The New Guy Newbie Guide to Fully Non Automatic Manual Play - Distant Worlds, The Manly Man Way to P

[Tormodino]

I know very little about coding, but could this problem be partially offset by weighting decisions differently? The number of ships moving around on the DW map seems excessive at times. If there were, not necessarily dramatically so, fewer objects moving around and performing the same tasks, wouldn't the number of calculations also be fewer?

Too many objects competing for the same algorithm considerations generally limit the effectiveness of each solution, do they not?
Not so much a solution as perhaps an alleviation of the computing problem could be to somehow throttle the number of objects performing the relevant tasks in the game.

[FingNewGuy]

Anyways, an i7 doesn't have 100 billion node neural network trained for billions of years to do the job unfortunately

Nothing human has been trained for billions of years. [8|]

[Gregorovitch55]

ORIGINAL: FingNewGuy

Anyways, an i7 doesn't have 100 billion node neural network trained for billions of years to do the job unfortunately

Nothing human has been trained for billions of years. [8|]

Well, since the Cambrian 600 million years ago really, billions is an exaggeration - the first multi-cellular species in the Cambrian started the process of building navigation systems and we've just inherited them.

[Tormodino]

I love the way you think

[FingNewGuy]

Well, since the Cambrian 600 million years ago really, billions is an exaggeration - the first multi-cellular species in the Cambrian started the process of building navigation systems and we've just inherited them.

I hoped that's what you were getting at. But, yeah. [:)]

[Gregorovitch55]

ORIGINAL: Tormodino

I know very little about coding, but could this problem be partially offset by weighting decisions differently? The number of ships moving around on the DW map seems excessive at times. If there were, not necessarily dramatically so, fewer objects moving around and performing the same tasks, wouldn't the number of calculations also be fewer?

Too many objects competing for the same algorithm considerations generally limit the effectiveness of each solution, do they not?
Not so much a solution as perhaps an alleviation of the computing problem could be to somehow throttle the number of objects performing the relevant tasks in the game.

I think one problem is players are very sensitive to the AI "cheating", this issue of exploration/scout ships in space 4X being a well known example. I seem to remember a big rumpus between players and Stardock over GalCiv2's AI scouting algorithms, which IIRC did pretty much what you are suggesting originally: it sent it's scouts flying around for show, mostly, and built as many colony ships as it could and sent them straight for the colonisable planets the location of which, obviously, the program knew. Needless to say players took an extremely dim view of this accusing Stardock of flagrant cheating on a par with early Civ games generating doom stacks out of thin air one hex outside the player's FOW. I seem to recall Stardock publishing extended patch notes on exactly how the "latest" AI routines for handling exploration and colonisation worked, down to C++ function names etc, to counter this charge.

Also I think part of the charm of DW, and indeed X3, is the sense of a huge living breathing universe you get from knowing there really are thousands of individual ships out there going about their business even if you can't see them all the time. It's the sort of immersion people are looking for in a sandbox simulation and i guess that's more important in the end than a few shaky exploration or autopilot algorithms.

[Tormodino]

Good point, but I was not suggesting giving the ai knowledge of the map. I was more pointing towards making the number of objects in the game fewer. 12 exploration ships seem to be the sweet spot. These are not such an issue. The fact that there are thousands of actual "travelling salesmen" on the map, however, must make the algorithm cringe. If the private sector ships were on average better armed, better protected, bigger and possibly travelling in armed convoys (private sector fleets), would this not necessitate less spamming of ships since their survivability would increase and the neccesity to constantly reevaluate their state would be diminished?

The mechanics for making the private sector have fleets is already in place for empires and pirate factions. Couldn't the same logic be applied to create a smaller and more cpu-manageable private sector? I might be off mark, but this seems for me to go directly to the issue of ship spam and optimizing the movement of assets in game. Not so?

[Gregorovitch55]

I think the vast bulk of ship movements in the game are simply of the "Fly there, shoot that" or "fly there, deliver, that" in response to events or circumstances where the flight end points are known. The exploration thing is different and because it does noticeably worse than we can everyone talks about it. If you scan this java program for solving the Traveling salesperson problem the scale of the problem becomes clearer:

http://www.sanfoundry.com/java-program- ... algorithm/

They key is that there are two nested while loops in the main function, TSP. In the example you have 9 cities and the corresponding 9X9 matrix of distances to work with. So it loops 81 or 72 or 100 times, dunno, haven't looked that hard, but it could probably do that quicker than you could notice if you ran it. However in DW we have up to 1400 starts each with maybe 10 planets and moons. So our distances matrix is 14,000*14,000 = 196 million loops. It might finish that by the time you finished the weekly shop, or maybe by tomorrow morning. And that's for one scout. Let's limit each of the 50 or so factions to five scouts each and we keep the total down to 49 billion loops.

Sure, they'll be a lot of optimizations one could do to this, but basically we're not even remotely in the ball park of an acceptable game algorithm here, this approach is a bust - we need something to decide where each explorer goes next more or less instantly, we need something like the autopilot algorithm from X3 above, two or three simple computations, hopefully one, but almost always less than five, loop per tick. A completely different approach to the problem.

Whatever we come up with it won't work anything like as well as the TSP solution and people will laugh at it on the forums, but if the explorers go explore and they visit practically all of the planets and moons they are supposed in whatever order, we'll call that a major result under the circumstances.

But to answer your question directly, when you find a really, really simple algorithm something like that autopilot one to do this, whether you run it on 10 ships or 100 is immaterial really 'cos it'll do it so fast the player won't notice, so you might as well have 100.

[Tormodino]

Okeydokey. I'll just have to take your word for it. I simply do not know enough about this to make a counter-argument.
What it boils down to is that a simulation on the scale of DW must function somewhat similar (read:quick and dirty) to the way it is currently set up to actually function as a game?

[Gregorovitch55]

Well, IMO it becomes an issue is where we humans are extraordinarily good at something without really trying: we can look at an explorer on a DW map and see instantly what the best exploration flight plan is. Another example is diplomacy: we are absolute masters (on this planet anyway) at driving bargains, lying, cheating, bluffing, detecting lying, cheating and bluffing, and indeed making friends if we want to. These things are very hard to simulate because they are inherently extremely difficult problems that happen to be vital to our survival so we've evolved specialized neural circuitry to handle them. On the other hand when it comes to distributing resources around your empire or handling 15 defense fleets in the face of 20 angry pirate factions simultaneously I would wager the DW AI does better, on average, than the average player would.

It's the instantly thing, I think. We can see the right flight path instantly without thinking about it so we are not conscious of the immense sophistication of the modules evolved over millions of years we use subconsciously to do it. Therefore we find it hard to accept that the problem is so difficult to solve whereas in fact we don't really understand how our own minds work half the time, if we did we probably could solve this sort of thing easily.

[Andy06r]

ORIGINAL: Spidey

I did some testing with with a +23% weapons research bonus, an energy research station, and a multi-school scientist.

Energy station with no weapons lab on weapon research bonus = no bonus at all.
Energy station with 1 weapons lab on weapon research bonus = +23% research bonus.
Triple-school scientist at energy station with 1 energy lab and 1 weapon lab = +18% weaps, +49% energy, +0% HT.
Triple-school scientist at energy station with 1 of each lab = +18% weaps, +49% energy, +23% HT.

So there you have it. A scientist needs access to at least one lab of the right kind to provide a bonus. Any kind of research station with at least one lab of the right type can provide a location bonus.

Where it gets tricky is the interaction of multiple starbases.

If you have a 20% weapons black hole lab, your empire will get +20% to weapons. The 18% on the tri-shook scientist is wasted because the game uses the maximum site (location plus staff) as the bonus.

I typically reduce the labs in my starbase and then add one of each to the research stations. You pay a little more in maintenance, but you avoid the eggs-in-baskets problem.

[BlueTemplar]

Looks like that there's a dilemma for DW : on one hand the defining feature for which it has been praised is the sheer number of systems and ships doing each their own business, but on the other hand this makes it impossible to apply any kind of computationally-heavy algorithm to each of them.

Ideally, you could have different, more computationally-intensive algorithms when there are less objects involved (early game or smaller maps), but then this means that CodeForce would need to develop and maintain multiple or more complicated algorithms... and for what? So that the (tiny?) minority of people caring about this (which am I a part of) is satisfied?

[janamdo]

Informative and extensive manual, but very timeconsuming.
I don't know how other people handle this information, but it seem to be not very strict structered the manual
It is a lot of stuff to learn, and takes time

[corwin90]

Thanks much for this guide. I'm enjoying the game much more since I read this guide and went full manual. Thanks again!

[Mortimer14]

ORIGINAL: EvilMonk3y

So just a couple of questions for anyone willing to help me...I have got to the stage where I am exploring the nearest star systems and have found some planets suitable for colonization and systems with a lot of resources in them.

1. If mining within a system do I also need to have colonized a planet within that system? My goods ships (the ones going back and forth from mines to planets) currently do not have warp drives in them, should I be designing ones with this capability (if I can?).

2. The BIGGEST problem I currently have is that I have found myself short of one resource that has halted my production of any warp enabled construction ships (and a lot of other stuff) which means even though I have now explored this resource I am unable to actually get to it to begin mining. Is there any way I can buy some amount of it to kick-start everything?

Thanks.

Not sure if this has been answered yet (still have 4 pages of replies to read) .. each type of hyperdrive (or other component) uses a different resource. If you are missing a resource to build a ship, try changing your ship design to use a different resource. Then when you do find the missing resource, change back to use whatever is cheaper/faster/stronger.



Somebody posted above a question on how to design passenger ships. What I did was start a game with level 4 tech as a starting point and full automation. I let it run for a minute and then checked to see what the AI did for passenger and colony ships. From this I loaded up my fully manual game and built my colony and passenger ships based on what I saw automated. I then tweaked the design a bit.

As to what passenger ships actually do...they move passengers from one colony to another, or from a resort to a colony and back. Those passengers may be emigrating from the original colony - reducing its total pop - and landing on the new colony - increasing pop. They may also be salesmen looking to set up new markets or servicing their customers. In the view of the game, it doesn't really matter why they are on a ship, they just are. The ship is private and the passengers like it that way.

[auryx]

Just wanted to say, as a nooby noob, I've just completed this guide and loved it. Brilliant way to learn the game! I will never automate....never. Now I can shave my neckbeard with pride!

auryx

[Keston]

ANSWER 1: This has considerable tactical importance depending on the nature of the opposing fleets.

ANSWER 2: AI handles colonization star hopping across voids poorly with colonization range limits, so Irregular is considered easiest for the AI because the stars are relatively evenly distributed.

ANSWER 4: Damage Control does not stack, but a backup is useful if the first is destroyed. Robotic repair bots each make repairs, so multiple components make repairs simultaneously.

ANSWER 5: It is important to understand all the functions needing energy on a ship or base. Energy generated goes into energy storage (like a "battery") and is drawn from there. Excess energy accumulates until it reaches the maximum storage capacity. Collectors work only when stationary[unless this has changed recently), and depend on the strength of solar radiation. Reactors are always available. This is why reactor output and not collector output is used to determine the excess energy number in green.

Beyond the static energy draw, usage is uneven because WEAPONS fire and recharge on different cycles, SHIELD recharge may kick in, THRUSTERS may be used, and maybe the HYPERDRIVE needs to be spooled up for a high-speed exit. Weapons "energy per second" is a statistic assuming simultaneous fire, but in fact they operate on different cycle times. Energy is also needed for SHIELD recharge (the rate is listed in the design), for the THRUSTERS, and to spool up the HYPERDRIVE. Most ships can't do all of these fully at one time.

ANSWER 8: Mining and Gas Mining Stations are "state" bases that can be built and retrofitted by the player.

Mining: I remember the old rule that marginal output drops a lot with the 4th Mining Engine for early tech, and that improvements mean fewer may be required before reaching the production limit. Check that Gas Extractors work the same.

[Skyjack]

Hi guys, new member and player here.

First let me say a google search brought me to this thread. TIMOTHEUS: thank you! Never tried Automatic, never will. What a great tutorial!

So questions for all:

1) so it's confirmed that having more than one lab type in a star base or port is a waste?

2) systems with existing peoples are marked with the grey dotted line? Without knowing, I colonies done of these but now when I look years later, I can't see the indigenous people...have I already absorbed them?

3) why do troop transports not fully load to the gills assuming I have enough troops on the supplying colony?


Okay I'll stop for now. I'm sure I'll have more questions in the future.

Thanks for any help.

[morik]

Hey guys, I just got this game last night and did a few tutorials. I plan to follow the guide on page 1 a bit... curious if the 3rd section is still in the works? Seems like its been a year and a half or so....

BTW: I'm planning on trying the AI improvement mod (sounds like it solves the 'AI is bad at designs' problem).
Can't post link cause new account... but if you type the forum's address (up to the
'tm.asp?m='), the message ID (to put after m=) is 3647528.


EDIT: I couldn't find a few things from the guide...
The 'newly built ships are automated', the default engagement stances, the fleet attack settings, nor discoveries... not sure where these went, have they moved with some new version? I am looking in 'empire policies' (I don't see an 'empire settings' button).

EDIT 2: D'oh, had started the game already. Found it in the options menu of the main menu.

EDIT 3: Regarding the "First assemble when this percentage of fleet dispersed". Shouldn't this be the inverse of what the guide says?
I.e., the lower this is, the more often it will trigger an assembly?
E.g., if set to 80, that means a whopping 4/5 of the ships have to be dispersed already to trigger assembling, right?
And the default 30 means that if at least 1 in 3 ships is missing, the fleet will assemble first?
(So setting to 1 would mean that if even 1% of the fleet has wandered off, they won't engage before assembling...)

BTW, how does this actually take effect? If I have manual control over everything, does this mean if I give an attack order, they will automatically either wait for assembly or go refuel if needed, and THEN perform my attack order?
What if it is an emergency? Is there a way to override and say "no you bastards, all available ships go immediately!"?

[Krystyn]

Picked up DW/U a few days ago on Steam. Was almost immediately intimidated and bewildered. Then I found Timotheus' incredible post/thread and felt like the Universe had tossed me a life preserver.

Thank you SOOO much for your outstanding help (and humor!). While I'm nowhere near being past the newbie stage, I'm way past where I would have been on my own, and having a super time. Heck, I haven't even had to restart after a Pirate wipeout -- yet...

[:)]

Krys