Battle 1062721 detail

sort

Anarchid

3 years ago

Bertha gets built approx 50% into this game.

Not recommending a watch, this is just painful.

+2 / -2

Bakuhatsu

3 years ago

This is why I always check to see if rewdrfe is online and in a game before queueing mm...

+1 / -0

PRO_Dregs

3 years ago

The rew is @ATOSTIC Major

+0 / -0

Bitterly

3 years ago

If anything shows how sad BB now is. The best move would've been to just ignore it after doing 3 wall-ups. It stood there for more or less 30 mins and didn't even directly make cost.

+1 / -0

Steel_Blue

3 years ago

I don't know how strong BB is when you let the AI pick targets. It makes it a micro sink, especially if the AI wants to decide to keep movint the turret forever. Rewdrfe2 had many player-issued orders on it that were effective, but then shot at empty space for extended periods of time.

Anyway, don't forget your lobsterpot countermeasures to BB. Teraform walls only protecting from the direction of the threat. It's easier to knock down a teraform spike than it is to knock down a 'deeper' tower that's like 4 squares thick. At some point with a thick enough wall it takes an insane amount of time to level the entire teraform blockage that it's not worth it. And if BB doesn't teraform ground under buildings reinforcing with a razor on top is the best bet. I'm saying your teraform was flimsy and inefficient.

+0 / -0

Anir

3 years ago

My question would be:
How could he hold the front and build a bertha at the same time?

+0 / -0

Steel_Blue

3 years ago

Inde pulled off a gs switch which did not pan out. He then discovered an 80% complete Bertha.

+0 / -0

Bitterly

3 years ago
(edited 3 years ago)

quote:
How could he hold the front and build a bertha at the same time?

Sounds like a question someone who hadn't played rew much would ask ;D

quote:
I'm saying your terraform was flimsy and inefficient.

Oh, I mean, for sure. Yet it seemed enough, I'd even say this game had excessive tera.

+1 / -0

PRO_Dregs

3 years ago
(edited 3 years ago)

The rew plays shieldbots on maps where he can identify a line which he does not advance beyond much, but moves his shieldbots generally from side-to-side as they amass, capturing would be raiding attempts. Doesn't need to raid, just needs to stop you from doing so whilst the ball becomes invincible. Then goes for artillery whilst keeping the ball generally available.

+0 / -0

Anarchid

3 years ago

Basically the good old Klon style, except less brutal.

+0 / -0

Silent_AI

3 years ago

I think better gunships to use would be nimbuses. @Inde_Irae built locusts, but their advantage is that they are fast, which comes at expense of power and range, they can be countered by lotuses and normal turrets.

When I was watching some big team game recently as a spectator, I was very surprised how effective can mass nimbus be. It outranges all static defenses, even stingers, and it has to be countered by dedicated aa, or some good long range units like lances (or raptors, but in this case, rew did not have air yet).

In this specific case, rew had cloakies and their only counter would be snipers, but they have very low dps, or dedicated aa units, but I think that aa-units from cloakbot factory have low range, which can be punished by nimbuses too. Besides, that gunships switch came as surprise, so before he would be able to build some aa units, his base would be dead.

What happened instead was that that mass of locusts rushed straight into static defenses in base and got killed by lotuses, com, and stardust.

+0 / -0

esainane

3 years ago

+0 / -0

Silent_AI

3 years ago

If you mean that razor annihilates locusts, I agree with you, if you want to say that one razor would be able to stop 3 nimbuses attacking base, I do not. First - there was no razor in the base to begin with, so if Inde_Irae micromanaged nimbuses correctly, it would not even get built, second, nimbuses, if explotiting their superior range, can be seen only as radar dots and still be attacking stuff in base, and razor frequently misses radar dots, and third, even if it somehow got built in time, it would still die to nimbus fire, 3 nimbuses together have much higher DPS that one razor and 4k vs 3*2.8k hp. The point is, no, I do not think that nimbuses would be counterable in that situation, they could have had annihilated the base.

And they would be superior raiding forces too, rather than bunch of locusts, they can take out whole expansions without punishment, if there is no dedicated aa. Locusts get shut down by a forest of lotus turrets, not mentioning even a stinger, nimbuses outrange stingers.

+1 / -0

Anarchid

3 years ago

Nimbus may not have been more efficient than Locust, which actually did reasonably well aside from attacking main.

But it would likely have fared much better than Revenant.

+0 / -0

esainane

3 years ago

Very well. Lets take a closer look to see how these counterfactuals work out in practice.

To replicate, /skip 682 and pause to quickly reach the point where the locusts are spotted at 11:22. /cheat, /godmode, /spawn 3 gunshipheavyskirm. Send the Locusts back, and send the Nimbus forward in their place. 2200 is close enough to 2280 to be a fair approximation.

Switch to rewd's perspective and unpause. alt-q then alt-d a few times to insert AA to the factory build queue. Select the base constructor, set it to high priority, build two Razors.

Lets first assume the three Nimbus beeline for the base, since the claim seems to be around three Nimbus being unstoppable in this situation. Taking a direct line along the bottom edge minimizes the time available to prepare. Once they reach the main base, one Razor and two Gremlins are out, and the second Razor is nearly done. By coincidence, the Reaver is in a very good position to intercept them as well.

This isn't a good position for three Nimbus to find themselves in. Gremlins on attack move can run in small circles to dodge Nimbus shots indefinitely, and they have the speed to keep up with Nimbus should they try to retreat. Nimbuses slowly turning around while directly over a Reaver isn't going to end well either.

What if they move keep moving forward, and fire at the factory? This keeps the Reaver away, and with few other outlets for buildpower, has a chance of making the best of a bad situation:

Well, down to 9% is pretty close, but ultimately that didn't end well. What if we restart, and go for the Advanced Geothermal instead?

That's a more satisfying result! Still, running into a cloaked Reaver at the blind spot directly underneath them was a bit painful. rewdrfe2 gets to take the reclaim, but lost a fair bit of buildpower in the process. It's hard to call this a tempo move, since the gunship side lost more army value, but maybe something can be made of it.

Perhaps trying to make an apples-to-apples comparison wasn't fair. Locusts are really good at quickly killing wind and caretakers in a run-by, taking out unprotected expansion, and forcing defense expenditure even in places they don't directly hit. Arguably this was the point, but Nimbus doesn't really try to do this, as somewhat hinted at by mentioning use of range. What happens if we just set the Nimbus on attack move, to slowly advance and take out the southern expansions?

Not bad! They quickly clear out the expansion behind the hill, including the Knight and Reaver in the area, but start taking fire once freshly produced Gremlins come out about half a minute later. They can take out the two Mexes and Lotus under fire, but pushing into increasing numbers of unhittable Gremlins is a losing game, and the Gremlins can do a very good job of tracking and intercepting whereever the Nimbus would try to move next. Not the most impactful payoff for ~2k of units, but you do get to keep the units and slowly wage a war of attrition, Pot-style, if that's your cup of tea.

Perhaps the specifics of this battle were unfortunate, particularly the timing of the Reaver's movements, and Nimbus is still a good idea in 1v1.

Prediction: while Locust will not be as effective in large Pot games, due to the higher relative density of defenses and units, they will be sufficiently effective in 1v1 that the overwhelming majority of first gunship strikes in skilled games will start with Locust.

At 2.5k+ elo 1v1, there are 1027 winning Gunship Plant as second factory moves in my dataset. Lets see what the consensus on the most popular Gunship switch opening is:

heap.get('gameid').dim.top(Infinity).map(d=>d.gameid).join('\n')

[Spoiler]

while read id; do sed -n '/^.*finished unit \(Locust\|Nimbus\)/{s/^.* \([^ ]*\)$/\1/
p
q}
$iNeither' /var/lib/zkreplay/stats/$id/events.log; done < gs-ids.txt | sort | uniq -c
    700 Locust
    170 Neither
    157 Nimbus

I'll freely admit that was more games starting with Nimbus than I was expecting, but the numbers still appear to be overwhelmingly in favour of Locust.

+6 / -0

Anarchid

3 years ago

Timeline editing used for good.

+0 / -0

Silent_AI

3 years ago

Wow, thank you for your analysis, but what I disagree with is that you already changed what was in a queue in that fac and already started making razor, so as you said, there already was one razor and gremlin and other gremlins on a way when gunships arrived. Of course that the outcome was different. If locusts arrived into that situation, it would be much different too. Rew did not change his queue nor started making dediated aa because the attack came as a surprise and he did not have time to react. You may say that nimbuses are slower and that gives him more time, but he did not start making gremlins during the whole attack, I doubt it would change.
Just look at the stats. Nimbus has 2800 hitpoints, 178 dps and range 600 (about that range - sorry, it does not outrange stinger, I was wrong), lucust has 800 hp, 57 dps and range 260. 10 locusts = 3 nimbuses. That is 8000 vs 8400 hp in favour of nimbuses and 570 dps vs 534 dps in favour of locusts. Bud main advantage of nimbus is its range, it can completely avoid lotus turrets and destroy targets from afar, that reaver is no threat to it. Unlike locust, its machine gun deals damage to a larger area and not a single target, which makes it slower at destroying small targets like lotuses, but if you want to destroy caretakers and a factory, its not a problem.

+0 / -0

esainane

3 years ago

No timeline editing happened until 11:22, when the Locusts had already been visible to rewdrfe2 for a few seconds.

In the real game, rewdrfe2 did notice and immediately responded by queuing up a Lotus in base with the base constructor. There was no need to edit the factory build queue, since Reaver does fine against Locust, especially with support.

In the counterfactual world where Nimbus was revealed instead, nobody is going to try to flash-build Lotus to prepare for Nimbus, they'll be making a Razor instead. The factory will have at minimum have dedicated AA queued up, and possibly have current projects cancelled to make them happen faster. Being conservative with the response, no project was cancelled here.

+2 / -0

MmmmM

3 years ago

#subscribe

+0 / -0

Title:	MM 4727: 1v1, Rank Singularity
Host:	Nobody
Game version:	Zero-K v1.9.2.0
Engine version:	104.0.1-1544-ge1f249f
Battle ID:	1062721
Started:	3 years ago
Duration:	63 minutes
Players:	2
Bots:	False
Mission:	False
Rating:	Competitive