While I introduced the Double Harvesting model, I am not an expert in game analysis. In fact, I am rather average player and I am sure I don't notice many details of high-level play. For that reason, if you have different thoughts about the games I talk about, or if I am blatantly wrong - say that! Still, I watched all the replays, trying to find situations where the model might have an impact on how the game unfolded.
Spoiler alert! If you haven't seen the tournament, don't know the result and want to watch the replays first - then please do so before reading this post.
Replays
On May 07 2015 04:04 Heyoka wrote: Here they are in all their glory.
Raw Data
Number of matches: PvP: 1 TvT: 17 ZvZ: 35 PvT: 27 PvZ: 27 TvZ: guess what... 27
Win-rate (note, that 27 matches is not much and may induce major error ) TvP: 63% (17 out of 27). TvZ: 63% (17 out of 27) PvZ: 59% (16 out of 27)
General Observations
Many games didn't differ from standard HotS. Players opted not to try to gain additional benefits that are possible because of the mod. There is absolutely nothing wrong in that! DH game alters the game most when: - early aggression - expanding without saturation (at 16 workers) both tactics involve a risk, that some --- give a situation --- may not want to take. However, if you and your opponent go for 16 workers in each base, the game will look very much as standard (it wouldn't differ in any other eco mod either). It is no fault of the mod, but also no fault of the players. Simply, there are situations where you want to just play Standard style!
There were however games where DH had an influence.
Worker cut for early aggression
In several games people cut their worker production early, in order to muster up a stronger army for a push. This was most common in 2v2 base play, where one player continued worker production, while the other made a stronger army. Naturally, this was most apparent in ZvZ. With the DH mod, getting the correct worker count of your enemy is more important than ever!
R2_Barrin_vs_ErikWM_Game_3 The red zerg stops drone production at 8-6 (8 mineram-mining drones in main, 6 in nat), over time replenish just to 8-8. At the same time, green zerg goes to 13-8. But having 50% drones no longer gives 50% more income, making the green zerg a bit harder to defend against such attacks. First red attack takes down a bit of green drones, reducing their count to 11-6. From that point the income is even, and a very intense zergling-baneling battle continues for few minutes until the winner is found.
R3_GoodSirTets_vs_Rasias_Game_1 This match probably highlights the best the mod, despite no 4th is ever taken! The purple zerg stops at 9-9 for zergling, and then zergling-baneling pressure. Green zerg struggles with defense, aiming to return to standard 16-16. But when that happens, purple is on 3 bases, with only 10-10-10 drones mining. This allows him to maintain the aggression, prevent green from expanding and then do a final blow with roaches.
R5_LaLuSh_vs_Liquid`MaNa_Game_2 Zerg is not the only race that can cut workers. In this PvZ match, MaNa cuts probes at 10-10 in order to snatch zerg's 3rd. Zerg appears to be unprepared, allowing him to go for the kill.
Cheese A nonlinear curve encourages various 1-base cheese and all-ins as well. Frankly, we didn't look at it much when designing Double Harvesting model. Lessons learned: it is an important factor to look at! The previous verson, DH10, promoted it even more! If I read it correctly somewhere, cheese was a problem in Starbow as well. But Starbow is modifying all the units and can nerf cheeses in other ways. DH is touching only the economy.
Some examples: R1_Big_J_vs_LEV_maid_Game_1 R2_GreenMash_vs_LEV_maid_Game_1 A proxy Void-Ray performed by the same player, cutting workers to 10. First game was straightforward, in second game the defender put much more effort and the whole match lasted over 10 minutes. Despite the long time, cheeser was committed, having very low worker count in the base.
R3_IzzoN_vs_ZZangDreamjOy_Game_2 A very straightforward proxy 3-rax. The attacker cut worker production at 10 and didn't even raise an orbital command.
R4_ZZangDreamjOy_vs_Liquid`MaNa_Game_2 While cheese hits a bit harder, it can still be defended. Here is an example of a very similar 3-rax on 10 workers, performed by the very same player. It was repelled without a problem, although player's skill was a factor here too
Stabilizing early aggression
What is unique about Double Harvesting mod, is that it is easier to stabilize after doing an early agreesion or even a cheese (talking about the attacker, not defender). Because of the decreasing efficiency, the commitment is smaller and ways of doing indirect damage is greater. The most important factor is: by how much the natural is delayed?
Scarlett vs PiliPili, match 3 The best example of it happening was in the previous DH tournament.
Scarlett went 10-poolgas on a 4-player map. She was able to get into enemy base but PiliPili took absolutely no damage. What happened however is that she delayed enemy expansion, while she expanded herself. As a result, at 6:30 she had a small supply advantage, noticeable income advantage (+15%) and PiliPili's natural was only just finishing.
R3_Beastyqt_vs_Rodzyn_Game_2 A less successful, but noteworthy example is this game where Protoss goes for a single proxy gate. He cut his workers at 11, does some serious damage while... expanding behind. By the time the attack is repelled, they were even in workers and bases. Defender's natural was delayed, but probably not as much as Protoss hoped for.
+1 base advantage Increasing the advantage of having an extra base was the primary goal of the mod. It should be noted however, that I am more interested in expanding when having less than 16 workers in bases. Expanding with 16+ workers can be considered a Standard play and will not differ as much.
The above requirement is easily satisfied with 4th mining base, but there were examples of taking an early 3rd for the additional economical boost. One of the examples is the R3_GoodSirTets_vs_Rasias_Game_1 which I discussed in the context of cutting workers for aggression. That's why I said that it is probably one of the best highlights of the mod.
But there were other examples: R2_IzzoN_vs_Fecalfeast_Game_1 Zerg goes for a greedy 3rd, but due to Terran's early agression the worker count remains quite low. Despite taking damage, he manages to keep up with the economy of 2-base terran. As the game unfolds, Zerg manages to set up 4th or even 5th, despite being under constant pressure. Some of the bases are lost, but he manages to keep going, expanding and containing terran on 2 bases only. It should be noted, that most of the time the Zerg didn't have any significant worker advantage. Usually, against Terran this is bad. But the number of bases gave enough economical benefit.
R3_Srilo_vs_ErikWM_Game_3 I am not sure, if those two players didn't set up their matches... they basically mirrored each other's builds 3 games in a row, but each time there was a small twist in how they do it, exemplifying the Double Harvesting mod. In this case: they both go "boring" pure-roach army, but red zerg takes 3rd early. They stay on the same amount of drones, thus any economical benefit is coming from the base count only. Slowly, but stedily red zerg gains advantage. And when the opponent expands - he attacks and wins.
R7_ROOTiaguz_vs_Beastyqt_Game_1 The finalist's first game. At 15:00 we have a red terran recovering from a rather painful drop. He is marginally behind in supply, has a bit less workers, but has 4 mining bases (main slowly depleting, but not there yet). He decides to move his army to contain the enemy on 3 bases. He maintains the contain for few minutes, keeping ~400 MpM income advantage. Afterwards he has enough strength to do a frontal push and win the game.
Overexpanding
While expanding is tempting, the benefits are not enormous and - more importantly - they do not appear immediately. It takes about two minutes (if not more) for a base to "pay" for itself, compared to a non-expanding strategy. For that reason it is easy to fall into a trap of overexpanding, and then having little to no army to defend oneself. This has been true in HotS and remains true in Double Harvest.
R3_LaLuSh_vs_larrybyerly_Game_1 The red zerg is going nuts, taking 5-th at 10 minute mark. A 2-base terran is responding aggressively, attacking at multiple fronts. Zerg has superior income, but not enough forces. He is doing his best and multiple places are defended, but a hive in the main dies which is a major hit. From that point he doesn't recover, although the terran still has hard time finishing him off. Overall a very interesting game, far from being one-sided. But some tweaking to the build would probably help. Maybe staying on 4 bases a bit longer would actually make this tactic viable?
R4_Arthur-_vs_zerge_Game_2 Not as extreme as the previous case, but the result was worse. Zerglings didn't do well against a sentry-heavy army, and the zerg didn't have money and tech to build anything else.
Splitting Workers
Another consequence of Double Harvest, that some omitted - is the split of the workers between the bases. By having 8-8 instead of 16-0 you get a 20% income boost. Some players have forgotten about it, leading to advantage taken by others.
R3_Srilo_vs_ErikWM_Game_2 Another mirror build from the same two people. They both go straightforward mass ling (no banelings) in 16 workers. One of the players however splits the drones 8-8 (potentially exposing them more to the enemy), while the other keeps 16 workers secured in their main. The 8-8 player gets an economical lead, but loses the match, probably due to micro.
Other Interesting Games
There were some other interesting games worth watching (and maybe casting? anyone?), although not necessairly that important for the perspective of Double Harvesting. My top pick is:
I think it is too early to make definite statements which race benefits most and which least from the mod. Before the tournament, the general concern was that the Zerg would be most favored, but with these matches I would carefully state - and you will probably agree - that this is not the case. My current attention is drawn to Terran's Mules. Those are not directly affected by the mod. Compared to a lone worker, Mule's efficiency is lower, but in a saturated base Mules actually help more (relatively). Consequently, Terran can turtle more than any other race. Moreover, morphing to orbital (and halting SCV production) is less punishing in DH than Standard. Out of all race macro mecanic, I fear that Protoss may get the short end of the stick: it is not helpful in low base count and does not help getting more bases.
In the end however I believe we need more games. The statistical error where only 27 games are played in each race configuration is still rather big. There is still 6% chance to obtain 17 heads in 27 coin tosses when using a fair coin (http://www.wolframalpha.com/input/?i=17 heads in 27 coin tosses). Usually, in statistics, a hypothesis is deemed false when the probability drops to at least 0.1%. We will need 5 times more matches to reach that point, assuming the same ratio remains. And this of course assumes an even distribution of player skill between the races.
This is really good stuff! Thanks for your hard work
DH was a mod which set out to allow players to exploit players who cede map control (e.g. a 3 base turtle strategy) by taking additional bases while preserving as much of HotS metagame as possible. How successful do you think DH was in doing this?
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^___________________________________^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the series against lalush was super fun. and the game one was very interesting to point out.
Thank you for these analyses! I was busy much of the day and only saw a few matches. At this point, I'm getting the sense that some tweaks should be done to Mules, but it's hard to say for sure. I'm inclined to say that they should compete with SCVs for patches, rather than mining right over them. This would encourage their use more on new bases or other undersaturated bases, as well as marginally increase the value of dropping supply depots. It's a bit of a severe change though, so a lot more testing is needed first. And I'm more interested in seeing Blizzard open up LotV to extension mods and trying DH9 on there. Or at least on the fan-made version of it.
TL is really pushing this double harvest model hard and I don't understand why. The issue isn't that expanding doesn't provide enough benefit, it's that you start with 12 freaking workers and end up being forced into opening at least 3cc and 3 hatch (and protoss 2 nexus because they can't defend 3 early bases). This double harvest thing is as novelty and as overhyped as Starbow, which didn't even get off the ground and was laid to rest within weeks of its inception.
On May 16 2015 14:19 ClaudeSc2 wrote: TL is really pushing this double harvest model hard and I don't understand why. The issue isn't that expanding doesn't provide enough benefit, it's that you start with 12 freaking workers and end up being forced into opening at least 3cc and 3 hatch (and protoss 2 nexus because they can't defend 3 early bases). This double harvest thing that worked in BW is as novelty and as overhyped as Starbow, which didn't even get off the ground and was laid to rest within weeks of its inception.
GG NO RE
You've confused two separate issues. Once is the 3 base turtle (which is what DH set out to solve) and the other is a hyper accelerated early game via 12 workers which DH makes no attempt to resolve. DH was not present in BW.
On May 16 2015 14:19 ClaudeSc2 wrote: This double harvest thing that worked in BW is as novelty and as overhyped as Starbow, which didn't even get off the ground and was laid to rest within weeks of its inception.
Starbow was laid to rest because it got rid of everything that SC2 fans enjoy that wasn't already in BW. There was no real plan to build a better SC2. They just tore everything down and built BW2. A lot of their ideas are great, but they are very much Starbow ideas and not Starcraft 2 ideas. The highground 50% miss chance could be worked into SC2, but not quite the way it was implemented into Starbow, where they completely removed the SC2 highground vision mechanics and implemented their own. It could've been combined and compromised and created something that would've greatly benefited Starcraft. They removed the controversial WM that was a critical component of the nail-biting 4M vs MutaLingBling TvZ. Then they also severely redesigned the largely beloved and uncontroversial Baneling, and after trying some fairly sensible rebalances to the unit, removed the somewhat bland, but purposeful and important Marauder. They made mech exactly how it was in BW, which is significantly better than in SC2, but at the cost of making bio much less fun (though admittedly, adding the Herc was a brilliant move on their part). They initially switched the Hydralisk and the Roach around, tech-wise, which could've easily been worked into SC2, but then they redesigned the Roach (which was already shown functional as a tier 2 unit in a different prominent mod, OneGoal) and later removed it. They made Protoss much more fun on the whole, but they didn't really take full advantage of the areas that SC2 did Protoss right, such as the Phoenix, the WoL Void Ray, and a unit with Blink as part of a core army composition. There's really no need for the Dragoon's return, other than the fact that it was necessary for their balance process, since they started it from BW and then tweaked the design from there. Starbow is a lot of fun, and I have an easier time talking some of my friends into playing it than talking them into playing SC2, but you're making some really off-base assumptions by thinking it ever claimed to be "SC2, but better".
Ultimately, the reason why it never really caught on, outside of a small but dedicated playerbase, is because it was never made for SC2 fans. It was made for BW fans, and BW fans often just preferred to stick with BW.
Unrelated, but anyone else notice that this thread just shows up as "Double Harvesting - Replay Anal..." in the community news sidebar?
I appreciate the hard work but I stopped reading after cheese. When making an rts game like starcraft no matter the system you are going to have cheese/all'ins. It's good that you want to eliminate it or reduce the amount of variation that could happen but the fact is players aren't scouting and not able to hold different timings / cheeses etc... Without a long test period I would it would be near impossible to say it has the say problem with Starbow. Starbow has had at least a couple of years to address this issue, and it grew into a great game (Still playing it.) I think having maybe 2-3 more tournaments of DH9 or w/e it was would be a good way to address such a time thing. The more games you have the more it becomes relevant to what the problem is. Personally I don't think you should address "cheese" as an issue until it becomes a notable thing. Plus Cannon rushing isn't something you can stop. It happens every where.
On May 16 2015 15:59 Rukis wrote: I appreciate the hard work but I stopped reading after cheese. When making an rts game like starcraft no matter the system you are going to have cheese/all'ins. It's good that you want to eliminate it or reduce the amount of variation that could happen but the fact is players aren't scouting and not able to hold different timings / cheeses etc... Without a long test period I would it would be near impossible to say it has the say problem with Starbow. Starbow has had at least a couple of years to address this issue, and it grew into a great game (Still playing it.) I think having maybe 2-3 more tournaments of DH9 or w/e it was would be a good way to address such a time thing. The more games you have the more it becomes relevant to what the problem is. Personally I don't think you should address "cheese" as an issue until it becomes a notable thing. Plus Cannon rushing isn't something you can stop. It happens every where.
I'm gonna go out on a limb and say you did not read the entire article.
And perhaps you should have, as BL explains in the very last paragraph that we need a lot more testing in order to get the raw data necessary to prove the system works.
I just don't understand the point of this double mining TL suggestion. Isn't this just equivalent to 4-mineral patch mining? (i.e. even more extreme than the 8-for-a-while, then 4-after-that which LOTV introduces)?
On May 16 2015 12:28 Plexa wrote: DH was a mod which set out to allow players to exploit players who cede map control (e.g. a 3 base turtle strategy) by taking additional bases while preserving as much of HotS metagame as possible. How successful do you think DH was in doing this?
Personally I think it does a good job, without overdoing it. 3 base turtling is still valid strategy. It just gives the expanding opponent a bit more chance. On the other hand, I think players were a bit more active and aggresive thanks to the mod, and we never hardly ever seen a completely passive play. I recall there was a single PvZ with completely passive 2-base Protoss. He was crushed by the Zerg without any problems. I think the skill was a major factor in that game, and not the mod.
On May 16 2015 14:37 Barrin wrote: Ooo the replays are out Maybe I'll get around to doing a BR or two on the better examples. Been waiting for an opportunity to do some more BR's.
What BR stands for? Battle Report of some kind?
On May 16 2015 15:59 Rukis wrote: I appreciate the hard work but I stopped reading after cheese. When making an rts game like starcraft no matter the system you are going to have cheese/all'ins. It's good that you want to eliminate it or reduce the amount of variation that could happen but the fact is players aren't scouting and not able to hold different timings / cheeses etc... Without a long test period I would it would be near impossible to say it has the say problem with Starbow. Starbow has had at least a couple of years to address this issue, and it grew into a great game (Still playing it.) I think having maybe 2-3 more tournaments of DH9 or w/e it was would be a good way to address such a time thing. The more games you have the more it becomes relevant to what the problem is. Personally I don't think you should address "cheese" as an issue until it becomes a notable thing. Plus Cannon rushing isn't something you can stop. It happens every where.
I agree with what you say!
I have put a point "Cheese" not because it is a problem, but because it is a strategy I have seen quite often in the replays. I wouldn't want to see cheese eradicated. But it is definitely something to look at, especially since DH9 buffs it a bit. So I included in the list to report it and draw attention to it. One of the replays actually shows that Cheese can be defended.
My current biggest concern is not cheese but Mules. I am thinking about reducing its efficiency by about 10%. But to make a definite statement we need more games. For that reason I do hope there will be more DH9 showcasts and tournaments!
On May 16 2015 15:59 Rukis wrote: I appreciate the hard work but I stopped reading after cheese. When making an rts game like starcraft no matter the system you are going to have cheese/all'ins. It's good that you want to eliminate it or reduce the amount of variation that could happen but the fact is players aren't scouting and not able to hold different timings / cheeses etc... Without a long test period I would it would be near impossible to say it has the say problem with Starbow. Starbow has had at least a couple of years to address this issue, and it grew into a great game (Still playing it.) I think having maybe 2-3 more tournaments of DH9 or w/e it was would be a good way to address such a time thing. The more games you have the more it becomes relevant to what the problem is. Personally I don't think you should address "cheese" as an issue until it becomes a notable thing. Plus Cannon rushing isn't something you can stop. It happens every where.
I'm gonna go out on a limb and say you did not read the entire article.
And perhaps you should have, as BL explains in the very last paragraph that we need a lot more testing in order to get the raw data necessary to prove the system works.
I'm gonna go out on a limb and say you're probably right. But strictly speaking about "cheese."
On May 16 2015 18:28 Haighstrom wrote: I just don't understand the point of this double mining TL suggestion. Isn't this just equivalent to 4-mineral patch mining? (i.e. even more extreme than the 8-for-a-while, then 4-after-that which LOTV introduces)?
The major difference is that a worker mines twice or even three times before returning to the main building. But between those mines there is a waiting period, during which the mineral patch is "unattended" and can be exploited by other workers. From the perspective of the mineral patch, there are several of those "unattended times" coming from each worker, allowing extra workers to still give extra income, but at gradually decreasing efficiency.
On May 16 2015 18:28 Haighstrom wrote: I just don't understand the point of this double mining TL suggestion. Isn't this just equivalent to 4-mineral patch mining? (i.e. even more extreme than the 8-for-a-while, then 4-after-that which LOTV introduces)?
The major difference is that a worker mines twice or even three times before returning to the main building. But between those mines there is a waiting period, during which the mineral patch is "unattended" and can be exploited by other workers. From the perspective of the mineral patch, there are several of those "unattended times" coming from each worker, allowing extra workers to still give extra income, but at gradually decreasing efficiency.
Yes, so if a worker mines twice on 8 patches, essentially 2 mineral patches are "busy" during that worker's trip. That means 2 workers would optimally mine 2 patches, or 8 workers optimally mining 8 patches. That's the same income as paired workers mining on 4 (except paired workers is more aesthetic, and so is the natural choice between the two options).
Sure, there are minor fluctuations due to travel time between patches, differences in patch distance from the nexus, etc, but on a macro scale they're equivalent.
Blizzard understood this in their response on why they would not be taking this proposal forward.
On May 16 2015 19:06 Haighstrom wrote: Yes, so if a worker mines twice on 8 patches, essentially 2 mineral patches are "busy" during that worker's trip. That means 2 workers would optimally mine 2 patches, or 8 workers optimally mining 8 patches. That's the same income as paired workers mining on 4 (except paired workers is more aesthetic, and so is the natural choice between the two options).
That's not true. In DH model a worker mines from a single mineral patch, making it "busy" twice in a row, with a small "not-busy" gap in between. It goes like this (values are rounded)
1.5s: Mine minerals 0.5s: Wait at mineral patch 1.5s: Mine minerals (from the same patch) 0.5s: Wait at mineral patch 1.5s: Mine minerals for the 3rd time 0.5s: Wait at mineral patch 4s: Round trip to the gathering building
This creates 2 periods of 0.5s where another worker can squeeze in, but it will reduce efficiency. This also gives 4.5s period where a worker can fit in, but also at reduced efficiency, because it takes 5.5s from beginning of first mining till the end of last mining. As a result 2 workers on single mineral patch (or 8 workers paired on 4 mineral patches) will mine slower than when each worker goes to a separate mineral patch.
I can understand that it could be confusing to you. But this is no excuse for blizzard for not understanding it. Please don't defend them like this.
Blizzard claims that DH model is extreme, and "In the Void model, we have something in between the above" [HotS and DH]. But the DH tournament showed that DH is rather small nudge to the economy, some players even expressing their dissapointment that it is not more extreme. There is no better proof than actual games, to show that the claim made by Blizzard is wrong.
Except that's not the case and travel time actually does matter. If you're flat doubling the returned minerals (i.e. DH10) then here is a graph comparing income rates. After factoring in travel time, 8 workers mining 8 patches in DH10 is roughly equivalent to 12 paired workers mining. We've since abandoned DH10 because it blows up the early game, but this is enough to show your understanding of the model isn't quite accurate.
So I've lost perspective on what you're trying to achieve here. You're only showing me the 8 mineral model for HOTS, not the 8>4 mineral model for LOTV. So you should be showing me the accumulated resources over time for various worker counts, because there is a big drop off in income at a point in time in LOTV, which is highly relevant to what their model is doing versus this one.
The objective for both this project, and Blizzard's LOTV model, is to make the income per additional worker decrease compared with HOTS, in order to encourage expansion, right? Blizzard want to do this by making 4 mineral patches disappear after a while, meaning >16 workers becomes less efficient per base at first, then >8 workers less efficient after some time.
The graph you're showing me is that after 8 workers mining becomes less efficient, but the reduction of income per additional worker is less than the LOTV model AFTER the patches expired (so in LOTV you get more income for workers 9-16 before the expiration, and much less income for workers 9-16 after the expiration - in fact none for workers 13-16).
So basically this model punishes players for not expanding sooner than LOTV, but not as much as LOTV once the 4 minerals have expired.
Wouldn't the same goal be achieved by changing a base to having, say, 6 or 7 mineral patches? Of course the graphs would look a bit different, but the objective was never "let's make the income this exact curve", was it?
I just feel like this whole project is spawned off the back of "we liked the economy in BW so we'll try and emulate that", and the actual objective and means of getting there have been lost within unnecessarily verbose posts full of graphs (why say with 100 words what you can say with 5, etc).
The best comparison between the three models would be to take say 12, 16, 20 and 24 workers (for simplicity) and then show cumulative income over time on 1 base for each situation (HotS, DH 8 normal minerals, LotV, maybe DH on LotV mineral patches?) and see how each one differs in cumulative income over time, to show the drop off in income as well as how efficiency works out as well, since you can't really compare HotS/DH with LotV when the LotV mechanism is primarily about reducing income after a period of time due to mineral drop-off, compared to DH which is about reducing income per worker when you hit a saturation point.
Not to say the DH vs HotS graph isn't useful for showing the impact DH has, but it doesn't really establish a comparison point to LotV due to the totally different approach to changing the income profile which is time base more than worker based.
On May 16 2015 20:24 Haighstrom wrote: [...] you should be showing me the accumulated resources over time for various worker counts. [...]
Sure, I ran a simulation for you a few weeks ago, based on the curves plexa posted. I also took the liberty to include the starting 12 workers, the cost and build time of workers and town halls, and how the left over minerals turn into army supply, up to the supply cap, a while after you get the minerals. And if you want to do more, the script is open-source for you to modify and rerun, so you can tell the simulation to stop at any number of bases and workers. + Show Spoiler +
On May 16 2015 18:43 Daeracon wrote: Really great analysis, I really hope we get to see more games! Lycanleague!? Was he not going to make a DH Lycanleague week? Hope that happens soon.
Any news on that one? Or maybe any other DH games/tournaments?
On May 16 2015 18:43 Daeracon wrote: Really great analysis, I really hope we get to see more games! Lycanleague!? Was he not going to make a DH Lycanleague week? Hope that happens soon.
Any news on that one? Or maybe any other DH games/tournaments?
Cheese A nonlinear curve encourages various 1-base cheese and all-ins as well. Frankly, we didn't look at it much when designing Double Harvesting model. Lessons learned: it is an important factor to look at! The previous verson, DH10, promoted it even more! If I read it correctly somewhere, cheese was a problem in Starbow as well. But Starbow is modifying all the units and can nerf cheeses in other ways. DH is touching only the economy.
On May 17 2015 20:45 sixfour wrote: Why is cheese a problem exactly?
It's not a problem per-se. It is a something you have to keep a close eye on, though. If cheese is too weak, it will be rare diminishing the need for scouting and safe play. If cheese is too strong, games will become coin-flippy and cheese-oriented.
DH9 gives a little bit more power to cheese compared to HotS. If cheese was at the right spot in HotS, then it may be too strong in DH9. A careful evaluation is needed to decide it.
I had a lot of fun playing in the tournament. Thanks to everyone who ran it! The mod felt pretty good while not being hugely difference from HOTS. Everybody should watch R3_GoodSirTets_vs_Rasias_Game_1 btw I'm going to try and do a community cast of the ro8 onwards and some of the games highlighted here within this week.
On May 18 2015 00:21 GoodSirTets wrote: I'm going to try and do a community cast of the ro8 onwards and some of the games highlighted here within this week.
That would be great! DH needs to be exposed more to people who may otherwise not notice it. Not everyone is on TL forums... Looking forward to watch it - where can we find your casts?
On May 18 2015 00:21 GoodSirTets wrote: I'm going to try and do a community cast of the ro8 onwards and some of the games highlighted here within this week.
That would be great! DH needs to be exposed more to people who may otherwise not notice it. Not everyone is on TL forums... Looking forward to watch it - where can we find your casts?
I'll link them in this thread and probably plug them on screddit once I upload them
On May 17 2015 20:45 sixfour wrote: Why is cheese a problem exactly?
It's not a problem per-se. It is a something you have to keep a close eye on, though. If cheese is too weak, it will be rare diminishing the need for scouting and safe play. If cheese is too strong, games will become coin-flippy and cheese-oriented.
DH9 gives a little bit more power to cheese compared to HotS. If cheese was at the right spot in HotS, then it may be too strong in DH9. A careful evaluation is needed to decide it.
Yeah, that seems fair, although there being too much of an opportunity for cheese isn't necessarily a bad thing. I'd much rather watch short games whereby people are doing something unorthodox and it's a case of the other guy recognising it and adapting to try to defend it, or at least there's the option of it so it keeps players honest, rather than the state of the game being such that there's negative play until 200/200 armies get thrown at each other
On May 18 2015 05:09 sixfour wrote: Yeah, that seems fair, although there being too much of an opportunity for cheese isn't necessarily a bad thing. I'd much rather watch short games whereby people are doing something unorthodox and it's a case of the other guy recognising it and adapting to try to defend it, or at least there's the option of it so it keeps players honest, rather than the state of the game being such that there's negative play until 200/200 armies get thrown at each other
Cheese is not the only alternative to negative play. I believe DH9 gives more options for early aggression that is not an all-in or cheese. You can cut workers for bigger army and then stabilize after doing moderate damage (sections "Worker cut for early aggression", and "Stabilizing early aggression")
On May 17 2015 20:45 sixfour wrote: Why is cheese a problem exactly?
It's not a problem per-se. It is a something you have to keep a close eye on, though. If cheese is too weak, it will be rare diminishing the need for scouting and safe play. If cheese is too strong, games will become coin-flippy and cheese-oriented.
DH9 gives a little bit more power to cheese compared to HotS. If cheese was at the right spot in HotS, then it may be too strong in DH9. A careful evaluation is needed to decide it.
Yeah, that seems fair, although there being too much of an opportunity for cheese isn't necessarily a bad thing. I'd much rather watch short games whereby people are doing something unorthodox and it's a case of the other guy recognising it and adapting to try to defend it, or at least there's the option of it so it keeps players honest, rather than the state of the game being such that there's negative play until 200/200 armies get thrown at each other
Totally agree with this, I actually used to love playing PvP in WoL since it was basically all about micro with small armies. Learning to defend 2 gate proxies, cannon rushes and continual 4 gates as I moved up the ladder was pretty much the best fun I have ever had playing SC2, I just loved the feeling of out-microing your opponent. Anything which makes maxing out slower is positive for the game imo, I guess with the qualifier that you're actually encouraged to do things with your units as you build towards max.
To be fair we have to remember that the current LotV model does a pretty decent job of this
So I've lost perspective on what you're trying to achieve here. You're only showing me the 8 mineral model for HOTS, not the 8>4 mineral model for LOTV. So you should be showing me the accumulated resources over time for various worker counts, because there is a big drop off in income at a point in time in LOTV, which is highly relevant to what their model is doing versus this one.
The objective for both this project, and Blizzard's LOTV model, is to make the income per additional worker decrease compared with HOTS, in order to encourage expansion, right? Blizzard want to do this by making 4 mineral patches disappear after a while, meaning >16 workers becomes less efficient per base at first, then >8 workers less efficient after some time.
The graph you're showing me is that after 8 workers mining becomes less efficient, but the reduction of income per additional worker is less than the LOTV model AFTER the patches expired (so in LOTV you get more income for workers 9-16 before the expiration, and much less income for workers 9-16 after the expiration - in fact none for workers 13-16).
So basically this model punishes players for not expanding sooner than LOTV, but not as much as LOTV once the 4 minerals have expired.
Wouldn't the same goal be achieved by changing a base to having, say, 6 or 7 mineral patches? Of course the graphs would look a bit different, but the objective was never "let's make the income this exact curve", was it?
I just feel like this whole project is spawned off the back of "we liked the economy in BW so we'll try and emulate that", and the actual objective and means of getting there have been lost within unnecessarily verbose posts full of graphs (why say with 100 words what you can say with 5, etc).
I'd be interested to see just how close the curves of DH (3x3) and 7 patch regular would actually be for a single base. I suspect full saturation would be about identical and that at least would mirror one of the benefits I see for DH in it's current form - a general slow down of the 3 base economy. This would throw up a couple new design choices: First, if you're going to use half patches a la LotV do you go with 3 or 4 half patches? Second, do you keep start bases at 8, thereby keeping the feel of the very early game or not?
I have no strong feeling as to whether this would be a good or bad change, I'm just interested in seeing outcomes. I do know that "Change available mineral patches to 7" would be an easier sell to the public and Blizzard as it's much much simpler to explain and simpler to try out especially as it can be tested in LotV at the same time as current options just by editing a couple maps.
edit: Just to note that having differing amounts of minerals in main / expansions was something BW did. This of course doesn't make it automatically good but it does provide precedent.
I'm really excited to see where this goes, I'm already starting to feel the strength of three bases in just about all matchups. It would be really nice to get away from 3 base 200/200 balls and get to a more broodwar mass expand style of game!
EPT
![[image loading]](/staff/Plexa/dh9.png)
![[image loading]](http://i.imgur.com/pjl6y2X.png)
![[image loading]](http://i.imgur.com/EWfXUqe.png)
![[image loading]](http://i.imgur.com/zPhxVd6.png)
