New Discovery? Costless Units!

[DarthVeda]

So what's the max shield cost in MGE/ToT?

250?

[klesh]

I am gonna have to have some time to play around and get some findings. Beautiful day here, and the gf wont stand for me sitting in front of the pc. I will kepp testing though...

Henrik, yes it does... for units anyway. Entering 0 for a cost and looking at the shieldbox for the unit, it was a standard depth 10 shield unit.

All of this is being done with MGE as that is all I have. You guys will have to test the ToT and FW ones...

I'll also check on if you get money or something from building these units. More to come....

-FMK.

[klesh]

Well I snuck some time for another quick test. The results are promising indeed.


When set to build these costless units any city really can build them. It isnt some graphic screw up where the units really cost 10 shields. A city that made 2 shields per turn could pump out these units.

Also, even rioting cities, which should have zero production anyway, can still pump out these units.

The units are supported as usual from the home city. Which makes me think that in the fanatic slot under a fundy gov., this technique could be deadly making free support/free build units from small little 'base' cities off out in nowhere or somesuch.

Now here is something strange indeed, when using the edit city function in cheat mode to reset cities's shield accumulation to zero, I found that cities which had just made one of these units had accumulation in the negative(-521 shields)! So I let those cities continue. They built another unit the next turn no problem. As I look now after the next turn... all the cities' progress is back to zero... no negative values anymore. Obviously, more in depth testing is needed. I would also like to see if the AI will choose these units to build. I will make their other stats (at/def) very appealing so they will want the unit. I would like to see if they are turned off to building it at all or something.

I also want to see how this effects the AI's use of cruise missile units, as it is commonly know they use a cost anaylsis to judge whether to send out a missle to strike.

Many things to test, and no time to do it right now. Argh!


-FMK.

[hardjoy]

Very interesting discoveries, FMK! Keep up the good work! I will be following the test results closely ...perhaps even do my own bits of testing! intriguing to say the least..

[ravagon]

I did some playing too and got some rather odd results.
Firstly I was NOT able to reproduce the screen with the NONE unit cost. Still don't know about this one. Costs were always negative.

Unit costs of zero default to 10 shields (although it doesn't do this for improvements strangely enough where 0=0) and follows the usual pattern up to 127 (1270 shields)
At 128 (2^7) it flips to -1280 which rounds up to give the zero cost unit and thereafter continues increasing. ie: 129 --> -1270 etc, up to 200 which seems to be the max and has a shield cost of -560 (IIRC). Hence all unit costs above 127 are 0.

[This flip happens in the terrain resources too btw at the same point. 127 = 127 shields. 128 = -1280 (rounds up to 0 production)]

It works this way for the AI too, even though AI cities seem to produce a full shield box every turn (Graphical anomaly?)

An AI oddity that showed up however was a strong tendency to NOT attack with zero-cost units. I was at war with them and cheated in a horde of spies to keep an eye on their city production from turn to turn and my spies/other units were never attacked/expelled. This went on until I made peace. The AI then began expelling spies like crazy before sneak attacking with its zero-cost units on both my spies and other units but only for this turn.
The next turn business as usual. ie: His units just moving around not attacking spies or others in open terrain right next to his cities.
[Incidentally I was using the Jihad scenario and the fundamentalist and horseman slots to test all this]

One interesting thing that came to light was the ability to rush-build these zero-cost units. Cost was 0 gold and it produced the negative 1280 (using the production cost = 128) shields required to build the unit (although this didn't show up in the production box).
If you the changed the production queue to something else it showed a much much higher number of turns to build the new something due to the negative shield deficit that had to be "worked off" before applying shields to actually build it.

This might be ideal if somebody were designing a scenario and wished for cities to be unable to build anything for a number of turns before having their production resume at full strength.
It is still very cumbersome though as the minimum shield deficit seems to be 560 (from rush building a unit with a cost of 200) which is quite sizable.
All of this is completely irrelevant however if civtweak can set a cities shield production to a negative value (which I forgot to check )

[Edit: Actually, rereading FMK's post, he covered most of this anyway ]

[DarthVeda]

Oooh.... that will be useful...

[Goingonit]

Neat.

IMO, the AI douesn't take cost into account when building units.

[Boco]

Watch out for Civ2 Editor resets

To make a 'slow road' terrain, I had set the movecost to 205 (Road bonus =25, set to >10 to take advantage of WK's fractional movement math). Civ2 behaves oddly but reproducibly with these settings. Regular roads (and rivers) cost 1/10 mf; the new 'slow road' costs 2/10mf.

Here's the kicker: the Civ2 terrain editor squawks when it encounters an out-of-range variable. I was tweaking the 'slow road' terrain image via the Editor menu in Civ2, when it spotted the movecost was set out of range. It did me the 'favor' of resetting the movecost to 0 in the rules.txt. I wouldn't be surprised if this were not an isolated incident.

[DarthVeda]

Watch out for the civ2 editors period. They are far too buggy to be trusted by any serious scenario creator.

I recall several incidents where it changes the name of your civ to that of a resource and vice versa.

[Heresson]

Nice discovery, I'm sure it'll be usefull in making some computer-played civ properly dangerous for You. I don't see any use for it
in human-played civ though; most of scns are enough easy already.

[Boco]

*Bump*

Has anyone used costless units in a scenario?

[William Keenan]

Greetings all. Its been awhile.

I think I can help shed some light on this topic. Most Civ unit values are stored as signed byte integers (-128 to 127). This conserves computer memory but produces odd results when larger than expected values are inserted.

More in a moment .....

[Goingonit]

That gives me a new idea...

Ravagon was suggesting in the engineer/city-production thread that a Xin Yu or similar strategist could disband a city by selling its city walls and causing it to be destroyed by combat. Well, that would be a lot more balanced strategically if the city walls had negative cost! (and thus cost money to sell!) I'll test this, but it sounds practical and useful.

[William Keenan]

..... back.

Computers do math using a base 16 system (hexidecimal), not the base 10 system (Decimal) we use.

One byte is 2 hexidecimal digits: 00 to FF

Decimal values: 0 to 255.

When a byte is signed decimal values are different
00 to 7F hex is 0 to 127 decimal and 80 to FF hex is -128 to -1 decimal. See Ravagons explaination above.

Civ2 uses signed bytes in unit values. It follows that a unit cost of 255 in the rules.txt is converted to a hex value of FF which to the computer is -1. Therefore the unit build cost would be (-1*10)=-10.

A city producing only one shield per turn will produce this unit evey turn because the total accumulated shield production of 1 is equal to or greater than the unit cost of -10.

Theoretically, a city with a -9 shield production would also produce the unit every turn. But a city with a -11 production would never produce the unit.

Originally posted by DarthVeda
So what's the max shield cost in MGE/ToT?

250?

The highest possible positive number should be 1270.

[Goingonit]

Darn, it turns out that this doesn't work. Improvements, at least, have positive and accurate costs to 255, and them positive but inaccurate costs thereafter (the cost is equal to the set cost mod 256, which leads me to think that the cost is stored as an unsigned byte.)