Fix exponential inflation

[michael blunck]

[...]

Yeah, nothing new, you already pointed this out in 2011 (or before): http://www.tt-forums.net/viewtopic.php?p=971953#p971953

But even then, most vehicle set developers have always been aware of this problem. E.g., the DB Set had a warning message about this (that it needs "startyear 1921") from the beginning.

Like so many other problems, this one is due to extending temporal and spatial limits of the original game in OTTD. Maybe it´ll help making the inflation setting readable by newGRFs.

regards
Michael

[Robbedem]

I have a brain wave after reading Robbedem's suggestion about pre and post 1950 calculation. How about making inflation a proportional exponent of the passed game years divided by the amount of years between 2050 (default end year) and the start year rather than the amount of passed years. To prevent unwanted effects from raising to the power of a fracture it can be expressed as a percentage. In short, instead of doing:

(inflation) = 1,1 ^ (amount of passed years)

do:
(inflation) = 1,1 ^ (((current game year counted from the start year) / (amount of game years from start year to 2050)) * 100)

You mean 1,02 instead of 1,1?
While this would indeed be an improvement, there would still be a difference for games started early:
start 1800
in 1950 cost = x3.28
income = x1.82
-> ratio = 1.8

start 1900
in 1950 cost = x1.93
income = x1.39
-> ratio = 1.4

[Robbedem]

Srry for double post, but wanted to show this graph

from 1800 to 1900 there was almost no inflation, perhaps that would be a solution? Start inflation only from 1900.

[Eddi]

any chance you could provide this in a "useful" file format?

[PikkaBird]

I have a brain wave after reading Robbedem's suggestion about pre and post 1950 calculation.

If there's one thing I've learned from creating NARS2 and UKRS2, it's that complex calculations with subtle effects are a waste of time.

The choices are;

A) Have non-divergent inflation.
B) Have rebalanced, epoch-based divergent inflation, give newgrfs the ability to detect the inflation setting, and get newgrf authors to provide balanced vehicle statistics for both no inflation and divergent inflation.

B may be more satisfying from a theoretical point of view, especially if you're not the one that's going to have to code it. But the net effect of "no inflation divergence" and "inflation divergence, but newgrfs will compensate for it" is almost exactly the same. The only difference is that the latter creates a lot more unnecessary work for newgrf authors.

If NewGRF authors want vehicular costs to increase over time, of course, they can do that already, without relying on a one-size-fits-all built-in model. UKRS2 does this. But personally I think it's a bad idea, from a gameplay point of view.

[Eddi]

How about these suggestions:

1. Have a fixed inflation rate (4%-ish) and instead make the income/cost spread as a configurable long-term-difficulty setting (current 1% is really difficult, 0% would be "pure eye-candy" inflation, make up steps inbetween as you please)
2. Have the inflation based on a fixed base year, not the starting year, i.e. adjust prices on starting the economy (the starting/max loan will be adjusted by this factor as well)
3. make the current spread factor available to NewGRFs (as fixed-point number), so they can do whatever with it
4. start charging a sizable interest rate on the loan

[Robbedem]

any chance you could provide this in a "useful" file format?

How do you do that?

Here is an excel graph, this time for the US

[planetmaker]

How do you do that?

Images are best provided in a format which can be shown by a browser...

docx or xlsx instead require every user to download a file which is in a proprietary file format, is known to be a major trojan and virus carrier (I don't say your files contain any, but how do I know?), requires everyone to start yet another programme, and worst, the file formats are not made to display graphics.

[Robbedem]

Images are best provided in a format which can be shown by a browser...

I've done some searching, and apparently I need a website to store the pictures on first and then link to that site, right?
What website would you suggest?

[Eddi]

what's wrong with attaching the image to this forum, like you did with the other files?

[Arie-]

I've uploaded the file to Google Docs, it shows a graph diagram which is attatched. I have no idea what the values on the Y-axis represent, the X-axis seem to represent years AD.

[michael blunck]

I have no idea what the values on the Y-axis represent, the X-axis seem to represent years AD.

Looks like depletion in monetary value, probably for the U.K. See here: http://www.tt-forums.net/viewtopic.php?p=875793#p875793

regards
Michael

[Robbedem]

I have no idea what the values on the Y-axis represent, the X-axis seem to represent years AD.

Thanks Arie

The graph is based on info from the Federal Reserve of Minneapolis (USA)
The y-axis represents the annual average of the consumer price index.

[Hyronymus]

I have a brain wave after reading Robbedem's suggestion about pre and post 1950 calculation. How about making inflation a proportional exponent of the passed game years divided by the amount of years between 2050 (default end year) and the start year rather than the amount of passed years. To prevent unwanted effects from raising to the power of a fracture it can be expressed as a percentage. In short, instead of doing:

(inflation) = 1,1 ^ (amount of passed years)

do:
(inflation) = 1,1 ^ (((current game year counted from the start year) / (amount of game years from start year to 2050)) * 100)

You mean 1,02 instead of 1,1?
While this would indeed be an improvement, there would still be a difference for games started early:
start 1800
in 1950 cost = x3.28
income = x1.82
-> ratio = 1.8

start 1900
in 1950 cost = x1.93
income = x1.39
-> ratio = 1.4

It's bot 1 value, the value changes every year. On a 1800 start in the first year it'll be:
1,1 ^ ((1 / 250) * 100)

On a 1800 start in the 56th year it'll be:
1,1 ^ ((56 / 250) * 100)

On a 1900 start in the 23rd year it'll be:
1,1 ^ ((23 / 150) * 100)

But as Pikka said this might be a waste of time and energy . It does solve the issue of not accounting for early starts though imho.

[Arie-]

I think he meant the inflation rate, you calculations assume 10%, he suggests using 2% for example calculations.

Edit: As a suggestion, ignoring realism, but for game purposes, would it be possible to make an agreement which says e.g. inflation only occurs between 1900 and 2100. Before and after those dates the rate would be 0%. Just an idea, haven't yet though about consequences of such a move.

[Hyronymus]

I think he meant the inflation rate, you calculations assume 10%, he suggests using 2% for example calculations.

Edit: As a suggestion, ignoring realism, but for game purposes, would it be possible to make an agreement which says e.g. inflation only occurs between 1900 and 2100. Before and after those dates the rate would be 0%. Just an idea, haven't yet though about consequences of such a move.

Not a bad idea though I think it may lead to questions why it doesn't work if people start earlier.

[Arie-]

Now I've come to think about this, after 2100 when playing with IS Maintenance costs enabled, your network probably is that large these costs start to kick in heavily. Also, if for calulation purposes inflation is always calculated from 1900 onwards, even when starting a game in 1950, NewGRF developers know what interest rates to take into consideration for balancing. A drawback of this would be that already developed NewGRF's may have to be taken a new look at to see whether these are still suited to this new situation.

[Robbedem]

I'm not sure, but doesn't a game always end in 2050? Or are there gamescripts that make it end later?
I suggest inflation from 1950 till game end (when you get you're score).
Perhaps use 2.5% or even 3% for costs and make the difference between cost and income inflation a difficulty setting (0%, 0.5%, ... -> 2%)

[Arie-]

Yes, for 'official' game scores. However NewGRF's make it possible to start much earlier, I think 1900 is a rather acceptable starting date because a lot of sets already provide vehicles at that date.

[Robbedem]

I think 1900 is a rather acceptable starting date because a lot of sets already provide vehicles at that date.

That's OK, too. I just took 1950 because from that time inflation became important in the real world (I posted examples for the UK and US earlier)