Powerful UI or Not?

I view this as kind of a must have feature given the scale that PA is aiming for. It's complicated enough to click fast enough to manage everything on a single map. On multiple maps simultaneously it's going to be a nightmare without improved UI features above and beyond what we are used to in a standard RTS.

Keep in mind, this isn't necessarily as far out there as many might think. SC2 very intentionally pruned away many user interface features to intentionally keep the difficulty up at a level they felt was appropriate for intense professional play.

The point there is that they easily could have included many features to make the game easier to play for mere mortals that lack 300+ APM but they chose not to.

With PA I think they will have to rethink where to draw that line a bit. multiple planets is going to be plenty complicated already without additional unnecessary hurdles in the UI to overcome as well.

In practice this has never worked well. It's currently impossible to tell an engineer to patrol and reclaim only. Area Commands will help give the player the ability to explicitly and exclusively repair, reclaim or assist multiple structures in an area with a single command.

This is so hard to get right, it's just easier to give the player tools that let them efficiently tell the computer what they want done.

This is exactly where Area Commands shine. You just select an existing command, and hold down a modifier key to make it applied over an area. The feature is being extended in a way that is easy and intuitive for players to understand.

I'm sure there are many other features that we could come up with that help add efficiency and power without adding complexity.

Not disagreeing with anything you said. If anything, i think you've added qualifiers to what im saying, however, area commands function just like patrol orders.

In TA, and Supcom, the AI still had a very odd priority for what it would attack/repair/reclaim automatically.

I don't think this is a bad thing. I think it takes up some micro slack, but leaves clear improvements for when the player does choose to micro it. However, I don't imagine it would be terrible if AI improvements were made in these areas also.

The reason I mentioned area commands is because they don't work just like patrol orders.

A patrol command tells the unit to walk along a path and target units, buildings or resources and do "what the unit does best" to them. So an engineer will walk around and target damaged units for repair, buildings under construction and factories for assist, and resources for reclaim. The priority it does these things in is supposed to be smart, but in reality it ends up being pretty retarded. It's a hard problem to solve.

Area commands are specific to the command, so you will have an area command for repair that will only repair, and it will target everything exactly inside the specified area, nothing more. I would imagine the priority the unit would target things in would merely be the shortest path to each unit, which is similar to the Travelling Salesman Problem.

It's explicit rather than magic.

Especially considering how situational the context can be.

Mike

I agree with your concern as it is a concern I have too. At no point do I want the UI capable of making strategic decisions for you. Only the player makes strategic decisions; it is just that the player can tell the UI in advance what those decisions are. Thus, I am very hesitant about any kind of choice/event-based/conditional kind of orders being added to the UI. Even though they may not be strategic choices, it could still lead to people playing the game through coding AI -- which can be fun in its own right but shouldn't be present by default in an RTS. Mods, and showing which mods people are using in a game or even requiring everyone use the same mods, let people write AIs if they want without upsetting non-mod users as much.

I do not see how your first example is 'bad': If the player has to manually specify the building the construction units are coming out of, the location (x,y), as well as the exact kind of building to build, the remaining things the UI does for you is just tedium now. Your second example I would concur: a 'when attacked' condition is a form of decision-making and is an iffy ability to give a UI.

If you wanted just a default patrol, you wouldn't need to do anything complex. If you did want to specify exactly how the patrol is done, then by nature it would require at least one extra piece information from you (a dropdown is 2 pieces of info but theoretically it can be as little as 1). The simplicity/efficiency can be maintained by having the most commonly performed tasks use as few actions from the player; and only if they want something more complex would they have to do some extra work, eg. a 'more...','advanced...','customize order...' button, placing extra options into context menus, etc.

It is a technical problem but it is a design issue as well. The default behaviors feel smart if it does things as good as or better than the player, and dumb if worse. So, when deciding what level of behavior to fix into the core game, how do you choose the smartness? Player ability is highly varied and changes over time for everyone. If the AI is too dumb and stuck that way, once the player improves past it they will get frustrated. If the AI is too smart and stuck that way, lower-ability players would not be required to improve because the computer plays it better than they do.

You can probably guess what I'm going to say next: have the default behavior be a reasonable level, ie. doesn't frustrate people playing the first time, and give the player a way to specify behavior so that the smartness of the units reflects player ability and grows better as the player improves.

From the technical side, I honestly don't know a lot about how to code a good AI. But I am pretty sure it is possible to do. After all, look at good old Unreal Tournament. Some of the smartest bots ever in a game. Actually often complained about as being too smart, and too good.

I know some pretty amazing stuff is possible there. And at least part of it I think is a question of developers intentionally holding back because they don't want the game good enough to play itself. But with PA, with the huge scale of it, better UI, better AI. these are things the game is going to need.

Without it, jugging a battle across numerous planets with units simultaneously doing things on all of them at the same time? It's going to be overwhelming. You need to keep the game at a level where it is playable.

As for it being challenging for the pros I'm not worried at all, the game is going to be plenty complex no matter how far they go with the UI and AI in this one. I'm sure the pros will still find the game plenty interesting. The only trick there is going to be balance.

People are going off in this somewhat unrelated direction of having an AI do work for the player. That's not really at issue for whether the UI is abstract or not.

Having an abstract, conceptually powerful UI lets the player assign specific orders to units, no AI intervention required. All the AI has to do is copy the commands into many units at a time, or duplicate a build template, or otherwise implement a structure that is completely player-defined.

We don't need AI at all to have a UI that gives us the kind of high-level control over large armies that we want.