EPT[q] Cost of worker loss compared to...
| Forum Index > Brood War Strategy |
|
DeepGreen
United States175 Posts
| ||
|
SiegeTanksandBlueGoo
China685 Posts
| ||
|
rkarhu
Finland570 Posts
| ||
|
Durak
Canada3684 Posts
Edit: Crap, you weren't talking about using it to scout :D. I suppose you could think of it as 50 minerals + the mining it could have been doing for the rest of the game. | ||
|
Avidkeystamper
United States8552 Posts
| ||
Xeofreestyler
Belgium6771 Posts
after a series where bisu used dts if im not mistaken! | ||
|
freelander
Hungary4707 Posts
On November 03 2008 02:25 DeepGreen wrote: I was wondering if somebody smarter than me could tell me what the cost of losing a probe is to an early game harass. Obviously the probe costs 50 minerals, but if a zealot snipes one (let's say when you have 15 probes, just for simplicity's sake...clearly 1 probe matters less when you've got 4 base saturated.), you also lose mining time. I'm assuming then, this number will be a fraction. How many minerals do you actually lose if you lose a probe in the early game? If a zealot snipes one of your a probe, and in that moment that zealot also dies, you are at 50 mineral advantage. Though this is turning into a disadvantage quite fast, because of your opponent's extra probe. 50 mineral is like 6.25 mining circle for his extra probe and after that he is gaining advantage I think. This was very rough estimation, it assumed that yours and your opponent's probelines weren't saturated. If they are, it is less disadvantage. I don't know exact numbers but one mining cycle is like 15 seconds, so you are losing ~0.5 mineral per second if you lost a worker early game (e.g. losing a building scv ) until your worker/patch ratio goes above 1.5-2, where it gets saturated. | ||
micronesia
United States24682 Posts
In pvp, if you harass with 1 zealot and 1 probe, you are obviously way ahead if you snipe 3+ probes. What about 2 probes? 1? Obviously you are behind if you snipe 0. Assume you lose the zealot and scouting probe. How do you decide who ended up ahead? | ||
|
KlaCkoN
Sweden1661 Posts
| ||
|
megastarcraft20
United States74 Posts
On November 03 2008 02:42 rkarhu wrote: The answer to your question and all things in life is 42. Correct. Even the answer to " How many chucks could a woodchuck chuck if a woodchuck could chuck wood? " is 42. | ||
|
freelander
Hungary4707 Posts
On November 03 2008 03:03 micronesia wrote: It's nearly impossible to come up with a proper analysis of the effects of sniping individual probes early in the game on the rest of the game. However, experienced players can make judgment calls pretty easily. In pvp, if you harass with 1 zealot and 1 probe, you are obviously way ahead if you snipe 3+ probes. What about 2 probes? 1? Obviously you are behind if you snipe 0. Assume you lose the zealot and scouting probe. How do you decide who ended up ahead? If you don't count fighting units: To be ahead, obviously you have to kill more workers than the number of your harassing worker. If you does count them: Theoretically, in the long run any worker loss hurts you much more than unit loss, because the worker production is being much more limited than unit production AND your saturation comes later. I saw a build made on this fact: In 2-3 years old Nal_rA PVP blizzcon matches! Look for the replays. He had a build on that 2 players map, where the guys spawn at bottom left-top right. I forgot the name of the map. He sent an early probe, one gate proxy, zealot harassment constantly. He usually sniped 1.5 probe with every zealot he sent, and meanwhile teched to standard 2 gate goon with reaver (without cutting probes, very important). But in practice, you have to take into consideration the current value of the harassing units, that's quite hard. (e.g slow zealots are worth less and less, first goon is worth a lot) | ||
|
SiegeTanksandBlueGoo
China685 Posts
| ||
|
micronesia
United States24682 Posts
On November 03 2008 03:48 SiegeTanksandBlueGoo wrote: Well actually since the efficiency of every worker decreases as you make more, I'm pretty sure you need some calculus to do this. I'll get on it once second semester rolls around or my college EA decision get back. It's not that simple. This is a combination of several fields of mathematics at once, combined with a tricky model. | ||
|
x89titan
Philippines1130 Posts
| ||
|
x89titan
Philippines1130 Posts
On November 03 2008 03:53 micronesia wrote: It's not that simple. This is a combination of several fields of mathematics at once, combined with a tricky model. its called the law of diminishing returns. | ||
|
micronesia
United States24682 Posts
On November 03 2008 04:02 x89titan wrote: its called the law of diminishing returns. No, it's called <several things including the law of diminishing returns as one part of it> | ||
|
hiroxx
Ireland115 Posts
| ||
GHOSTCLAW
United States17042 Posts
Rough cut though, you can estimate that every probe killed in the first 5 minutes is worth double the price (100 mins), in the same way that you can assume that gas is worth 3x minerals. The figures above are kinda sorta random, but it lets you establish even a "reasonable guess". As freelander pointed out, you actually don't care how much every probe costs, but you care about the difference in mining time; Aka how much is your opponent getting ahead with each probe he kills. If you kill the same amount of probes that he kills, your life is much better. | ||
|
DeepGreen
United States175 Posts
On November 03 2008 04:20 waterGHOSTCLAWdragon wrote: As freelander pointed out, you actually don't care how much every probe costs, but you care about the difference in mining time; Aka how much is your opponent getting ahead with each probe he kills. If you kill the same amount of probes that he kills, your life is much better. Yeah, I'm curious to see just how close 1 probe kill early on comes to being worth the cost of a zealot. | ||
|
GHOSTCLAW
United States17042 Posts
On November 03 2008 04:44 DeepGreen wrote: Yeah, I'm curious to see just how close 1 probe kill early on comes to being worth the cost of a zealot. depends at what point in the game is the easy answer. I believe that's a poorly phrased question... | ||
|
Pyro]v[aniac
United States147 Posts
On November 03 2008 04:44 DeepGreen wrote: Yeah, I'm curious to see just how close 1 probe kill early on comes to being worth the cost of a zealot. you would have to calculate the cost of a probe in terms of "peon time" or the amount of time each probe mines average in the time it takes to build a worker. every time a worker pops youve lost that many minerals untill satuturation, at which point the average amount of minerals mined per worker goes down about 20% each worker you make. obviosly if you lose 20 probes to a storm your way the hell behind, but early game when every unit counts it is harder to decide when your behind. try this: loss = rN + 0.5gN(N+1) = N(r+0.5g) + 0.5gN2 Where: N = number of workers killed r = cost of replacing each worker g = minerals a worker can gather in the time it takes to build a worker | ||
|
freelander
Hungary4707 Posts
On November 03 2008 04:17 hiroxx wrote: honestly - who cares? honestly, why the fuck are you posting here. honestly, there are the ten commandments of tl.net what you should read once before you start posting. honestly, there are at least 10 people in this topic who posted actual thoughts, opposed to you. | ||
|
gm.tOSS
Germany898 Posts
| ||
|
Empyrean
16987 Posts
On November 03 2008 05:26 gm.tOSS wrote: This question can only be answered (more or less correct) in specific situations, but not in general. This question can be solved in general solutions. All you need is the marginal utility in terms of minerals/minute of having an extra probe at various points (e.g., how much more useful is the fifth probe when you have four, how much more useful is the ninth probe when you have eight, etc., in terms of minerals/minute). You could set up a function for both sides that tells you how many minerals/minute you're losing, then take the difference of the functions and integrate them. It's an easy concept, but very difficult to apply in practice if you take into account, for example, having eight probes target the zealot, etc. The models start getting very complex, and honestly, at this point it's probably easier to just use intuition. | ||
|
mjh
United States133 Posts
say X is the number of probes you have, min(X) is the rate of minerals, T is the time it takes to make a probe, and X >= 15 (you lose your 15th probe). the overall loss is then T*min(15) + T*min(16) + T*min(17) + ... = T*(min(15)+min(16)+...) = loss. min(X) is a decreasing function, but this quantity is going to be more than 100, which is the amount your opponent needs to spend to make up for a lost zealot. yeah there's other factors (some minor and probably negligible), like your opponent having 1 less zel, you have 1 more supply, etc but in terms of minerals sacrificing a zealot for a probe is probably worth it. | ||
|
Raithed
China7078 Posts
1zealot for 2probes, well, you know, its common sense, the loss of the probe is at a disadvantage. 2zealot for 2 probes you are still ahead as you(same BO for instance) have two more zealots vs his. 2zealot for 3 probes, you are evened out. 3 zealot for 3 probes, still even. 3 zealot for 4 probes, well, start microing. after you lose 5+ probes to any # of zealots, imo, youre losing the eco-game slowly and slowly and with that, when he is able to reproduce troop count much faster than you, you must have to play catch-up in terms of probes AND army size. | ||
|
Elvin_vn
Vietnam2038 Posts
Make a game, mine for 5 mins, record the minerals. Say it's (X) mineral. Make another game where you spawn at the same location, mine for 5 mins but take out a probe at around 12 food, record the minerals at 5-mins mark, say it's (Y) X - Y = cost of a loss probe | ||
|
Ceril
Sweden1343 Posts
But on average and just one and having 15 one base? The time your put behind due to his succesfull harass killing one probe is not many seconds. If you get his harrassunit, then your ontop again granted you got the micromakro for the early game down tight. The other option here is to try strengthning your position by advancing your line. | ||
|
mjh
United States133 Posts
On November 03 2008 07:11 Elvin_vn wrote: uhm you don't need to calculate anything Make a game, mine for 5 mins, record the minerals. Say it's (X) mineral. Make another game where you spawn at the same location, mine for 5 mins but take out a probe at around 12 food, record the minerals at 5-mins mark, say it's (Y) X - Y = cost of a loss probe yes you do after the five minute mark your opponent will still have 1 more probe than you. if all you and your opponent do is build probes, he'll always have 1 more than you. the difference later on becomes negligible because of saturation, but it makes the cost more than just "X-Y" | ||
|
GHOSTCLAW
United States17042 Posts
On November 03 2008 07:58 mjh wrote: yes you do after the five minute mark your opponent will still have 1 more probe than you. if all you and your opponent do is build probes, he'll always have 1 more than you. the difference later on becomes negligible because of saturation, but it makes the cost more than just "X-Y" At least this way, you could see what the difference would be/how close it is. Real numbers always help more than made up numbers. The real problem is that you need to determine when (what time) you end up killing off 1 probe/2 probes/etc, and how much it matters I think that you would end up with a 2 dimensional graph- one axis for the total number of minerals after 5 minutes, one axis for number of minutes that went by before probe got killed. You would need a different graph setup depending on how many zeals got killed, as well as how many probes you lost. | ||
|
ShmotZ
United States581 Posts
but if thats not the case i dont think 1 probe will destroy you no matter how early, unless you 4 probe rush like nal_ra ;D | ||
|
Elvin_vn
Vietnam2038 Posts
On November 03 2008 07:58 mjh wrote: yes you do after the five minute mark your opponent will still have 1 more probe than you. if all you and your opponent do is build probes, he'll always have 1 more than you. the difference later on becomes negligible because of saturation, but it makes the cost more than just "X-Y" after 5 minutes mark, the difference of 1 probe doesn't matter | ||
|
VorcePA
United States1102 Posts
Larva. Each one used to produce a drone/overlord is not being used to produce a fighting unit, delaying the time it takes to build an army. I realize we've mostly been talking about PvP, but that adds a whole new element to the equation. Of course, all that means is that drones are an even more significant target compared to the other two races. How much more? That's where all this math comes in to play. | ||
|
Leath
Canada1724 Posts
Surely, it costs 100, but maybe it would kill 4 probes, or 2 dragoons in 2 lucky hits... maybe it will force the opponent to get a bunker or a sunken colony... im drunk nvm me | ||
|
Empyrean
16987 Posts
On November 03 2008 07:17 Ceril wrote: I seriously see where I should be posting all my calculus questions O_O TL.NET ftw. No Raithed please. | ||
Chill
Calgary25980 Posts
90n + 25n*(n+1) as taken from http://www.zileas.com/strategies/wc3/realcost.phtml | ||
|
Chill
Calgary25980 Posts
| ||
|
Archaic
United States4024 Posts
| ||
|
HooHa!
United States688 Posts
On November 03 2008 05:11 freelander wrote: honestly, why the fuck are you posting here. honestly, there are the ten commandments of tl.net what you should read once before you start posting. honestly, there are at least 10 people in this topic who posted actual thoughts, opposed to you. stop saying honestly. I wouldn't know the exact equation, but in mirror matchups, these worker kills mean everything. I generally just focus on the workers more than anything. I guess thats a big "duh" though, but recognizing the economic impact a little more closely makes it much more worthwhile. Like, I just killed 10 probes. Thats 500 minerals. Thats an expo and a pylon. Whatever they were doing is going to be slowed down. Maybe not tech wise, but their actions will deteriorate due to your superior economy. | ||
|
lololol
5198 Posts
On November 03 2008 14:35 Chill wrote: Zileas solved this equation a LONG time ago. 90n + 25n*(n+1) as taken from http://www.zileas.com/strategies/wc3/realcost.phtml This is for wc3, it would have quite different numbers in bw, because income progression isn't linear and costs are different, e.t.c. | ||
|
Shauni
4077 Posts
I know it might be obvious for some, I'm just saying... or mathematically formulated: calculate the time to replace the lost units from the economy advantage gained | ||
|
Chill
Calgary25980 Posts
On November 03 2008 23:24 lololol wrote: This is for wc3, it would have quite different numbers in bw, because income progression isn't linear and costs are different, e.t.c. OMFG I'm so dumb. He did it for BW too first. I never knew he did it for WC3 so when I saw "The true cost of Peons" I just assumed it was StarCraft without reading it. Goddamn. | ||
|
Elvin_vn
Vietnam2038 Posts
| ||
|
Guilty
Canada812 Posts
[Edit] Yeah i just saw that lil while go.. | ||
|
KOFgokuon
United States14893 Posts
that's the only way to do it | ||
|
HooHa!
United States688 Posts
Plus ramps and high grounds and natural defenses give you a little more advantages and hindrances that they would have to deal with if they wanted to end the game. Same would apply for them if they utilized it against you. Examples like neutral buildings, wallins, sim city's, mining an area that arbiter would recall. Manner pylon to increase efficiency of mining, or decrease theirs.. Rush distances. We all know this in the back of our minds, but it can heal a broken economy. little by little. I always feel pretty lame if I get caught in a worker swarm with a zealot. cause it isn't too hard to dodge the probes or whatever, and like the potential is lost from the zealot. | ||
| ||
