|
Not sure why people don't stay on the subject.
+ Show Spoiler +that is not an 8% economic deficit. for a start not all of your economy is drones, its also hatches overlords and queens. not all your money over the test period was put into econ and there us no reason to justify applying that assumption for the difference. the discrepany is 6 seconds / 6:29 = ~1.5%. id also like to see your complete results for each individual test as judging by the 3 second difference between each tests average and best, your error bars will render this result close to useless .
I think he has a point, wouldn't it be more meaningfull to see how long it takes to a specifique build like 14gas/14pool or 15hatch (with everything that is done) depending on pos?
However nice and interesting post. Thx
|
On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete.
Umm... Your econ maybe doesn't consist solely on drones, but your income does.
Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10.
|
On November 26 2011 05:00 Klondikebar wrote:
I feel like spawn locations is more comparable to white vs black in chess. Doesn't white have an extra 4% win rate or something? Based entirely upon "spawn color."
White has the higher winrate because white moves first.
Spawn color is how we identify who has the advantage, but it is not the advantage itself
|
On November 26 2011 05:23 Fleebu wrote:Show nested quote +On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete. Umm... Your econ maybe doesn't consist solely on drones, but your income does. Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10.
funnily enough your income actually also depends on hatches. but for the broader picture I think you may have misunderstood me good sir;
a player who has made 11 drones and 50 roaches has NOT had a 10 % economic advantage over a player who has made 10 drones and 50 roaches.
|
On November 26 2011 05:29 TERRANLOL wrote:Show nested quote +On November 26 2011 05:00 Klondikebar wrote:
I feel like spawn locations is more comparable to white vs black in chess. Doesn't white have an extra 4% win rate or something? Based entirely upon "spawn color."
White has the higher winrate because white moves first. Spawn color is how we identify who has the advantage, but it is not the advantage itself
True. Thanks for the clarification. I'll just keep reading the thread.
|
Are we talking a purely mining test here? cause surely the zerg with the larvae distant from the minerals has a reinforcement advantage?
|
On November 26 2011 05:36 shammythefox wrote:Show nested quote +On November 26 2011 05:23 Fleebu wrote:On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete. Umm... Your econ maybe doesn't consist solely on drones, but your income does. Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10. funnily enough your income actually also depends on hatches. but for the broader picture I think you may have misunderstood me good sir; a player who has made 11 drones and 50 roaches has NOT had a 10 % economic advantage over a player who has made 10 drones and 50 roaches.
In your example the player with 11 drones will have a 10% economic, income, advantage assuming everything else is equal. I assume I could include the 1/8 of an overlord cost but there is still a significant difference between close and distant larvae-mineral positions. Also hatcheries and queens are one time costs that are going to be built at specific points in economic development regardless. It will stall your growth for a short period of time, but a static 450 minerals at x income affects both timings equally so it can be disregarded. What's your point?
|
On November 26 2011 05:44 Duban wrote:Show nested quote +On November 26 2011 05:36 shammythefox wrote:On November 26 2011 05:23 Fleebu wrote:On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete. Umm... Your econ maybe doesn't consist solely on drones, but your income does. Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10. funnily enough your income actually also depends on hatches. but for the broader picture I think you may have misunderstood me good sir; a player who has made 11 drones and 50 roaches has NOT had a 10 % economic advantage over a player who has made 10 drones and 50 roaches. True, but they will have a 10% economic advantage soon assuming everything else is equal. I assume I could include the 1/8 of an overlord cost but it's still a significant amount. Also hatcheries and queens are one time costs that are going to be built at specific points in economic development either way. It will stall your growth for a short period of time, but it doesn't have a baring on my work. What's your point?
they will have a no economic advantage for the most of the game, a short 10% economic advantage for a few seconds whilst.their opponent creates their 11th drone and then none again. the result is an incredibly small economic advantage. my point amongst others which were in my original post is that your logic takes this to be a 10% economic advantage which is erroneous.
|
Nice analysis. 
Theres a shortcut though. 
Once you have measured the 2.5 second delay, you can calculate the delay as function of number of built drones:
The first 7 drones (including the first you built) will be 2.5 seconds late. Thus, i fyou stay on a 7 drone eco (ohai 7 pool!! ^^) everything you build will be 2.5 seconds late. If you build more drones, your 8:th drone will also be 2.5 seconds late to build, and with the extra walk distance, it will be 5 seconds late to the minerals, thus your 8 drone eco will be 2.5*7/8 + 5.0*1/8 = 2.8 seconds late. By generalising this argument, you can define a function d(N) which is the dealy in yuor total economy after N built drones. We know that
d(7) = 2.5
and we know that drone number N+1 will be built with a delay d(N), and arrive at the minerals with a delay d(N)+2.5. Thus, your N+1 drone eco will be delayed by
d(N+1) = N/(N+1)d(N) + 1/(N+1)(d(N)+2.5) = d(N) + 2.5/(N+1)
from this, by recursion it is easy to see that
d(N) = 2.5 + 2.5*sum_{n = 8 to N} 1/n
So the result is:
number of built drones --> delay of economy in gs
6 --> 2.5
10 --> 3.3
15 --> 4.3
20 --> 5.0
25 --> 5.6
30 --> 6.0
40 --> 6.7
50 --> 7.3
60 --> 7.7
80 --> 8.4
which is consistent with your result within your error margin, as I guess your build had around 30-40 drones.
Feel free to add it to the OP if you want, and also feel free to rewrite it better formulated consistent with your format.
|
On November 26 2011 05:40 Nymbul wrote:
Are we talking a purely mining test here? cause surely the zerg with the larvae distant from the minerals has a reinforcement advantage?
advantage over what? Terran and protoss do not make units from nexuses or CCs, you can't say it's advantage for zerg.
|
On November 26 2011 05:49 shammythefox wrote:Show nested quote +On November 26 2011 05:44 Duban wrote:On November 26 2011 05:36 shammythefox wrote:On November 26 2011 05:23 Fleebu wrote:On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete. Umm... Your econ maybe doesn't consist solely on drones, but your income does. Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10. funnily enough your income actually also depends on hatches. but for the broader picture I think you may have misunderstood me good sir; a player who has made 11 drones and 50 roaches has NOT had a 10 % economic advantage over a player who has made 10 drones and 50 roaches. True, but they will have a 10% economic advantage soon assuming everything else is equal. I assume I could include the 1/8 of an overlord cost but it's still a significant amount. Also hatcheries and queens are one time costs that are going to be built at specific points in economic development either way. It will stall your growth for a short period of time, but it doesn't have a baring on my work. What's your point? they will have a no economic advantage for the most of the game, a short 10% economic advantage for a few seconds whilst.their opponent creates their 11th drone and then none again. the result is an incredibly small economic advantage. my point amongst others which were in my original post is that your logic takes this to be a 10% economic advantage which is erroneous.
Economy is synonymous with income in this game. You’re thinking of the word macro includes production, resources, and position. In the end you will be in a much better position as the game progresses with 8% more income and economic growth that increases with economic position.
|
yep in numbers in chess white starts with 1/3 of a peon more. But here is the fun part, even in highest level it doesn't really mean something, mostly the ones thinking i am behind lose more often with black.
Anyway on topic. Zerg starts with enough larva to even overlap production, means they are ahead eco wise from the start over terran or toss. So i wouldn't complain or zerg might start with no larva at the start ^.^ to make it more fair.
That you might be behind from the start location is pretty common, for example that you have to expand towards your opponents main etc. (which is 200 times worse then having larva in the evil position).
Or that your addons points towards the map middle in tvt. There is a reasons some sports flip coins at the start. And sc2 does it for you.
|
When I played my first Zerg game against the AI I thought, that it is way harder when your spawningso, that your larva is in your Droneline, because its harder to click the Larva (with your mouse), because I didn't know about the select Larva Button lol.
nice analysis though
|
On November 26 2011 05:50 Cascade wrote:Nice analysis. Theres a shortcut though. Once you have measured the 2.5 second delay, you can calculate the delay as function of number of built drones: The first 7 drones (including the first you built) will be 2.5 seconds late. Thus, i fyou stay on a 7 drone eco (ohai 7 pool!! ^^) everything you build will be 2.5 seconds late. If you build more drones, your 8:th drone will also be 2.5 seconds late to build, and with the extra walk distance, it will be 5 seconds late to the minerals, thus your 8 drone eco will be 2.5*7/8 + 5.0*1/8 = 2.8 seconds late. By generalising this argument, you can define a function d(N) which is the dealy in yuor total economy after N built drones. We know that d(7) = 2.5 and we know that drone number N+1 will be built with a delay d(N), and arrive at the minerals with a delay d(N)+2.5. Thus, your N+1 drone eco will be delayed by d(N+1) = N/(N+1)d(N) + 1/(N+1)(d(N)+2.5) = d(N) + 2.5/(N+1) from this, by recursion it is easy to see that d(N) = 2.5 + 2.5*sum_{n = 8 to N} 1/n So the result is: number of built drones --> delay of economy in gs 6 --> 2.5 10 --> 3.3 15 --> 4.3 20 --> 5.0 25 --> 5.6 30 --> 6.0 40 --> 6.7 50 --> 7.3 60 --> 7.7 80 --> 8.4 which is consistent with your result within your error margin, as I guess your build had around 30-40 drones. Feel free to add it to the OP if you want, and also feel free to rewrite it better formulated consistent with your format.
I like it, I finished with 27 drones plus pool extractor and hatchery. Still I feel people would be critical if the analysis didn't include practical examples and two independent tests. Practical evidence is good, right?
|
On November 26 2011 05:54 Duban wrote:Show nested quote +On November 26 2011 05:50 Cascade wrote:Nice analysis. Theres a shortcut though. Once you have measured the 2.5 second delay, you can calculate the delay as function of number of built drones: The first 7 drones (including the first you built) will be 2.5 seconds late. Thus, i fyou stay on a 7 drone eco (ohai 7 pool!! ^^) everything you build will be 2.5 seconds late. If you build more drones, your 8:th drone will also be 2.5 seconds late to build, and with the extra walk distance, it will be 5 seconds late to the minerals, thus your 8 drone eco will be 2.5*7/8 + 5.0*1/8 = 2.8 seconds late. By generalising this argument, you can define a function d(N) which is the dealy in yuor total economy after N built drones. We know that d(7) = 2.5 and we know that drone number N+1 will be built with a delay d(N), and arrive at the minerals with a delay d(N)+2.5. Thus, your N+1 drone eco will be delayed by d(N+1) = N/(N+1)d(N) + 1/(N+1)(d(N)+2.5) = d(N) + 2.5/(N+1) from this, by recursion it is easy to see that d(N) = 2.5 + 2.5*sum_{n = 8 to N} 1/n So the result is: number of built drones --> delay of economy in gs 6 --> 2.5 10 --> 3.3 15 --> 4.3 20 --> 5.0 25 --> 5.6 30 --> 6.0 40 --> 6.7 50 --> 7.3 60 --> 7.7 80 --> 8.4 which is consistent with your result within your error margin, as I guess your build had around 30-40 drones. Feel free to add it to the OP if you want, and also feel free to rewrite it better formulated consistent with your format. I like, it was 27 drones. Still I feel people would be too critical if I didn't include practical examples and a second independant test. Practical evidence is good, right?
Yes you are correct, it does not make your tests obsolete, rather the opposite imo.
Strongest argument is a model consistent with measurements, which is what we have here. The model explains why this is what we expect, the measurement shows that the argument actually works. They are strong together.
Also, the calculation is general for any situation, while you only tested a specific case. You did all the work though (i threw this together in 20 mins...), so I am by no means trying to steal your fame. Good job on the analysis!
|
So what you're saying is that a Zerg spawning in top left of Shakuras should lose 100% of the time. Between this and the left/right combat disparity*, the disadvantage is insurmountable :D
+ Show Spoiler +Clearly trollin. For those curious, left/right disparity is just that when things attack-move each other, more often than not right will attack first. It's an incredibly miniscule difference (roaches will still kill each other) but feel free to try it out in a unit tester with single unit vs single unit battles :D For the record, banelings provide the most consistent result.
|
On November 26 2011 05:52 Duban wrote:Show nested quote +On November 26 2011 05:49 shammythefox wrote:On November 26 2011 05:44 Duban wrote:On November 26 2011 05:36 shammythefox wrote:On November 26 2011 05:23 Fleebu wrote:On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete. Umm... Your econ maybe doesn't consist solely on drones, but your income does. Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10. funnily enough your income actually also depends on hatches. but for the broader picture I think you may have misunderstood me good sir; a player who has made 11 drones and 50 roaches has NOT had a 10 % economic advantage over a player who has made 10 drones and 50 roaches. True, but they will have a 10% economic advantage soon assuming everything else is equal. I assume I could include the 1/8 of an overlord cost but it's still a significant amount. Also hatcheries and queens are one time costs that are going to be built at specific points in economic development either way. It will stall your growth for a short period of time, but it doesn't have a baring on my work. What's your point? they will have a no economic advantage for the most of the game, a short 10% economic advantage for a few seconds whilst.their opponent creates their 11th drone and then none again. the result is an incredibly small economic advantage. my point amongst others which were in my original post is that your logic takes this to be a 10% economic advantage which is erroneous. Economy is synonymous with income in this game. You’re thinking of the word macro includes production, resources, and position. In the end you will be in a much better position as the game progresses with 8% more income and economic growth that increases with economic position.
that comment isnt really relevant to the quote. im writing on a phone here and short of words I cant really think how to explain this much better. An 8% income would be huge if it existed however your ignoring all income generated which isnt poured back into drones and are also ignoring the fact that each drones effect on your income is only relevant with respect to the time it was created and the duration of the game.
|
Ah well it's one of those many quirky things that have very little effect on the game but could be noticeable.
I still think some of the other spawn assymetries for other races are more annoying:
- the reactor/techlab always being on the right of a rax can have a big impact for walling on some maps. I recall a game of Jinro on bel shir beach for example where he basically lost because his reactor spawned on the outside of his wall and got killed..
- many naturals have just tiny differences which make it much harder to FFE on some maps. It's much harder to memorise some exact FFE wallin for example because for some maps you need up to 4 different ones.
|
On November 26 2011 05:50 Alpina wrote:Show nested quote +On November 26 2011 05:40 Nymbul wrote:
Are we talking a purely mining test here? cause surely the zerg with the larvae distant from the minerals has a reinforcement advantage? advantage over what? Terran and protoss do not make units from nexuses or CCs, you can't say it's advantage for zerg.
advantage over its larva near mineral counterpart? not sure why you bring terran or protoss into this
|
On November 26 2011 06:01 shammythefox wrote:Show nested quote +On November 26 2011 05:52 Duban wrote:On November 26 2011 05:49 shammythefox wrote:On November 26 2011 05:44 Duban wrote:On November 26 2011 05:36 shammythefox wrote:On November 26 2011 05:23 Fleebu wrote:On November 26 2011 05:19 shammythefox wrote:
this is not an 8 percent deficit. your econ doesn't comprise solely of drones, it's also hatches queens and overlords. there it's no justification to ignore army spending when assesing this discrepancy and also of course the benefit of adding more drones is below linear (20 drones vs 40 does most usually NOT equal double the income). your discrepancy should be 6 seconds / 6:29 ~ 1.5%. id also like to see your full table of results for each test as judging by the3 second discrepancy between best and average times, your error bars could easily render this result obsolete. Umm... Your econ maybe doesn't consist solely on drones, but your income does. Whether you have 5 queens, 20 overlords or none 11 drones will always mine more than 10. funnily enough your income actually also depends on hatches. but for the broader picture I think you may have misunderstood me good sir; a player who has made 11 drones and 50 roaches has NOT had a 10 % economic advantage over a player who has made 10 drones and 50 roaches. True, but they will have a 10% economic advantage soon assuming everything else is equal. I assume I could include the 1/8 of an overlord cost but it's still a significant amount. Also hatcheries and queens are one time costs that are going to be built at specific points in economic development either way. It will stall your growth for a short period of time, but it doesn't have a baring on my work. What's your point? they will have a no economic advantage for the most of the game, a short 10% economic advantage for a few seconds whilst.their opponent creates their 11th drone and then none again. the result is an incredibly small economic advantage. my point amongst others which were in my original post is that your logic takes this to be a 10% economic advantage which is erroneous. Economy is synonymous with income in this game. You’re thinking of the word macro includes production, resources, and position. In the end you will be in a much better position as the game progresses with 8% more income and economic growth that increases with economic position. that comment isnt really relevant to the quote. im writing on a phone here and short of words I cant really think how to explain this much better. An 8% income would be huge if it existed however your ignoring all income generated which isnt poured back into drones and are also ignoring the fact that each drones effect on your income is only relevant with respect to the time it was created and the duration of the game.
I didn't ignore the income that wasn't poured back into drones. I made specifically sure to point out that if the amount of resources you spend on upgrades, tech, and army stayed the same with time your economic growth was actually LARGER than 8%. The amount of resources available to be poured into economy is based on what's left over after army/tech/upgrade development and an 8% bonus to economy has a disproportionally large effect on resources devoted to drone production if all other costs stayed constant.
If both the 10 and 11 drone players built armies at the same rate the 11th drone would grow on itself. The 11 drone player’s advantage in the long game would be disproportionally larger than the 10 drone player. If 5 drones work to produce army for both players, at the same rate of army production, it would effectively be 6 drones vs 5 drones. It would effectively mean a 20% bonus to economic growth.
P.S. As your opponent begins making his 11th drone you should have already started on your 12th in this case.
|
|
|
|
|
|
EPT
