1. We have added a Gift Upgrades feature that allows you to gift an account upgrade to another member, just in time for the holiday season. You can see the gift option when going to the Account Upgrades screen, or on any user profile screen.
    Dismiss Notice

Pollution bug <- NAILED

Discussion in 'Civ1 - General Discussions' started by darkpanda, Oct 14, 2014.

  1. darkpanda

    darkpanda Dark Prince

    Oct 28, 2007
    Although I never put it at the center of my focus, stories about the "pollution bug" have been kept roaming at the back of my mind ever since I became active on those forums, and read people mentioning it here and there.

    Being more or less done with the City Process routine (I'm not done yet, in fact, but a bi tired ;)), I thought it would be more fun to investigate this so-called pollution bug one way or another.

    So, first thing, I looked up the forums for related threads, and here is what I found:

    Some of the above links contain detailed game logic, extracted from "Rome on 640K a Day" regarding pollution, namely how known Techs influence pollution, how a city's power source influence chimneys, how population influence pollution, as well as shields production (or did it?).

    Anyhow, I had already identified a very small routine in CIV.EXE, that I bluntly called "AddPollution(x, y)". It seems this routine is used anytime pollution should be added to the map (nuke, city process, ...).

    Back-tracking the call from the City Process routine, I could finally dig out what I believe to be the cause of this famous "bug".

    Here the raw piece of assembly, for the die-hards out there:
    Spoiler :
    		seg007:633F     seg007_633F:                            ; CODE XREF: cityProcess_cityID_mode_?1_?2_?3+62F7j
    		seg007:633F 10C                 mov     ax, cityShieldsProd_dseg_705C
    		seg007:6342 10C                 cwd                     ; AX -> DX:AX (with sign)
    		seg007:6343 10C                 mov     cx, dseg_6C18_CityPowerType
    		seg007:6347 10C                 idiv    cx              ; Signed Divide
    		seg007:6349 10C                 sub     ax, 14h         ; Integer Subtraction
    		seg007:634C 10C                 mov     [bp+var_pollutionProd], ax
    		seg007:6350 10C                 mov     ax, 1Ch
    		seg007:6353 10C                 imul    [bp+arg_cityID] ; Signed Multiply
    		seg007:6356 10C                 mov     bx, ax
    		seg007:6358 10C                 mov     al, CityData.ActualSize[bx]
    		seg007:635C 10C                 cbw                     ; AL -> AX (with sign)
    		seg007:635D 10C                 imul    pollutionFactor_dseg_C7A2 ; Signed Multiply
    		seg007:6361 10C                 cwd                     ; AX -> DX:AX (with sign)
    		seg007:6362 10C                 xor     ax, dx          ; Logical Exclusive OR
    		seg007:6364 10C                 sub     ax, dx          ; Integer Subtraction
    		seg007:6366 10C                 mov     cx, 2
    		seg007:6369 10C                 sar     ax, cl          ; Shift Arithmetic Right
    		seg007:636B 10C                 xor     ax, dx          ; Logical Exclusive OR
    		seg007:636D 10C                 sub     ax, dx          ; Integer Subtraction
    		seg007:636F 10C                 add     [bp+var_pollutionProd], ax ; Add
    		seg007:6373 10C                 mov     bx, cityOwner
    		seg007:6377 10C                 shl     bx, 1           ; Shift Logical Left
    		seg007:6379 10C                 mov     ax, perCivTechCount[bx]
    		seg007:637D 10C                 imul    difficultyLevel ; Signed Multiply
    		seg007:6381 10C                 cwd                     ; AX -> DX:AX (with sign)
    		seg007:6382 10C                 sub     ax, dx          ; Integer Subtraction
    		seg007:6384 10C                 sar     ax, 1           ; Shift Arithmetic Right
    		seg007:6386 10C                 sub     ax, 100h        ; Integer Subtraction
    		seg007:6389 10C                 neg     ax              ; Two's Complement Negation
    		seg007:638B 10C                 push    ax
    		seg007:638C 10E                 call    random_N        ; Call Procedure
    		seg007:6391 10E                 add     sp, 2           ; Add
    		seg007:6394 10C                 mov     cx, [bp+var_pollutionProd]
    		seg007:6398 10C                 shl     cx, 1           ; Shift Logical Left
    		seg007:639A 10C                 cmp     cx, ax          ; Compare Two Operands
    		seg007:639C 10C                 jg      short seg007_63A1 ;

    This code block is one of the first condition to be assessed when checking whether pollution should be spawned.

    Re-writing the condition in a single formula, it gives this:
    IF ( 2 * CityPollution > Random(256 - CityOwnerTechCount * difficultyLevel / 2) ) THEN AddPollution

    In the above, CityPollution is computed as explained in other threads:
    CityPollution = CityShields / CityPowerType - 20 + CitySize * PollutionFactor

    CityPowerType depends on the power infrastructure in the city (default/power plant = 1, hydro/nuke plant = 2, recycling center = 3)
    PollutionFactor depends on the "polluting" techs known by the player (INDUSTRIALIZATION=1, AUTOMOBILE=2, MASS PRODUCTION=3 and PLASTICS=4), and whether the city has a MASS TRANSIT (= 0).

    So we can see that multiple combination are possible, but the key to this bug lies in the RANDOM() call: what if CityOwnerTechCount * difficultyLevel / 2 is more than 256?

    Then the RANDOM() call has a negative argument, for which it will always return zero.

    In other words, as soon as one gets (256*2/difficultyLevel) Techs, pollution starts to get triggered at every turn.

    Let's make a list:
    • Chieftain = 0: (256*2/0) = +infinity -> pollution bug never triggered
    • Warlord = 1: (256*2/1) = 512 -> pollution bug triggered after Future Tech. 444 (512 - 68 non-future techs)
    • Prince = 2: (256*2/2) = 256 -> pollution bug triggered after Future Tech. 188 (256 - 68 non-future techs)
    • King = 3: (256*2/3) = 170 -> pollution bug triggered after Future Tech. 102 (170 - 68 non-future techs)
    • Emperor = 4: (256*2/4) = 128 -> pollution bug triggered after Future Tech. 60 (128 - 68 non-future techs)

    The theoretical numbers seem to more or less match what people have reported in the various threads above, although some other unknown conditions may add variations.

    Anyhow, for me, the riddle is solved.

    The riddle in the riddle that isn't solved, though, is: was this done on purpose? I guess we'll never know... ;)

    Now, how would you guys out there suggest to patch it?
  2. Tristan_C

    Tristan_C Emperor

    Aug 16, 2006
    I can't speculate on what might be wrong, but this formula suggests that if a Mass Transit and Recycling center are both present, a city of any size can produce up to 60 shields before 2*CityPollution > 0. I guess if we're dealing with a &#8805; or the engine is allowed to generate a signed random value (you'd know better than me) then I think the bug would occuring without fail through that formula.

    This is a magnificent find by the way. One patch proposal would be to throw a line returning the lesser of CityOwnerTechCount and 68 and use that value in place of the direct TechCount— freezing the effect of future techs.

    But if the formula is exactly right, then I think the in-game, unmodded work-around is more than sufficient: ensure that cities have the Transit and Recycle and aren't breaking the ceiling with shields. If you can get to future tech 60 than surely you can afford a recycling center...
  3. SWY

    SWY Prince

    Oct 3, 2014
    The Netherlands
    Isn't it possible (and realistic) to just not take future technologies into account when calculating total advances? I would hope that the future brings us cleaner and greener technologies.
  4. darkpanda

    darkpanda Dark Prince

    Oct 28, 2007
    I originally thought that the CIV.EXE's Random routine would simply return 0 if the argument is negative, but my Java port of this routine will actually return negative values when the argument is negative...

    If it returns negative values, then it makes the bug all the more prominent since even when benefiting from 20 free chimneys, 2*CityPollution might be higher than the Ranom result!

    Needs to be checked in-game...
  5. Tristan_C

    Tristan_C Emperor

    Aug 16, 2006
    So as things stand, I suppose this would be a prime suspect on why the bug is occurring. Speaking from some experience (future tech 64 on my part) this bug is very prominent indeed, but one big issue is lack of documentation. Few games are played out to this point and of those next to none have been walked through carefully for testing.
  6. darkpanda

    darkpanda Dark Prince

    Oct 28, 2007
    Yes this is what Tristan_C is proposing by replacing "CityOwnerTechCount" with "MIN(CityOwnerTechCount, 68)"... But to implement this in assembly is difficult, mainly because of the lack of room to put the patch bytes.
  7. Svetkavitsa

    Svetkavitsa Chieftain

    Sep 12, 2010
    Utah, USA
    The simplest patch would be to either replace the difficulty with 0 (always chieftan), or to change one of the modifiers to push the bug back further (divide by 4 or 6), but those both reduce all pollution everywhere, affecting game difficulty.
    IF ( 2 * CityPollution > Random(256) ) THEN AddPollution
    IF ( 2 * CityPollution > Random(256 - CityOwnerTechCount * difficultyLevel / 4) ) THEN AddPollution

    A cool solution (doubtful to be able to be done in place) would be to take the absolute value before calling the rand(), so that future techs eventually reduced pollution. There would still be a period where pollution would be rampant...
    IF ( 2 * CityPollution > Random(abs(256 - CityOwnerTechCount * difficultyLevel / 2) ) ) THEN AddPollution

    Something that might be possible and more in line with what the developers wanted, would be using PollutionFactor*multiplier instead of CityOwnerTechCount, where multiplier would probably be 16 or 17 (researching plastics gives you the same value as having all techs researched). Now, I'm familiar with assembly (from a CS class almost 10 years ago), but far from fluent, so I'm not entirely sure if PollutionFactor is available for use at that point in the calculations. It would also be nice if mass transits didn't factor in (otherwise having one would give you Chieftan difficulty pollution levels), but I think it would be awesome if it could be implemented this way.
    IF ( 2 * CityPollution > Random(256 - PollutionFactor * difficultyLevel * 8) ) THEN AddPollution
  8. Folket

    Folket Deity

    Jan 7, 2010
    Would you not be able to add a procedure at the end of the file and rewrite the function there? Then just call that.
  9. darkpanda

    darkpanda Dark Prince

    Oct 28, 2007
    I think I would be able to do something like that, but it is definitely not simple - and I didn't try it yet.
  10. MiGaNuTs

    MiGaNuTs Chieftain

    Jan 3, 2011
  11. Tristan_C

    Tristan_C Emperor

    Aug 16, 2006
    As we know (bump), on Emperor this returns 0 at Future Tech 60.

    This RNG call as described returns a range,
    [0,-2] at FT 61
    [0,-4] at FT 62
    [0,-6] at FT 63
    ...and so on.

    A city with 0 production and a Mass Transit is defined as having -20 pollution as per the CityPollution formula... -20 being the integer 'buffer' added when computing pollution from shields.
    So if I am reading everything correctly so far, when you reach FT 71 (range [0, -22]), even a 100% clean city has a chance of adding a pollution tile.

    As an aside, and just to be clear, at FT 71 a city generating 60 shields, with Recycling and Transit, shows exactly 0 smokestacks on the city screen. But it only has about a 4% chance of not adding a pollution tile. With higher FT the chance of polluting continues to approach but not reach 100% as it becomes less and less likely the RNG will roll you a 0 in range [0, -x].

    This leads up to a question. What happens at FT 189?

    At FT 189 the CityOwnerTechCount (which includes the 68 non-future techs) is over 256. Does it overflow? It tells the engine to return the range [0, -258] on paper, but does it? If it's possible to bug this value twice, then it's possible to research beyond the pollution bug and eliminate it.

Share This Page