Changes to the Social Engineering models

[Nexii]

Sure you can. Just give all the faction files that you aren't playing against +3 SUPPORT. The faction you aren't playing as remains as default.

[bvanevery]

That's ridiculous.  "Normal" players aren't going to put up with that level of futzing, just when they want to play a game.  I ship working products to people, even at the $0 price point.  Not broken junk that makes people hate life.

You could legally ship a tool that will perform the file modifications automatically.  However, there's no legal way to integrate it into the game menus.  They'd have to go outside of the game to run it.  I think in the real world, players won't do it.  YMMV, and you're welcome to take the risk of writing and distributing such a tool.  If you even cared about the legal separations.

I'm slightly sorry to sound this dismissive, but we're not even talking about the same level of design professionalism here, on some of these things.  It's one thing to make 'product' for modders who like to hack around with their own *.txt files, and quite another to ship product to players.

[Nexii]

/shrug it wasn't even my idea. Taken from SMANIAC's faction set which is the most popular faction pack to date. It's really not that much work to copy over a few files before a new game. "Monster" factions are popular, people want a challenge

[vonbach]

False.  It takes longer to produce a Clean unit.  When a Clean unit gets killed, it's more of a production loss, due to its embedded cost.  Currently I have Clean units as 50% longer to produce.  I can go to 100%.  I have not received enough player feedback to feel a need to do it.

I'll cheerfully pay more for a clean unit. They're worth more in the long run.

[Nexii]

Yea for any unit that isn't offensive military, like formers or base defenders, Clean is great. Pays for itself extremely fast. Where I have dual ability units late-game, I think it's even better than Super on formers. Which may be a minor issue. Exempting non-combats is more of a wishlist thing..

[vonbach]

Yea for any unit that isn't offensive military, like formers or base defenders, Clean is great. Pays for itself extremely fast.

There are some things I kept from bvanevery's mod and thats the clean formers and the synth garrisons. They are so useful. I like to put it on everything I can though.
Its minerals saved later.

[Nexii]

Another wishlist thing but it would have made a bit more sense if some abilities like Clean only worked when you had +PLANET SE rating. Non-lethal methods kind of works like this, it's only good if you can make use at your current POLICE rating. I forget if Trained was affected by MORALE, I think it isn't. Most of the rest are just combat things which would apply all of the time.

[vonbach]

Another wishlist thing but it would have made a bit more sense if some abilities like Clean only worked when you had +PLANET SE rating. Non-lethal methods kind of works like this, it's only good if you can make use at your current POLICE rating. I forget if Trained was affected by MORALE, I think it isn't. Most of the rest are just combat things which would apply all of the time.

I never really thought about that one. Here's a question. Does the AI build clean units on its own?

[bvanevery]

It's really not that much work to copy over a few files before a new game.

I've been asked to "tone down" my level of disagreement about such matters.  What I will say, is that this doesn't rise to the level of professional software release.  I don't just mod, that's not my goal.  I'm trying to build a reputation.  When you build for yourself a reputation that you don't do "fit and finish", then your reputation only goes so far.

"Monster" factions are popular, people want a challenge

Not relevant to our previous disagreement.  But for the record, I really hate those things.  As well as all the 4X TBS games that implement "difficulty" by keeping the AI dumb, but cranking up the AI resource bonuses to Eleven.  I've "forcibly retired" plenty of units over my many years of gaming.  I don't want or need that.  SMAC does it some.  I mean, I do play on Transcend and couldn't possibly consider playing on a lower difficulty level with the stock binary.  I've squeezed as much AI performance out of the stock binary as I can.  It required embracing things like a thorough understanding of Clean Reactors and SUPPORT.

[Nexii]

The AI doesn't like to make Clean units. It would have to be a basic unit type.

[bvanevery]

Another wishlist thing but it would have made a bit more sense if some abilities like Clean only worked when you had +PLANET SE rating. Non-lethal methods kind of works like this, it's only good if you can make use at your current POLICE rating.

The problem with this kind of fine tuning and making game rules 'picky', is that AI doesn't get written to handle it.  I had to change my Democratic from -2 POLICE to -1 POLICE, because I couldn't get the AI to know when it needed to build Non-Lethal Methods units.  So instead I made all early governments able to use such units, because the POLICE penalty was never lower than -1.  If some factions then had trouble with my Knowledge, which is -2 POLICE, I'm like eh, can only do so much.  They'll just have to suffer in dumbness.  At least I delayed such dumbness, and plenty of factions would never have an issue because they wouldn't pick my Knowledge or Cybernetic.  Although, in my 1.43 it will be Eudaimonic instead.

If you're a binary modder, in principle, you can write the code to handle the case.  In practice, did you do it?  Did you kick your own patched version of Thinker Mod out the door, or get it to adopt your patch?  I'm not clear on just how "open sourcey" anyone's treating the project.  Since I'm not planning on participating in that project myself.

This is the general form of the problem of why AI is bad in 4X games.  People think up neato game design rules to implement, and no AI person comes along to handle that.

[bvanevery]

Does the AI build clean units on its own?

Yes, but, in the stock game it cannot be accused of being a conscientiously good designer in that regard.  I've seen it hand me designs, and use designs, that made me think "why did you bother" ?  One possible excuse is that on Transcend, the AI may be paying such lowered unit costs, that it doesn't really have to be careful about designing anything.

When I last played straight Thinker Mod more than a year ago, it did design some units better for some purposes.  I can't comment greatly on what it managed to do, since it's been awhile, but for instance it would make intelligent use of a Fusion Reactor if it had that capability.  This is something I can't get the stock binary to do.  Probe teams, for instance, kinda have to suck as fission.

If one is willing to require Scient's patch, that particular problem can be solved with predefined Fusion unit designs.  But I'm not willing to require Scient's patch.

[vonbach]

I've been asked to "tone down" my level of disagreement about such matters.  What I will say, is that this doesn't rise to the level of professional software release.  I don't just mod, that's not my goal.  I'm trying to build a reputation.  When you build for yourself a reputation that you don't do "fit and finish", then your reputation only goes so far.

I understand  the desire for professionalism believe me. I'm a painter and there are projects I'll want to do over again just because I can do better.

[bvanevery]

I have love-hate with SMAC, going on 2 decades now.  It bankrupted me in the early 2000s.  When will I ever tame the beast??  Can't / won't do much more for $0 anymore, that's for sure.  Trying to wrestle with its game design, and with 4X TBS game design in general, has been frustrating lately.  But part of me feels that before I die, I should put the 4X TBS genre right.  Perhaps the time is now??  Perhaps I need to proceed on faith, that the "subtractive sculptural approach" to game design can work here.  There is a better game sitting inside this SMAC thing, somewhere.

[Nexii]

Yea you can only mod around the AI so much. It avoids negative MORALE more than it ought to, as a prime example.

Like I said elsewhere, I think AI was probably limited a lot at the time. Not just the algorithms but it has to make calculations very fast. For example, chess engines today are amazingly strong. But even they take time to play at their very best. And that's just one piece on a board, not dozens of cities and hundreds of units. A human playing a 4x game isn't going to sit there and wait 5 minutes for the AI to calculate all the best moves. Chess, sure, because that's just expected.

I think you'll see 4x AI improve a lot more over the next 20 years. Self-play neural networks are getting very strong at this sort of problem.