Victoria 3 - Dev Diary #68 - Patch 1.1 Changelog (part 4)

[tahlaskerssen]

I hear you on that ultimately it does come down to the player experience of the system, but if we do have a fundamental balance issue in the economy system - which is not unlikely to be the case - doing a proper fix early in the game's lifecycle is definitely preferable to a smoke-and-mirrors solution. We also want to take care to not give countries full control of every nuance of their trade policies without pushback, since it's completely contrary to both modeling the international market forces that lead to economic imperialism and the political interests of your pops. I have no doubt we'll find a solution to the balance issues we currently have with trade, we just need a bit more time.

Lovely stuff, thanks. I agree that a lot of control is also not healthy.

 

[sloanjh]

First, thanks for all the hard work, both in this patch and overall ! I spent a lot of hours on Victoria 2, and the improvements in the game play, especially in the underlying economic modeling, are truly fantastic!! Great job!!!

That being said, since this seems to have turned into a bit of a "I hope you fix XXX next" thread , I wanted to call out the suggestion by @neroden here (sorry - spam filter won't let me post real link): forum/threads/victoria-iii-most-of-the-world-has-no-ports-1920s.1551913/ (that would make Convoys a "normal" good Ports a "normal" factory that convert Clippers into Convoys in order incentivize the AI building Ports) and ask if it's on your radar for 1.2 or beyond.

I think it's a great idea, and that it has the potential to also lead to a fix for an issue that I don't see a way around: I don't know how to open a trade route with non-neighbors if I'm a land-locked country.

If you extended @neroden's suggestion (as he suggests) to also include land transportation (i.e. transportation is represented as a Caravans/TrainRoutes good that is produced by Railroads), then I think the Convoys-as-a-good mechanic has the potential to solve the land-locked trade issue. You could set up the game mechanics so that if a country had a neighbor with ports, the cost of the trade route would include paying the neighbor for Caravans to transport it across the appropriate states, and paying someone's Convoys to send it overseas if necessary. A different version of this idea would to allow a country to negotiate a special kind of "transportation" trade deal with another neighboring country for access to purchase that country's Caravans (for transit across land states) and/or a maritime 3rd party country to purchase Convoys (for transit across sea states) that would then be consumed as part of a normal trade route for goods. A trade route would be possible if there was a transit path where for every leg of the path either the buyer or the seller had such a transportation deal in place. I suspect this sort of mechanism would have a lot of potential to simulate "carrying trade" economics (in the sense that there's a 3rd party to the trade route who is supplying the transportation) and/or the Great Silk Roads.

I did see a "yes, we're looking at this" message in the original thread, but just wanted to bump it here. Thanks!

 

[foamingjetty]

I see lots of good changes here. Thank you - it's great to see progress on so many issues!

When people ask me whether I recommend the game, I've responded that the issues I care about are on the roadmap and I hope that the major patches bring this title to a better state. So all of this is positive except for one glaring tidbit: the war system and automatic removal of generals from fronts is such a severe problem that it basically requires general teleportation as a manually-applied hotfix. I've said many times that if the hotfix is removed before the need for it goes away, I couldn't play or recommend the game until these are brought back in alignment.

​

- Fixed general travel exploit that would allow to teleport generals by alternating orders towards the same destination
- Fixed a general travel duration exploit that was possible when giving several orders targeting different destinations, causing GENERAL_TRAVEL_OVERSEAS_SCALE to sometimes not be applied
- Fixed a bug where making a general return to a previous location during travel would sometimes apply GENERAL_TRAVEL_OVERSEAS_SCALE

This is critical - if war and frontlines are waiting until 1.2, then players need to keep the current capacity, however unintended, to hotfix their own general behaviors via the teleportation bug until then.

Have I missed a dev response to the repeated player posts on this subject in the patch thread? I didn't see one

 

[mongolio]

So many basic things in 1.1

 

[Sighfer]

What about it?

Italy gets formed reliably and quietly by Two Sicilies, despite in reality being a French project for Sardinia-Piedmont that involved quite a bit of fighting Austria.

Prussia is currently turning the French and Germans into fertilizer on the Rhine for four years straight with two million dead each. While trying to annex a German minor with 250k population.

 

[Amorenkaire]

This essentially lets them execute a number of profitable transactions "before" the final market prices are actually set, leading domestic market agents to have to buy or sell at (probably) unfair prices by comparison, particularly when one market is much bigger than the other.

If I could make a suggestion, when this is looked at could that sort of behavior be left as an option for the right economic or trade laws? Quite a few countries during the time period pursued a policy of aggressive exportation of certain resources, usually to fund government expenses. Most notably the USSR's grain exportation to fund its technological and industrial imports from the west.

 

[alex huang]

When do you guys get the chance to fix the 'regime change' war goal? Right now, it just changes nothing(only a few laws)

 

[TempestM]

When do you guys get the chance to fix the 'regime change' war goal? Right now, it just changes nothing(only a few laws)

I think this also requires an update which actually makes government in other countries matters. Like, you're a monarchy, and your neighbor became council republic. So what? There's nothing worth fighting for, his government doesn't affect other countries. The only way to use it is rp as I see it now, if you want to spread revolution (which can rather hurt you than help if everyone will go isolationist and protectionist)

 

[durbal]

The pricing function is linear, ranging from 0.25 to 1.75 of base price and reaching those bounds when sell / buy is 2:1 and 1:2 vice versa and progressing linearly based on the ratio between buy and sell orders in between. Not sure where the claim to a concave pricing function is coming from, this is pretty well outlined in both tooltips and the defines file (the PRICE_RANGE and BUY_SELL_DIFF_AT_MAX_FACTOR defines in particular).

The price inversions are rather originating from traders optimizing their volume based on an average of pre- and post-trade prices, but then applying the full brunt of this volume as Sell and Buy orders in the respective markets. This essentially lets them execute a number of profitable transactions "before" the final market prices are actually set, leading domestic market agents to have to buy or sell at (probably) unfair prices by comparison, particularly when one market is much bigger than the other.

That's fine for one country, but when accounting for trade route volume between markets this is no longer the case since the prices will vary according to the sell and buy orders in their respective market (in other words, Russia selling 1000 wood out of the 10000000 they have to Timbuktu will flood their market and tank the price to the minimum, but if the opposite happened the price in Russia would barely budge). You're correct in stating where the price inversions come from (albeit in a slightly different way), but what I'm trying to explain is that using that straight arithmetic mean doesn't produce 'good' outcomes just like using some other arbitrary value doesn't produce good outcomes. If you want a mean such that the accurate revenue is the mean price * volume you'd need to have very tiny steps and sum them, or since that's just basically integration then you could integrate (see @Tempscire's post below and I'd urge the dev team to take a deeper look at the analysis in that thread even if just to tell us we're all totally wrong).

Thanks to some insightful posts in this thread (especially by @Ekcrbe ), I think I have found a solution which would solve the problem while not requiring a complete rework. The idea that some sort of average value should be used to calculate profitability is sound in my opinion. Because if you export as much goods as needed to balance the prices between the two markets then you should at least generate some profit along the way. Otherwise the prices should have been the same even before trading. The problem lies more within the particular choice of average price. The average price is the arithmetic mean of the prices before and after trading. This would be correct, if the function that represents the price difference between the two markets was linear. But the price functions in Vic3 are not linear. Therefore, the total amount of profit is just not the volume times this average price.

As an example consider the following:
Say A sells 500 units of good G with base price 1 and buys 100 units of this good, while B sells 100 and buys 100. Now A exports the good G to B until the prices are equal,i.e. it exports 200 goods.
Then the original prices are P_A=1+0,75*((100-500)/500)=0,4 and P_B=1. After trading both have the price P_A=P_B=1+0,75*((100-300)/300)=0,5.
If the prices where linear we would then have the average prices in the exactly in the middle of this trading process, i.e. when 100 goods are traded.
But then we get P_A=1+0,75*((100-400)/400)=7/16 and P_B=1+0,75*((100-200)/200)=5/8.
Both of these are clearly not the average prices, as 7/16<0,45=(0,4+0,5)/2 and 5/8<0,75=(1+0,5)/2.

The correct profit is more complicated to calculate. As it is not linear a simple product does not suffice and the correct way to is to take the integral over the price difference function from 0 to the trade volume V.

So let us determine the price difference function PD: [assuming that the price formula with max is correct]
Let t be the trade volume, BP the base price and B_I,S_I,B_E,S_E the buy/sell orders of the importer/exporter
PD(t)=BP(1+3/4*(B_I-S_I-t)/max(B_I,S_I+t))-BP(1+3/4*(B_E+t-S_E)/max(B_E+t,S_E))=3BP/4*((B_I-S_I-t)/max(B_I,S_I+t)-(B_E+t-S_E)/max(B_E+t,S_E)).

The profit E(V) for trading V goods should now be the integral from 0 to V of the function PD(t) over t.
I.e. E(V)=int_0^V PD(t) dt.

The function PD(t) is continuous in t and defined on 4 (non-necessarily) distinct subsets of the real axis. Therefore, the integral exists and on each of these subsets the function can be expressed as a linear combination of rational functions. Therefore, the integrals are fairly easy to compute. I have computed all the relevant integral for PD as above per hand, if there is any interest. If tariffs or transport costs enter the picture, the formula has to be adjusted accordingly. Currently this amounts to a sum of natural logarithms, constant and linear functions, so it should not be that hard to implement. [I do not know how this is implemented in the new engine, but if it is necessary, there is always the option to approximate the ln parts via taylor exopansions or similar methods.]

The definition of the profit given above fulfills all the desired properties to a good degree, in particular:
(i) The profits are a lot higher than without averaging,
(ii) It remains profitable to expand the trade route as long as there is a price difference,
(iii) It is always unprofitable to export goods to a market with a lower price, as then the integrand is smaller than 0.
(iv) It is not profitable to create loops, as at least one of the trade routes will not be profitable, as the price in one market will be higher than in the other one and (iii) holds.

There is obviously a lot more detail in the integrals, but I wanted to keep the post short and mostly free of formulas. In addition writing down integral equations in the forum is a nightmare. If anyone is interested I can do a write-up of the relevant math with a better tool to display the formulas at some point.

 

[MarkBrandenburg]

Legitimacy has a tendency to drop steeply during the election period, due to IGs leaving parties and thus increasing the size of government. Is that going to be a problem or is legitimacy ignored during elections?

 

[Soranya]

I see the trade changes and it looks like you 'fixed' circular trades while leaving the ability to raid someone else's market by buying high and selling low (i.e. the price inversion problem). Now we have no workaround to re-import goods being effectively stolen from our markets when fixing price inversions would've killed both birds with one stone.

I hope this is not true. But I fear it is.
Same with the teleporting generals exploit -> Fixed the easy bug that made the huge bug bearable before tackeling the huge bug. ;(

 

[Soranya]

Same here. Some of these problems really killed the fun for me, so I haven't played in a while. Can't wait for 1.2 to finally dive deeper than ever in this game. Paradox games are always like wine, after all: the older the better.

Until they turn to Vinegar...

 

[Soranya]

Fractal rebellions and uprisings should also be fixed, yes! Countries created due to civil wars should no longer be able to have their own civil wars.

Wouldn't be a baetter solution to allow multiy way wars (3 or more sides fighting in the same war?) - this would solve a lot of issues with break of wars, rebellions and so on (which have been a problem in all PDX games so far)

 

[bitmapmedivh]

Does anyone know if the rulers start to die? I won't touch a game where my king lives for 110 years and gets a bunch of negative modifiers.

You can try reading the changelog in the OP?

 

[TempestM]

Wouldn't be a baetter solution to allow multiy way wars (3 or more sides fighting in the same war?) - this would solve a lot of issues with break of wars, rebellions and so on (which have been a problem in all PDX games so far)

Considering it's probably very hard coded (since no pdx games have 3ways) any workaround will probably be much easier to do than allow this

 

[Soranya]

Considering it's probably very hard coded (since no pdx games have 3ways) any workaround will probably be much easier to do than allow this

An Engine upgrade would be interesting then.
Could solve thsoe oos/numberoverflow issues as well... (properly designed/done from a softwareengineering perspective)

 

[andseanfor]

The pricing function is linear, ranging from 0.25 to 1.75 of base price and reaching those bounds when sell / buy is 2:1 and 1:2 vice versa and progressing linearly based on the ratio between buy and sell orders in between. Not sure where the claim to a concave pricing function is coming from, this is pretty well outlined in both tooltips and the defines file (the PRICE_RANGE and BUY_SELL_DIFF_AT_MAX_FACTOR defines in particular).

The price inversions are rather originating from traders optimizing their volume based on an average of pre- and post-trade prices, but then applying the full brunt of this volume as Sell and Buy orders in the respective markets. This essentially lets them execute a number of profitable transactions "before" the final market prices are actually set, leading domestic market agents to have to buy or sell at (probably) unfair prices by comparison, particularly when one market is much bigger than the other. We'll try to address it like I said, but please understand that even if a different algorithm was to work better and look more sane at large volumes, there's no guarantee this will magically solve all problems without causing new ones.

The pricing function isn’t quite linear. If there are 10 buy orders for Wheat, each additional sell order from 5 to 10 will decrease the price by 0.15. Each additional sell order from 10 to 20 will only decrease the price by 0.075.

Think of it as 1:2 and 4:2 rather than 1:2 and 2:1. The distance between 1 and 2 is half the distance between 4 and 2, yet the distance between 0.25 and 1 is equal to the distance between 1 and 1.75.

I don’t know the details of how you calculate your trade routes but that shift in slope will throw off your calculations if it isn’t taken into consideration. Odds are you’ve already thought of this but I figured I may as well say this in case you hadn’t.

 

[gkbba]

I definitely understand that perspective but I'm not sold on the thesis that every real life historical outcome represented the most likely scenario. Over the course of history plenty of 30% and even 10% outcomes have actually happened and shaped what came after.

yes, exactly.

I see a lot of suggestions focused on buffing mechanics such that the historically accurate result is almost forced. I personally find it more interesting to have the game represent the historical dynamics well (it also adds replayability). of course some might feel more immersed if AI produces the same results most of the time, thats also a perspective.

I think an example where this actually happened is the unification of Germany: Because Prussia gets German minors for free it will also form Germany most of the time, which is a good way to make the historical result more likely but belies the process a bit, which was not some semi-colonization mixed with the leadership question (which is represented fairly at least and gives Austria a shot on forming Germany), but rather a bottom up political movement (so far nationalism and pan-nationalism mostly is a tech that allows for these things rather than actually ideologies that affect things) that was more hindered by the leadership question of Austrian and Prussian rulers who didnt want to give up power in favor of a unified Germany.
@PDJR_Alastorn already stated that some flavor might get added though, so I am hopeful in that regard.

 

[anbeck]

​

- Multiple changes to which notification types display as Toasts (middle of the screen) versus Feed (bottom right)

Could we get an overview of what these are (or rather: what the rationale is for a message to appear as Toasts)? Any chance we might get customizable notification settings in the future?

 

[gkbba]

First of all, these are great changes all around as far as I can see. nice. One pretty big bug, that I dont see adressed however is that defaulting doesnt seem to work properly. is this planned for 1.2?

The pricing function isn’t quite linear. If there are 10 buy orders for Wheat, each additional sell order from 5 to 10 will decrease the price by 0.15. Each additional sell order from 10 to 20 will only decrease the price by 0.075.

Think of it as 1:2 and 4:2 rather than 1:2 and 2:1. The distance between 1 and 2 is half the distance between 4 and 2, yet the distance between 0.25 and 1 is equal to the distance between 1 and 1.75.

I don’t know the details of how you calculate your trade routes but that shift in slope will throw off your calculations if it isn’t taken into consideration. Odds are you’ve already thought of this but I figured I may as well say this in case you hadn’t.

its piecewise linear, which is still "quite" linear. but yes, the slope is 2x steeper for a shortage than for overproduction, which makes sense as a general thing, but could be rebalanced actually.
/edit: to create an everywhere differentiable linear pricing function, would result in either the price for a 50% shortage being just 37.5% higher and the price hitting +75% just at 0 sell orders

also in regards to discussion about linear price ranges I wanted to throw in the idea (pretty sure it has been suggested a few times, just to be sure), of making different orders of price elasticity a thing. linear price ranges are a good abstractions for most goods, not for all though.