A ginger's thoughts on Spam-Building...

Hey guys,

So I'm writing to you in the midst of a completely unbearable game, where the lag is so terrible that it makes the game not worth playing. In fact, the only reason I haven't just quit yet is because I want to see if I can get one more nuke out the window before I do (while writing this, my secret silo was discovered and I was killed).

I think it'd behoove this game to discourage bot/vehicle spamming. From what I've seen, players rarely upgrade their units to the 2nd tier, because it's more effective to simply build 20 Vehicle Factories and spam tanks (or bot factories / doxs). This results in what I like to call "stop-motion gameplay," where the server just can't even calculate what the f*** is going on because there are way too many units.

Lamer games who I won't name here remedy this with unit limits of 200 or so, and players obviously hate that. I don't think you should put a unit limit.

However, I think you should make the defensive turrets (like the Lobber, etc.) do way more splash damage than they currently do. If I have 2-3 Lobbers (or even P... Pummelers? you know the ones), they should decimate tanks at an incredible rate. Right now, they don't take out very many at a time. This means that you can swarm over them and they're basically useless.

Buffing these def. turrets up would discourage people from simply spamming their cheapest unit, and would encourage people to upgrade to the 2nd tier or try some different attack strategies, making the game (in my humblest of opinions) more enjoyable.

Also, the game would (hopefully) be able to be played as it is intended, instead of the stop-motion gameplay that every game devolves into.

I simply couldn't imagine the server trying to handle a multi-planet battle if everyone is simply building as many factories as they can, then spamming as fast as they can.

Rule number 1 of Alpha Club, you do not talk about performance.

Rule number 2 of Alpha Club, you do not talk about per-, oh wait, I got that wrong... second rule is no smoking.

Advanced units aren't meant to be better units. They're intended to be more specialised at what they do.

Also, network bandwidth has been my biggest problem (I live in a 1st-world country that has 3rd-world Internet, try to guess which it is).

Yeah I ignorantly used an EU server instead of Australia (I'm in Hong Kong). After switching to Australia, I find games to be much more playable, or at least they don't hit the stuttering point until a lot more units are in play.

I believe (correct me if I'm wrong) that if you're experiencing stuttering in that the game still seems to be running the same speed, but you're only seeing snapshots of whats happening, that is, units are jumping as they move rather than a smooth progression, this is highly likely to be bandwidth related.

If the simulation itself was being slowed down, then you would still see very smooth movement, but time would appear to be much slower.

The advanced laser turrets chew pretty quickly through t1 tanks, a few of those in a bunch make a really hard target to punch through. For both turret types though, depending on where they are first facing, sometimes by the time they've turned to face their threat they are nearly dead which is no good.

I think turrets in PA have a different component of balancing compared to other RTS as well, in that most other RTS games, a turtler can generally rely on the edge/corner of the map for impassible terrain. In PA, there's a real danger that creating a wall of turrets is not going to be sufficient unless you create the wall 360 degrees around your base. This is a new problem which the devs will need to balance for, but I look forward to their solution.

My comment wasn't about performance, really, more about gameplay. That being said, no matter how much they optimize the game, in a 5x2 game with 250+ tanks each, the network is going to bring the server to a crawl while it tries to calculate all the interactions.

This I disagree with on a fundamental level. Advanced units are supposed to be better.

Yes, I see where the confusion is coming. The consequences of having something like 1.5k or 2k units is twofold. One of the problems stems from network latency, as you said. The other, however, stems from the computational power on the server-side, which is what I was referring to.

Games of this caliber do not calculate damage, etc. on the client-side because that would lend itself to hacking and manipulation on the client's machine. Instead, there is a simulation running on the Ubernet servers which calculates damage, movement, etc. and then informs everyone of what's happening. There were too many actors in the game for the server to keep up with what was going on, and that's what I was experiencing.


I still believe that it would be better to increase the splash radius and decrease the damage attenuation from the impact point.

Not only from a performance standpoint, but also from a gameplay standpoint. I love having epic battles with lots of tanks / bots / planes, but I'd like to see a bigger incentive to get those advanced units in play and also demand a better strategy than "build as many tanks as you can."

They are fairly formidable, but I'd still like to see splash damage.

Your note about the turning times is interesting, though. I think this would be well-remedied with a modification to their range. As far as I know, they won't turn to face their opponent until it is in firing range. If there were a larger "attention" radius, that would give them some time to get facing their target.

This would be esp. helpful if used in conjunction with Radar or Adv. Radar.

Then you disagree with Uber. They have stated they wish to reduce redundancy of most units to keep them relevant. Advanced units will provide unique or additional attributes, and will in most cases not simply be Tank + better gun. Part of this reasoning, aside from the good but laborious materials of @Nanolathe, is to reduce redundancy of development time.

Games of this "caliber" do in fact calculate damage on the client-side, as Supreme Commander did in an attempt to reduce network traffic, see http://en.wikipedia.org/wiki/Lockstep_(computing). The downside to this approach was that the game ran as slow as the slowest PC in the game. They also bring other challenges such as port-forwarding nightmares. I'm not fully aware of Uber's decision to go back to the client-server model, but it certainly gambles that network traffic is not a bottleneck today as it was in 2007.

There are also a number of possible factors to remember when evaluating performance:

- Alpha is highly unoptimized.
- Alpha servers appear to be fairly underpowered, at least the specs of the AWS instances I read about.
- Bandwidth is (or was at one point) artificially restricted to encourage bandwidth optimization.

I'm not saying simulation speed is not an issue, however in my experience, I've never noticed the game speed drop indicating server power was a bottleneck. I've only ever witnessed stuttering or "dropped frames" where units jump around, indicating bandwidth constraints.

The third option here of course is simply low FPS on your local machine.

*reads* oooaaaahhh...

In this game, t2 is more specialized, but aren't stronger repeats of t1. T1 you get bread and butter, t2 you get single snipe utility. Yet, this makes t2 required for you to even get artillery.

I sometimes think it should somehow require multiple branches of t2 to upgrade, because currently people build a single t2 air engi for access to all the mexes and artillery they want. True complaint.

If one looks at this game, it is very easy to build mexes energy and a factory, and make that factory build the engis to build 5 other mex-power-factory sections, and spread like the plague. It is easier and the purpose to build large army on this game. The metagame is composition and strategical movement in order to get inside the enemy. Covering surface area will always be the way its played.

However, its easier to do now. T2 Mex is overproduction. Abundant metal spots. The such. When things are touched on, it will take longer to get the same high economy, and hopefully impossible to have an effectively infinite economy. The game is surprisingly well balanced atm with absolute time, with how factories have a big delay each unit even if it build instantly. A larger delay will encourage more factories rather than heavy assist on one factory. All these things will extend time before armies achieve the size of over a hundred units, hopefully it will take longer than fifteen minutes to do by game launch.

Btw, have you tried to learn to flow with this play model? Try the building of five mex 4 power a bot factory, and then infinite queue engi bots, then send them in groups of 5 to establish more pockets of mex power factory to infinite produce fighters on large circle of patrol and tanks and stuff like that. You can still quickly upgrade to t2 if you like at same time as you're second factory.

It has also been stated by Uber that, at least at a fundamental level, that multi-threading is implemented as a thread-per-planet and that current single planet games are only single-threaded as a result. The ultimate solution may be more complex than this, but this is a good opportunity for coarse multi-threading, as long as the single-planet game isn't forgotten in the excitement

T1 and T2 turrets are very effective against tank or bot spam.

T2 Levellers are actually a very good counter to t1 spam. Upgrading to t2 is essential if you want to last longer than 30 minutes.

Buffing lobbers will not promote people trying "different attack strategies" it will just promote turtling and slow, slow attrition through artillery defense and nuke bombardment. They work just about fine at the moment.

Remember nothing has been optimised regarding how much data is being sent around and the servers are currently bandwidth limited.

*grumble grumble*

That whole sentence just gets my goat.

Perhaps add "in it's current implementation".

Needs to be the first thing changed as soon as new units are added IMO.

Specialisation is fine for t2, but for the higher cost you reasonably expect them to be superior to t1. The important thing is that t2 not make t1 irrelevant - ie there should be cases where 1000 resources poured into a large set of t1's is strategically superior to spending that on a small set of t2's.

This is completely the wrong way to think about things IMO. If something is trying to 'make up' its cost then increase their drawbacks, make them more specialised by making them less able to cope outside that area of specialisation and lower the cost!

(I'm aware that you're not thinking that way glinknot, I'm just using that thought as an example)

Advanced units do not need to cost your entire economy to produce, they don't need to cost more than Basic Units, nor should they be able to account for their cost unless they're being used correctly.

Advanced units should be 'undercosted' for how powerful they are within their area of expertise, and 'overcosted' for how powerful they are outside their area of expertise.

Well, this one doesn't. See this quote from Neutrino, where he says "the clients don't do much other than render."

And the whole reason they have an Alpha is so you can tell them when you disagree with them, and that's what I'm doing here.

roflroflroflrofl. yeah, my machine is handling it just fine, my friend. the stuttering I'm referring to is not accompanied by a drop in FPS, because what I'm experiencing is due to the server crunching numbers, as mentioned before and cited above

Please give us a reason why you disagree. Just stating that you want 'Upgrade' Technology Tiers doesn't help your argument, nor Uber to recognise their 'error' in not doing it that way.

For the record; I'm firmly against an upgrade tiered technology system since it changes quantities of play, but changes nothing when it comes to the quality of play.

Perhaps I wasn't clear but I thought I stated that PA's networking model is contrary to Lock-stepping:

Fair enough, I'm just letting you know the Uber have stated their stance on the topic, and the majority of forum feedback to my knowledge is behind their choice on a basic level. I'm not telling you you're wrong, I'm just informing you you're going against the grain so expect resistance.

You have not accurately or concisely described the lag you are experiencing, and the term lag is ambiguous so what am I supposed to do other than list the potential causes for "lag"? I apologize if you took my listing of a third option as some sort of attack on your integrity as a human being.