ec I can figure, aside gifting them to your enemy.

---

Purely specualtive, I found out that at the end of the game, if upgrading crawlers for the projects is disallowed, if you upgrade form a Good-Good-X plain unit to a Best-Best-X +CleanDrop unit, even losing half of its minerals when you disband it into a project it will grant you a cost per mineral lower than 4ec.

---

Before Fusion, the most cost-effective crawler upgrade is from your plain SC-1-1 to ECMTrance SC-3+t-1. (before fusion using rover crawlers is a true waste, do the math)
After Fusion, as now also Speeder Crawlers cost only 3rows, the MOST cost-effective crawler upgrade of the whole game is from SC-1-2*2 to CleanDrop (or ECMTrance) SC-1-2*1, for a cost of 12ec per added Row (then it depends on your SE Industry setting).
If someone can get below that cost for cralwers I'll publicly declare what he'll ask me to .

---

True, as the NanoFactory cuts youe upgrade costs, it makes this technique *de-va-sta-ting*. It's almost a cheat, and some valuable SMACer as Bingmann indeed consider this whole technique as cheating, although I disagree.

---

Thank you all for the attention you gave to this record-long post from me.
I did it because I thought that I might contribute to the technical understanding of this game, and I did it here because thhi si the most serious site to discuss about SMAC facts.

If you disagree with my firm conclusions, I'd say that they are not my opinion, but FACTS which can be verified by anyone with some testing in the Scenario Editor.
So I'm open to confutations, after all I always made lots of mistakes and I'll keep doing them, but before telling "I think you're wrong", please do the TESTS, AND do the Maths.

MoSe

[edited to fix typing errors with the sub/sup HTML code, and forgive me for the many typos I still didn't hunt down]
[This message has been edited by MariOne (edited July 17, 2000).]

[MariOne]

Enigma,
I don't have SMAX, so I don't know if "pulse
"armor" is something new, or if you meant plasma armor.
Indeed an ECMTrance SC-3+t-1 crawler (considering this can happen way before Fusion) costs 160ec to upgrade form a plain basic crawler. Its new cost is 14 rows, thus you added 11 rows from the 3 present in the original crawler.
I don't think that you should consider SE Industry switches when you evaluate the convenience of upgrade costs.
And I can't see where you too that 144 minerals figure from. It could fit with 18 rows of 8 minerals. Or with 16 rows of 9 minerals. but I can't figure how to fit it with your crawler upgraded from 3 to 1r4 rows.
Anyway, there you had paid 160ec (regadless your Industry SE setting, for upgrades too) to add 11 rows, thus you paid 14.5454... ec per Row. This is indeed the most cost-effective crawler upgrade I was able to figure out before the discovery of Fusion. The cost per minerals has to be determined according to the number of minerals per row you have when you actually cash in the crawler.
Hive or Sparta apart, this can range from 8 to 10 in the most cases. So your cost per mineral can go from 1.4545... to 1.8181...

With Fusion you can get down to 1.2ec per mineral, with standard industry.

I realize now taht I didn't perform extensive tests with Quantum, so maybe my bet in the previous post was a bit bold... ;^)

[Helium Pond]

Nice work, there, MariOne,but watch it calling people embarassed. I said right up front there that my formula was wrong. I gots no "embarassment at making wrong statements." You say, "Didn't it dawn to verify it in a sistematic [sic] way?" Oh please. Sure, it "dawned." Then I thought, hmmm, sounds like work. Mayyyybe I'll just let someone else do it. And look, you did! Thanks again. Excellent work.

[Adam_Smith]

MariOne,

There is nothing wrong with the answers that we came to in this thread. The information here should give the reader a correct understanding of upgrade costs. We did do plenty of experimenting. There is no exact formula.

[MariOne]

HP:
True, you stated it upfront.

quote: These formulas are accurate for me about 70% of the time. I would say the facilities and Secret Projects are always right, the unit costs seem to be really close but sometimes off.

Anyway, not all THAT work was required. It took me 10 minutes to chart the values up to 40 mineral missing and to devise the underlying pattern. Reflecting on it a bit more allowed me to verify the trend for higher values, ad to organise the matter in a systematic way ( ). It did it for my own interest, I'm glad that my work is useful for others too. (It took me a LOT more time and effort to write my big post here! )

AS, I'm sorry but I dissent:
Frankie got to the right answers at the end, despite the many objections he had to come over.
I found the approach you took and some conclusions, misleading. Indeed, they actually mislead someone.
I recognize, when one has found the way the things ARE, all the random trial and errors and unverified assumptions seem aimless, but each one has his way to get to things.

I thought that the main issue of this thread were the hurrying costs. Upgrading costs were a side issue. So, I'm not sure if I got your reply right.
Anyway, as for Hurrying costs, my opinion indeed is that the information here will make the reader bump into many walls and wrong paths before finding the right way.
About Upgrading costs, Enigma's reported formula was a paritcular, and not general case. The formula which I simply reported, I read it form these forums sometimes ago, it didn't come from me, and I don't take the merit for it. I simply used it, and verified it worked.

You wrote "There is no exact formula".
For Units Hurrying costs, Frankie and I got separately to the same formula, and answering to this thread I put it down in its final form, using the INT operator (integer part).
That formula IS exact.
For Upgrading costs, the Apolyton formula I reported IS exact, at least for the many tests I performed with it, checking all the significant situations I could think of.

I'd be curious, at the end of this duiscussion and all that has been said about the approach to such issues, to know how did you got to make that statement. Common sense should require that you have found at least ONE case in which the formula we provided (btw, which one did you refer to?) was not exact. If you have found such case to support your statement, please show us and I'll be the first to reconsider. Or did you say that there is no exact formula (and with this you include also the "exact" formulas we reported) out of an impression?
---
About the specific issue of crawlers upgrades, one can't deny the simple maths which state how many ec you paid for each mineral you gained, in the several cases you propose, and that should determine which is the most efficient option. Then I admit that many variables may influence the optimal choice, which is not always availabe. And that each one may have his opinion about what's better to do in the actual circumstances, there is room for different playing styles and personal preferences.
I'd only care to add that IMHO, personally I would NOT advise someone to take THIS thread as reference for the best way to use crawlers and upgrade them.

---
Thank you for the kind words, I don't think I did anything special.

MoSe

PS: Helium Pond, I'll add to the proper thread, but I tell you from now that I also find your work with the tech trees outstanding.
From the beginning I put myself at work on a similar endeavor, but that was harder, and I left it somewhere halfway. Now maybe your thread will give me the impulse to finish it and offer it too the the community.