Randomness in early game

I'd like to see spawn position choice made fair, instead of spawn positions being mathematically fair. Right now players are limited to pre-existing spawn points at certain positions, and I think that changing this would be a better solution than trying to balance some pre-defined spawn points on a randomly generated planet, or changing the planet itself. The obvious problems with free spawn choice would be that players may want to choose the same spawn position (or spawn positions very close to one-another), especially if metal spot density can vary as much as it does now (hopefully that changes as well). I'm not sure what the best solution to that problem is, but I don't think it's giving players a set of predefined start points as it is now.
As for planet generation, I think it needs to be improved but not changed as drastically as some are suggesting. Random maps do add depth to the game, so long as the variations can affect the game in an important way. I'm hoping that we get improvements that will lead to this fairly soon.

I feel like the randomness is part of what makes this game infinitely re-playable. Maybe the system needs tweaking , but fundamentally I like the idea for spawns as they are now. (still waiting for spawns on different planets... ^_^)

For competitive play, I agree that spawns need to be fair, but there can be special maps for this, much like those in Sins of a Solar Empire. Either way, I think a map editor would solve this problem pretty completely.

I have to disagree with "only one good spot".

On many maps the option to place your base on an island or peninsula can have a huge impact on how you play. Your enemy doesn't know you're on an island and you can focus your early game into naval and air without worrying about his DOX swarm. This can also help you get off world more quickly as you won't be investing as heavily into defenses and early ground units.

But it's equally important that this remains a choice, not a mandate. Some players may prefer the center of a large land mass for fastest possible expansion on the starting planet, but in PA that's not the only way to win.

The issue at the moment is that players don't rally have much choice and can occasionally find themselves at a disadvantage. In my experience, the relative disadvantages tend to be quite small. I have yet to see a replay where one player is at an irrecoverable disadvantage across all their spawn points compared to the other players, excusing the obviously unintended bugs where the algorithm has clearly broken itself (such as spawning metal inside mountains).

I find that the only people who have a real problem with such disadvantages tend to be those who care the most about pure competition, while many people are willing to put up with the risk of being at a slight disadvantage, in exchange for getting to play a different game each time. Those who enjoy the competitive side tend to want to focus on improving their skills and timing, and are able to gain satisfaction from developing better execution of a repetitive set of tasks. Others find variety and adaptability as more engaging qualities. To put it bluntly, there are a lot of games which cater for the former, and few which cater for the latter. I would personally hate to see PA lose one of it's strongest unique points in trying to out compete other games which do a far better job of catering to that audience. I recognise that this is just my own personal preference for this, and it's one that often isn't in the majority on the forums, as forum-goers will tend to over-represent the competitive scene while the others are a silent majority. I do believe that PA can and will cater to both, but it's pretty clear that procedural generation is a compromise that some people are just going to have to accept. I'm sure that the competitive scene will be happier once a library of "fair" seeds has been compiled and mirror maps are a possibility, if not a priority.

However, there are still ways to deal with the random nature of spawn points. One way is to allow player choice to smooth over the issues of a bad spawn or two by giving a much greater area to choose from. This works because risk-recognition/managment are still skills that players have a degree of control over, so they don't feel gypped if they still have a large choice.

To do this, the spawning points simply need to be changed so that instead of each team getting a couple of different small areas, they instead get a single large area which covers most of one hemisphere of the planet. For games with more than two sides, the surface of the planet can be split up accordingly. This way, players can choose the most metal rich points or they can decide to take a less metal rich point if they see some other value, such as having a more defensible position. They can also estimate their enemies actions to a greater degree, as if there is a large metal deposit on the edge of their hemisphere, they know there is a greater chance that their enemy will choose to spawn near the deposit and cause conflict. Players would also know that unless they are on the edge of their spawn zone, they are guaranteed to be a reasonable distance from their enemy. As long as the spawn zones are large enough, players will have enough choice to even out inequalities in the procedural generation. If the spawning zones are too small, then chances are that players are playing on a world too small for the number of players, and the game is probably not going to be that good anyway. It's something I would really like to see tested as I think it may fix a number of concerns that people have.

To me, that sounds like a good way to lose the game. And unfortunately, expanding the fastest IS the only way to win.

The questions are :
- Is it "more different each game" than any other RTS?
- Is it related to the random generation?

My answers would be :
- No. It's the same game development without any variation each game, with the same set of units.
- Yes, it's due to the random generation, because each planet generated is a slight variation of a dull, flat map. And playing a dull map doesn't make any game exiting or unique.

They are so many things to do (because so few is made so far) to make the game more interesting than putting imbalance in the game.

If it's still not enough, maybe a random map generation would give a new breath to the game. But we are not remotely there yet (if so, it would be sad, because it would mean that the rest of the game is boring).

After 7 years of existence, the *idea* of generating map automatically was raised for FA.
It's not impossible, but the consensus would be that it's more better to play 500-1000 interesting maps than 1.000.000.000.000.000 of boring ones.
And if efforts have to be make, it's toward a better map editor.

Actually, the guy who has given the example of the "island spawn" is a good one.

That's a rare occurrence (if it was really a competitive spot). And you remember it because it made your game more interesting than usual.

How often these kind of things really happens? How often would it happen if the maps are handcrafted?

Pardon the naivety, but how are the procedurally generated maps any more or less boring than 2D or designed maps? I mean, they have ocean bits, land bits, choke points, mountains...and even the latter features will have more impact as we go along in development. It feels kind of like people are getting misty eyed about the merits of user created maps, when in reality there isn't much in it. Presumably, it's more complicated when it comes to competition play, but just thought I'd voice my opinion, FWIW. not trying to start an argument either, just saying.

You are basically saying that some people prefer to have some random handicap, because in their head, it's challenging?
Well, obviously.
Any kind of handicap make anything more challenging, it's the purpose of it.

That doesn't mean that handicap had to be a core element, and even less a obligation.

Then you are saying that on the other side of the rope, they are players that like to repeat the same thing over and over again.
You should read the "Being good at "strategy"" thread.

Because even you if are what Sirlin is describing as a scrub, at some point, if you are not brain-dead, some more efficients patterns will emerge from the chaos of your "challenging and adaptive starts" (while all they are is disguised handicap selection).

And at some point, you will find that some way are better than other and... oh my god, you've just reached the first category, but in 10x more time and efforts! And basically, you've just learn how to select the difficulty...

That's why I'm saying it's bad : it makes the game harder to learn for less focused players, and the skill gap, at some point, will be that huge that there will be no competitive scene at all (too few players), and the low end of the scope will abandon the game because nobody is making them dreaming about what the game could be if they play better.

Because an human brain can create. A computer can't.
Sure, both can have the same bricks.

Well, it's easy, make a program that can pile up lego bricks. And run it 1000000x times.
How many times the program will do something interesting?

Then give these bricks to a kid. After how many tries will he get something worth the effort?

In a RTS, you can create unique situations that a computer can't do. Simple as that.

Take a simple map like Theta Passage on FA or Loki on starcraft.
Spend 10 minutes to analyze every tactical or strategic situation that emerge from their layout.
You will understand what I mean.

If both players have to adapt, how is it a handicap? Pardon my misunderstanding....

You are assuming that there are strategies that are global optimums under all circumstances. If these are the overall focus of PA, then I would view that as a failure.

I really think you should look up the techniques that can be used in procedural generation.

Fair enough. At my amateurish level of play I only really notice the glaringly obvious map features, and take most of the enjoyment out of the rest of the game. I don't think I would be suited to competitive play lol.

Because it's more than probable than your opponent won't have to adapt, because he already learn the more efficient layout, and will apply his perfect BO on it.

That's basically what happen in any game, but it's exploding in PA because of the extra-layer or complexity (and not depth).

I think you should teach Uber developpers then, because they are not there yet...

I have several good experiences in this regard.. i remember especially one where i chose a spawn with three lakes around me, which i immediately saw as a great way to gain map control and defend my base. Id say terrain in PA already affects strategies somewhat, could be better of course. To me user created maps often feel like they are designed to be played in a specific way. But I'm still sour about XCOM:EU predefined maps so..

I am currently trying to create cool looking maps, and I can say that the randomness makes it horrible. Basically it is trial error and all I want are balanced maps. :S

I will do the what I've proposed :
Let's look at Theta.



This is the top area. A wide open area with a little cliff at the end. On that cliff, two extra mexes.
That area is the vital : It's the main access to the enemy base, and as there isn't a lot of mexes, these two often make the difference.

A big mountain formation is cutting the map in half past that open area.



There is a little opening in the middle on the mountain, leading in the enemy base, near mexes.
It's a chokepoint not really hard to defend, but it can't be ignored.



Another opening on the bottom of the mountain. A chokepoint, but this time, it's on a little bump, making it harder to defend if the enemy is on top of it.
Also, it's leading to the back of the enemy base, and it's far from your base, making it harder to defend.

The difference between winning and losing is often determined by who is attacking at the right time on that chokepoint, and if you are splitting your force to send the right amount of units to defend (too few or too many, and you either lose the back or the middle chokepoint/top area).



The mountain is of course offering a protection that allow you to build a radar behind safely. But they are bombers in the game...


It's really as simple as you can get, and I still, I have yet to see that level of deepness in a PA map.

Other that that underpass, that is all in PA.

Seed?

Of course all these elements are in PA. That's not the point.

I'm talking of what the procedural is doing : The LAYOUT. The layouts that PA is generating are simple, dull, and mostly boring. And on top of it : imbalanced.

You can't beat a map layout designed by a human. I don't get how you can argue with that actually.

That's as dumb as saying that a program could generate a house plan that is as interesting and viable than a decent architect plan.