Now that this is a huge issue in all match ups involving toss, I wonder if we can expect any sort of changes to address this. Thoughts? It seems particularly bad in ZvP now. Every game is toss turteling to mass air. So boring.
6 Likes
The defacto counter to deathballing is splash. Deathballs happen because armies become exponentially stronger when grouped up. Splash punishes this to balance it out. For example, a cluster of marines could plow through an infinite number of lings if not for banes, which force split micro, which increases the surface area for lings. Skytoss deathballs create this problem on steroids because the air units can occupy the same space on the map and have such high HP that terran/zerg splash options are a complete joke.
They could simply make it so toss capital ships can’t stack, or they can give Terran/Zerg viable splash options. In the past, fungal filled that role combined with infested terrans. They nerfed the ever living crap out of fungal and just straight up deleted the infested terran. They could buff parasitic bomb. Give it +50% damage to shields or make the split micro harder some how like increasing the radius or maybe reducing the duration of the effect by 50%. Another option is to buff the ravager’s bile. It, combined with fungal, can be used to punish Protoss deathballs. If they brought back rooting for fungal, you could 100% beat skytoss using ravagers and infestors.
For Terran, they just need to change the relationship between the thor and the disruptor. Thors beat skytoss if not for disruptors. They should consider an upgrade for the Thor which makes it impervious to disruptor shots. Call it “Reinforced Armor” or something and have it cut the disruptor damage by 75%. Disruptors could still hit everything else, such as tanks, just fine. It just means terran’s anti-air can’t be bothered by such a long-range siege AND splash unit.
2 Likes
That happens due to the nerffs that each and every protoss unit got those 3 last years. Basically, protoss race is not able to take serious offensive moves before having 150-180 supply. Also we have no luxury of attacking with only 2-3 type of units like terrans.
Our compositions are an order of magnitude more complicated because our units suck in isolation and have to overlap and cover for each-other shortcomings (Colossi have to be covered with Stalker or/and Phonexies).
This delays the unlocking of protoss power and by the time it happens protoss has reached beyond 150 supply.
The design of the race is such that it requires a protoss serious force to be a deathabll.
Creating 2-3 self-sufficient Protoss battle formations is extremely difficult.
2 Likes
It’s not that bad, though. atleast within my diamond league. I was pretty happy to play all kind of strategies except for cheese, DTs and turtle carrier.
But since you can get to GM with sentries only, well…
That’s a very misleading statement. Yes a 6000-6500MMR high-GM can do that (and much more). But that 6000MMR player would be obliterated if he tried to pull the same trick with a player of the same skill.
That is valid for all three races…
1 Like
That’s unequivocally false. Those “X to GM” series are extremely misleading. First of all, you have a person who is much more talented than the minimum threshold for GM, so they do take a huge performance hit as a result of doing such a bad strategy. Whether or not you can get GM with sentries depends on if you are substantially above the minimum GM threshold. The idea that “you” meaning anyone can get GM with sentries is an absolutely absurd proposition.
Second of all this only works over the short term. The strength of a strategy in the long term only goes down as people run into it enough that they start to plan for it and craft better and better responses. Playing against Ruff, for example, and his very strange 1 base ghost rush, was quite difficult for me for awhile. I would be sitting at 5.7k then lose a game to a 5k terran just because his build was really strange. Eventually I ran into him enough that I was able to test a variety of responses and came up with something that counters him with a 95-99% win-rate.
The idea that mass sentries are a viable strategy at a GM level is plain false. It works only if given every advantage conceivable including a massive skill disparity and the opponent being completely unprepared for such an unusual play style. 99.99%+ of games would not meet that criteria - the ladder has to be artificially manipulated to put a much higher skilled player at a lower rank just for it to even be possible. That doesn’t represent the organic nature of SC2 in the slightest.
That was one of my biggest criticisms for AlphaStar, by the way. The long-term viability of a strategy is very low because opponents adapt to how you are playing and they didn’t test AlphaStar’s long term performance. Getting a certain ranking temporarily is completely different from being able to hold it long term. Long term, maintaining a GM rank is about adapting faster than your opponents. Since there is a very high cost to update / retrain the AI’s it’s extremely unlikely that AlphaStar could maintain its GM ranking due to its inability to adapt as fast as humans can. For some people, all it takes is running into a new strategy 3 or maybe 5 times and they can then beat it with a very high win-rate. AlphaStar has to train quite likely hundreds of thousands of bots in probably a billion games or more. That’s because it’s essentially a randomized process meaning it makes randomized changes to the bots and then tests them to see if it increases their win-rate. It has absolutely zero clue how the game works which means it can’t make targeted adjustments to its play to adapt to specific instances. It has to make loads and loads of randomized changes until, statistically, it finds the 0.00001% of them that improves the play vs a particular strategy. This mode of “problem solving” is called numerical computing and it simply tries to solve a complex problem by spamming such a huge volume of potential solutions at it that, statistically, it is bound to find something eventually.
3 Likes