Fleet commonality exploit

[JumboShrimp]

This was discussed in game threads, but official bug report was never filed.  So here it is:

Fleet commonality calculation is completely broken.  It allows small airlines have 10+ fleets, less than handful of each with little or no penalty.

It only penalizes large airlines, even if they have 100s of aircraft in common.  Knowledge of this exploit allows airlines grow fast at the beginning, knowing that fleet commonality does not apply to small airlines, only to large airlines.

Airlines growing with a sane strategy, concentrating on growing with few fleets are penalized by much slower growth following what in RL would be a winning strategy.

Please review these posts for for proof of this:

tiny increase for tiny airlines (review any of the MT6 airlines with less than 100 aircraft across many fleet groups).
medium increase for medium airline: https://www.airwaysim.com/forum/index.php/topic,34106.msg184179.html#msg184179
huge increase for huge airline: https://www.airwaysim.com/forum/index.php/topic,34106.msg184238.html#msg184238

As the airline grows and has 100s of aircraft in common, the penalty for adding a 4th type is 576% (retroactive to to first 3).  This penalty does not exist (or is minimal) when airline has less than 100 aircraft across 10+ fleets.  Basically, fleet commonality works in exactly the opposite way as it should.

[Sanabas]

Also see lots of evidence in https://www.airwaysim.com/forum/index.php/topic,34837.0.html and I still have a chunk more data saved on my computer showing the same thing. The bigger your airline, the more you pay (in both straight $ and in % terms) to add a 4th fleet. A small airline adds a 4th fleet, the cost of the other 3 barely changes. A large airline does it, and the cost of maintaining their first 3 fleets can suddenly be 6 or 7 times what it was before.

[JumboShrimp]

Excellent summary is in your post here within the thread:
https://www.airwaysim.com/forum/index.php/topic,34837.msg184443.html#msg184443

[JumboShrimp]

Reviewing the post here:
https://www.airwaysim.com/forum/index.php/topic,34837.msg184443.html#msg184443

Not all fits a straght line, but looking at the last 3 entries:

JA5, Sanabas, 115 planes, 4 fleets = 129% of 3 fleets.
JA5, Sanabas, 108 planes, 4 fleets = 121% of 3 fleets.
JA5, Sanabas, 121 planes, 4 fleets = 101% of 3 fleets

kind of says that the number of fleets is more or less ignored up to about 100 aircraft.  Then, it grows exponentially

MT5, Jumboshrimp: 376 planes. 4 fleets = 355% of 3 fleets.
MT5, Jumboshrimp: 680 planes. 4 fleets = 676% of 3 fleets.

So the exploit is, grab whatever you can, ignore fleet commonality, grab everything that flies, grow to 100 aircraft.  The 100 aircraft will have given you enough profits to have another couple of hundred aircraft on order way before other people who actually maintain fleet commonality.  Once your good aircraft starts arriving from the new market, reduce your fleet groups to 3.

[Sanabas]

Reviewing the post here:
https://www.airwaysim.com/forum/index.php/topic,34837.msg184443.html#msg184443

Not all fits a straght line, but looking at the last 3 entries:

JA5, Sanabas, 115 planes, 4 fleets = 129% of 3 fleets.
JA5, Sanabas, 108 planes, 4 fleets = 121% of 3 fleets.
JA5, Sanabas, 121 planes, 4 fleets = 101% of 3 fleets

kind of says that the number of fleets is more or less ignored up to about 100 aircraft.

A couple of those numbers were actually anomalous, the jump should have been bigger (though still irrelevant compared to what happens for big airlines) according to the rest of the data. There's still an active bug report about it: https://www.airwaysim.com/forum/index.php/topic,35033.0.html

Seems like one of those things that needs a test world airline with god mode switched on, so you can simply add & subtract planes from an airline and see exactly what commonality does. Or get the formula out of the underlying code.

But even with a bug in how the current system is being applied, like a few of those numbers, the working as intended bit of the current system, where economies of scale work in reverse, needs to be redone.

[EGSC]

Fleet commonality seems to be a buzz word here In real world many operators don't care about it and have a multi-type fleet with no really significant burden on their overhead costs, same with multiple base airports. Smth that should be changed in AWS imho.

[Sigma]

Fleet commonality seems to be a buzz word here In real world many operators don't care about it and have a multi-type fleet with no really significant burden on their overhead costs, same with multiple base airports. Smth that should be changed in AWS imho.

Just because some, or most, airlines operate multiple-type fleets doesn't mean that they don't care about it or that there's no appreciable cost to doing so.  Of course there's a cost to operating multiple models; it's basic common sense (though the penalty is about 10,000 times more complicated than how it's done here in AWS).  But there are costs of operating single-fleet models too -- potentially MUCH larger ones.  Namely what happens if some problem crops up on the model that makes up all or most of your fleet -- see American Airlines' Super-80 issues a couple years back.  Or imagine what would happen to Southwest if some issue popped up with the 737.  In reality, those costs (and others, like the security of using multiple suppliers, the acquisition cost advantages of multiple suppliers, and man others) are weighed significantly with the costs of operating different fleet models.  AWS just looks at the operating costs -- all those other variables simply do not come into play here because there are no potential issues that can pop up with a particular model, you cannot pit suppliers against one another, you can't get a supplier to give you a steep discount to "get in the door" with you, etc, etc, etc.  As a result, AWS drives decisions to minimize fleet types, as only financial factors come into play; things like service, security, etc are not modeled factors here.  In real-life, the other factors are weighed in to the decision model, and that's why you get airlines operating two nearly identical models even though using just one would technically save money.

And the exact same arguments come into play with bases as well as some others like efficiency improvements to your existing bases one gets by opening new ones due to decreased volumes, improvements in passenger satisfaction due to reduced travel times with more/shorter options, and numerous other changes that come into play when weighing the costs of an additional base.  Again, those aren't factors here as those aren't modeled -- all we get is the financial side.

Just because airlines all over the world operate multiple bases, doesn't mean it's not a consideration and certainly doesn't mean there's not huge overhead costs related to doing so.  ALL airlines give HUGE consideration to the costs of operating additional hubs -- it's the primary point of discussion when airlines need to cut costs, or when acquisitions/mergers occur and redundancies are sought to be removed, as the costs of operating hubs is staggering.  

Just as an example, when Delta was reorganizing out of bankruptcy, two of its primary savings drivers as part of its $4 billion per year savings package, was closing its DFW hub and reducing its operating fleet types from 12 to 8.

[JumboShrimp]

Seems like one of those things that needs a test world airline with god mode switched on, so you can simply add & subtract planes from an airline and see exactly what commonality does. Or get the formula out of the underlying code.

Well, this is not a black box, that we can't look inside.  Sami has the code, and it is not that difficult to look at the formula directly.  The data we already have:

MT5, Jumboshrimp: 376 planes. 4 fleets = 355% of 3 fleets.
MT5, Jumboshrimp: 680 planes. 4 fleets = 676% of 3 fleets.

clearly shows that the formula does a complete opposite of what it is supposed to.  There is a lot more commonality with 680 aircraft in 4 fleets (simple average of 170 aircraft per fleet) than in fleet of 375 aircraft in 4 fleets (~94 aircraft per fleet).

And the fact that number of fleets is almost completely ignored up to ~100 aircraft:

JA5, Sanabas, 115 planes, 4 fleets = 129% of 3 fleets.
JA5, Sanabas, 108 planes, 4 fleets = 121% of 3 fleets.
JA5, Sanabas, 121 planes, 4 fleets = 101% of 3 fleets

indicates that this is an area that can be exploited.  After all the discussion about ways to limit growth of airlines, the most reasonable one of all is to make it expensive to get 5 or even 10 fleets initially, the system makes it possible (or even encourages) players to get a lot of fleets, expand quickly, and start to consolidate number of fleets when the number of aircraft gets to 200-300 range.

I think this really needs to fixed.  In my opinion, this is the #1 priority (of all the items I see in Feature Requests and in Bug forum).  The current formula is not only contrary to the real world, it is contrary to game playability, and one of the objectives AWS has had, which is to slow down the initial rush, prolong the build up phase, and to encourage careful fleet planning.

[Sanabas]

And the fact that number of fleets is almost completely ignored up to ~100 aircraft:

JA5, Sanabas, 115 planes, 4 fleets = 129% of 3 fleets.
JA5, Sanabas, 108 planes, 4 fleets = 121% of 3 fleets.
JA5, Sanabas, 121 planes, 4 fleets = 101% of 3 fleets

The three numbers before that were very different though, and fit well with the numbers from your MT5 airline.


JA5, Sanabas: 82 planes, 4 fleets = 213% of 3 fleets.

JA5, Sanabas, 87 planes, 4 fleets = 219% of 3 fleets.

JA5, Sanabas, 92 planes, 4 fleets = 225% of 3 fleets.



I would love to see the actual formula used in the code.

I certainly agree it needs to be fixed ASAP, and would be my #1 priority as well.

[Sami]

Can't be done to current game version as that would cause so large sudden changes, so is standing by for next larger update

[JumboShrimp]

Can't be done to current game version as that would cause so large sudden changes, so is standing by for next larger update

I kind of figured that a change like that, which could potentially leave 1/4 of the airline facing bankruptcy would probably need to be done in a new game world.

It seems to me that the current formula is something like this:

Cost of Aircraft 1 = Fixed Maintenance Cost + Training Cost

Cost of Fleet 1 = Administrative cost + Aircraft 1 + Aircraft 2 + Aircraft n

Total Cost = (Fleet 1 + Fleet 2 + Fleet n) * Fleet Commonality Multiplier Function

This Fleet Commonality Multiplier Function is a function of both:
- number of fleets
- number of aircraft

It may be ok for the Fleet Commonality Multiplier to be a function of number of fleets (reflecting need for cross training of crew - ground maintenace crew and flight crew), but it should be gradual, without this huge 3->4 fleet jump, and all the others being very small.  Each additional fleet should increase the cost by the same proportion.

As far as the this Fleet Commonality Multiplier also being a function (exponential at that) of numer of aircraft, it is just working the opposite way of what it should.  If anything, there should be a discount.

So the simplest change would be to make the fleet component a gradually rising multiplier, and removing the number of aircraft component completely.

In my opinion, to implement it correctly the fleet commonality should be implemented within individual fleets.  the Administrative Cost per fleet should be substantially higher, and there should be a discount for number of aircraft within fleet.  And each fleet cost would be multiplied by the Fleet Count multiplier.

If all of this is done, the tradeoff would be straight forward.  Player can opt for slower growth (limited by aircraft delivery rate) with small number of fleets, and be rewarded by greater profitability (due to lower costs).  Alternatively, player can opt for faster growth, with more fleets, but at much lower profitability (higher costs).  Right now, there is no such tradeoff early in the game.

[jkotinis]

just to add some new numbers . I suppose this is the same problem reported here

few days back I did the test with a total number of planes <300
3 fleets commonality costs total : 2831188
4 fleets commonality costs total : 7438251 ( by just adding one plane from the 4th fleet)
so 4 fleets = 262% of 3 fleets

Yesterday I did the same exercise ( I just have more planes now >300)
total costs with  3 fleets   : 3.487.496
total costs with  4 fleets   : 12.313.421  (by just adding the same exact plane like before)
so 4 fleets = 353 % of 3 fleets

as it is now  it seems that each time the number of planes pass the 00 mark  (100, 200, 300 etc)  you just get an extra penalty on top of the previous one.

[Jona L.]

just to add some new numbers . I suppose this is the same problem reported here

few days back I did the test with a total number of planes <300
3 fleets commonality costs total : 2831188
4 fleets commonality costs total : 7438251 ( by just adding one plane from the 4th fleet)
so 4 fleets = 262% of 3 fleets

Yesterday I did the same exercise ( I just have more planes now >300)
total costs with  3 fleets   : 3.487.496
total costs with  4 fleets   : 12.313.421  (by just adding the same exact plane like before)
so 4 fleets = 353 % of 3 fleets

as it is now  it seems that each time the number of planes pass the 00 mark  (100, 200, 300 etc)  you just get an extra penalty on top of the previous one.

Simple reason:

To prevent big airlines from growing bigger too soon, to give n00bs and smaller airlines a chance...

[jkotinis]

hi, agree on the reasoning, I don't complain on that

I just thought that this was a bug and it would be fixed, based on the comments by sami on the same thread.

And again, nobody stops a big airline to become bigger by just using 3 fleets I believe. If the question of bringing difficulties to big airlines to prevent them from becoming bigger, then the right approach would be to just increase the commonality costs just based on the number of aircrafts and not fleet types.

So if an airline has <100 then commonality costs = standard ones
from 100-200  30% increase
from 200-300  60% increase
from 300-400  100% increase

and from over 400 planes just put a 900% increase so nobody woule even think to have more than 400 planes
(could also put a variable based on the type of the planes, small, big, etc   ex  0,5 for small planes, 1 for medium and 1,5 for large so
 if x*0,5+y*1+z*1,5 > 400 then commonality costs = commonality costs * 9)

[Jona L.]

hi, agree on the reasoning, I don't complain on that

I just thought that this was a bug and it would be fixed, based on the comments by sami on the same thread.

And again, nobody stops a big airline to become bigger by just using 3 fleets I believe. If the question of bringing difficulties to big airlines to prevent them from becoming bigger, then the right approach would be to just increase the commonality costs just based on the number of aircrafts and not fleet types.

So if an airline has <100 then commonality costs = standard ones
from 100-200  30% increase
from 200-300  60% increase
from 300-400  100% increase

and from over 400 planes just put a 900% increase so nobody woule even think to have more than 400 planes
(could also put a variable based on the type of the planes, small, big, etc   ex  0,5 for small planes, 1 for medium and 1,5 for large so
 if x*0,5+y*1+z*1,5 > 400 then commonality costs = commonality costs * 9)

I disagree... The factor should be mainly on the number of fleet types. A slight increase per each 100 a/c, okay, but you cannot expect a limit to 400 planes per airline, if you don't allow subsidiary companies, which would reset the number to 0 for the new company.

[jkotinis]

I understand.

I just thought that the idea was to prevent big airlines become bigger. The issue is not solved anyway by penalizing the ones who use many fleet types.
Nobody can prevent me as it is right now to just keep growing on 3 fleet types.

[JumboShrimp]

Exploit is in the fact that a new (small) airline is immune to these fees.  This exploit gives a green light to new airlines to use 10 fleet types early in the game in order to:
- buy up slots
- crush the competitors
- place unlimited orders to 3 desired lines

This happens early in the game.  It puts airline following a sound strategy of 1-3 fleet types at a disadvantage.

[ArcherII]

I'm based at HKG in MT6, and for the entire game I've run 3 fleets. Just for the sake of it and to get some action till the end of the game, I decided to try the A380 and the A350 (those two for the first time for me).
Well, when running 3 fleets my commonality costs where around 90mil per month, which was good. I had 147 A330/340s, 160 757s and 118 777s. You would say that it's a pretty tight fleet.
When the first A380 was scheduled, it added so much commonality cost that virtually duplicated the expense. So the jump was from 3 fleets to 4. When the A350 came online, it added to the cost but at a same rate as if you were adding a 2nd fleet.
Now my commonality costs have risen to the double because of that. Nonsense.
I'd prefer a larger amount of personnel hired to overcome the new type and its particular commonality costs to be very high at the beginning and decreasing on a per plane basis. But NOT rise maintenance to the entire fleet.
Maybe that's why I run 3 types most of the time *sigh*.

[JumboShrimp]

The cost of 4th fleet goes up exponentially with total number of aircraft your airline has.  Since it is exponential, you don't have the same penalty for 4th fleet early in the game.  That's the exploit I am talking about above.

Joe

[ArcherII]

The cost of 4th fleet goes up exponentially with total number of aircraft your airline has.  Since it is exponential, you don't have the same penalty for 4th fleet early in the game.  That's the exploit I am talking about above.

Joe

I know. I'm just pointing out what you said previously in this thread, that this should be the other way around.



edit: grammar