Pre-YAGBU / Bug report: Appearance and shop prices
Erik Piper
SLIGHTLY
SPOILY
ARGUABLY
SLIGHTLY
SPOILY
ARGUABLY
SLIGHTLY
SPOILY
ARGUABLY
SLIGHTLY
SPOILY
ARGUABLY
SLIGHTLY
SPOILY
EXECUTIVE SUMMARY OF THE BELOW:
There seems to be a bug that makes shopkeepers' buying and selling
prices move in the same direction for female characters once
Appearance exceeds 20, instead of in opposite directions as one would
expect -- at least under some circumstances, which I perhaps only
happened to encounter. If the bug affects all female characters, then
it is wise to reduce Appearance to 20 when *selling* items. The bug
does not affect buying prices. No idea if a similar bug exists for
Charisma.
---
Fairly recently, I think, someone noted--maybe even Andy in the
Guidebook--that the effects of (females') high Appearance on shop
prices are unpredictable. I'd never wanted to believe it, but since
I've been in a swing of doing a bit of safe experimenting in this
game, I thought of this when I realized I'd forgotten to put on my
Pendant of Beauty. Well, I ran a short test, and as they say in the
Bible, my countenance fell:
Appearance 30, leather cap [0, +3] -- Leggot will buy it for 37
Appearance 25, same item -- Leggot will buy it for 44
Incidentally, I tried repeating and refusing the offer several times
(mostly to ensure myself that the behavior was stable and really was
the inverse of what I expected), and I don't seem to be seeing an
effect I've also read about somewhere: worsening of deals after the
player refuses an offer. Of course, this is something that one tends
to ad-hoc verify in MOST games (we've all offered something by mistake
and caught it at the last minute once in a while), and I never noticed
a worsening of deals in the past, either. I wish I could remember
where I read that rumor...
I decided to "research that," putting together the table you see
below. I was all ready to state that the AMW must be the ultimate gear
when selling off your stuff. But nope -- the oddity disappears as
Appearance drops below 20.
Here are some semi-random raw data for anyone who wants to research
further, AND FOR ALL YOU MATHEMATICIANS OUT THERE, HINT HINT. I
deliberately bought back some appearance-altering items I had sold in
the past in order to reduce gaps in the scale below, so you all must
now worship me and my dedication.
Character: level 22 human witch, alignment C-, charisma 21; version:
ADOM 1.1.1 winbeta 4; shop ever invisirobbed? - probably (err, that
was real helpful); shop at "no more haggling" point, haggling having
probably succeeded 50-60% of the time prior to reaching that point;
shop probably restocked once; Leggot: human; item considered for
purchase/sale: 1 aquamarine
Ap 31 - 379 sell, 1084 buy
Ap 30 - 392 sell, 1120 buy
Ap 29 - 405 sell, 1159 buy
Ap 28 - 418 sell, 1195 buy
Ap 27 - 431 sell, 1234 buy
Ap 26 - 444 sell, 1269 buy
Ap 25 - 457 sell, 1307 buy
Ap 24 - 470 sell, 1344 buy
Ap 23 - 483 sell, 1382 buy
Ap 22 - 496 sell, 1419 buy
Ap 21 - 509 sell, 1456 buy
Ap 20 - 522 sell, 1494 buy
Ap 19 - 520 sell, 1531 buy
Ap 18 - 511 sell, 1569 buy
Ap 17 - 502 sell, 1606 buy
Ap 16 - 493 sell, 1643 buy
Ap 15 - 485 sell, 1683 buy
Ap 14 - 476 sell, 1718 buy
Ap 13 - 467 sell, 1757 buy
Ap 12 - 458 sell, 1793 buy
Ap 11 - 449 sell, 1832 buy
Ap 10 - 441 sell, 1870 buy
Ap 9 - 431 sell, 1961 buy
Ap 8 - 422 sell, 2055 buy
Ap 7 - 413 sell, 2150 buy
Ap 6 - 404 sell, 2244 buy
Ap 5 - 396 sell, 2337 buy
Ap 4 - 387 sell, 2431 buy
Ap 3 - 378 sell, 2524 buy
Ap 2 - 369 sell, 2616 buy
Ap 1 - 360 sell, 2711 buy
I hope I'm not doing this unnecessarily. I searched for the text
"appearance" in the entire Guidebook and could find no entry on the
effect of Appearance on shop prices. I'm on dial-up at home (I usually
post from work), so I couldn't be bothered to search the newsgroup
before assembling the list. In any case I doubt I would have missed a
thread regarding a bug that, once pointed out, is so glaring.
AFTER I was halfway done with this testing, it occurred to me that a
pile of sis would allow me to meet both the goal of buying/selling the
same thing and the goal of having large numbers to work with. Oh well.
Erik
Andy Williams
> Fairly recently, I think, someone noted--maybe even Andy in the
> Guidebook--that the effects of (females') high Appearance on shop
> prices are unpredictable. I'd never wanted to believe it, but since
> I've been in a swing of doing a bit of safe experimenting in this
> game, I thought of this when I realized I'd forgotten to put on my
> Pendant of Beauty. Well, I ran a short test, and as they say in the
> Bible, my countenance fell:
My testing was limited to PCs with Ap levels 1, 50 and 99. This
probing was done several gammas ago. At that time, there seemed to be
a factor of about three, at most. I will reinvestigate this, but it
is not going to happen soon, since I am on the road.
--
Andy Williams
ADOM Guidebook: http://www.andywlms.com/adom/
Mirror: http://users.rcn.com/andy.williams/adom/
Laszlo
Very interesting. Analysis of data follows:
> Ap 31 - 379 sell, 1084 buy 36
> Ap 30 - 392 sell, 1120 buy 39
> Ap 29 - 405 sell, 1159 buy 36
> Ap 28 - 418 sell, 1195 buy 39
> Ap 27 - 431 sell, 1234 buy 35
> Ap 26 - 444 sell, 1269 buy 38
> Ap 25 - 457 sell, 1307 buy 37
> Ap 24 - 470 sell, 1344 buy 38
> Ap 23 - 483 sell, 1382 buy 37
> Ap 22 - 496 sell, 1419 buy 37
> Ap 21 - 509 sell, 1456 buy 38
>
> Ap 20 - 522 sell, 1494 buy 37
>
> Ap 19 - 520 sell, 1531 buy 38
> Ap 18 - 511 sell, 1569 buy 37
> Ap 17 - 502 sell, 1606 buy 37
> Ap 16 - 493 sell, 1643 buy 40
> Ap 15 - 485 sell, 1683 buy 35
> Ap 14 - 476 sell, 1718 buy 39
> Ap 13 - 467 sell, 1757 buy 36
> Ap 12 - 458 sell, 1793 buy 39
> Ap 11 - 449 sell, 1832 buy 38
> Ap 10 - 441 sell, 1870 buy 91
> Ap 9 - 431 sell, 1961 buy 94
> Ap 8 - 422 sell, 2055 buy 95
> Ap 7 - 413 sell, 2150 buy 94
> Ap 6 - 404 sell, 2244 buy 93
> Ap 5 - 396 sell, 2337 buy 94
> Ap 4 - 387 sell, 2431 buy 93
> Ap 3 - 378 sell, 2524 buy 92
> Ap 2 - 369 sell, 2616 buy 95
> Ap 1 - 360 sell, 2711 buy
Excellent work. What the data seems to suggest is that Appearance
modifies the _buy price_ of items downward. It also modifies the
sales price of items upward. HOWEVER, there is an overriding function
that says the sales price of items can't be less than a certain number,
based on the buy price. (This appears to be 35% of the buy price).
Let's try to set up an exact function (pseudocode):
Buy price: The item has a base price.
If Apprearance is 10 or less:
Buy_Price = ( (30-App)/30 ) * Base_Price
If Appearane is higher than 10:
Buy_Price = (2/3) * ( (85-App) / 75 ) * Base_Price
If this seems a little abstruse, here's what it means in English.
Each item has a theoretical price, that would be its Buy Price for
a PC with Appearance 0. This one's is 2805, by the way. Every point
of Appearance up to 10 knocks 1/30 off the price, and every point
of App above 10 knocks another 1/75 off.
This holds true until an App of 31, at least; I'm pretty sure
there's another break in the function somewhere, and since I'm
starting to understand TB's thinking, I'd bet that it's at App
35.
In other words, every point of App up to 10 counts for quite a lot,
but it becomes less and less important after that.
Now, Sales Price.
Max_Sales_Price = 0.35 * Buy_Price
Base_Sales_Price = Base_Buy_Price/8
Sales_Price = ( (40+App)/40 ) * Base_Sales_Price
If Sales_Price would be bigger than Max_Sales_Price
Then set Sales_Price to be equal to Max_Sales_Price
In other words, The theoretical Sales price at Appearance 0 would be
an eighth of the Buy price. Every point of Appearance adds another
1/40 of this base Buy price to the Buy price. However, the Buy
price can never go above 35% of the Sales Price. (This explains the
break at App 20).
My figures are not absoutely 100% exact, but they work _very_ well.
Any discrepancy is probably due to me and ADOM ounding differently.
You'll never find more than about a 2 GP difference.
> I hope I'm not doing this unnecessarily. I searched for the text
> "appearance" in the entire Guidebook and could find no entry on the
> effect of Appearance on shop prices. I'm on dial-up at home (I usually
> post from work), so I couldn't be bothered to search the newsgroup
> before assembling the list. In any case I doubt I would have missed a
> thread regarding a bug that, once pointed out, is so glaring.
>
> AFTER I was halfway done with this testing, it occurred to me that a
> pile of sis would allow me to meet both the goal of buying/selling the
> same thing and the goal of having large numbers to work with. Oh well.
If anyone gives me more data to work with, I'll gladly revise my
equations as needed. I still don't know whether Charisma works in
exactly the same way for males, for example, or what effect being a
merchant has on these functions.
> Erik
Again, great job!
Laszlo
Malte Helmert
[snip discussion of shop prices]
> If anyone gives me more data to work with, I'll gladly revise my
> equations as needed. I still don't know whether Charisma works in
> exactly the same way for males, for example, or what effect being a
> merchant has on these functions.
Very nice work, Erik and Laszlo! If you look into this, don't forget the
effect of the Haggling skill.
Malte
Erik Piper
> er...@sky.cz (Erik Piper) wrote in message news:<c605db35.04021...@posting.google.com>...
> > ARGUABLY
> > SLIGHTLY
> > SPOILY
> >
> > ARGUABLY
> > SLIGHTLY
> > SPOILY
> >
> > ARGUABLY
> > SLIGHTLY
> > SPOILY
> >
> > ARGUABLY
> > SLIGHTLY
> > SPOILY
> >
> > ARGUABLY
> > SLIGHTLY
> > SPOILY
> >
> >
and especially how they go haywire at the border between 11 and 10!
This might have been because I was focussing much more strongly on
Sell prices, and might have been the effect of losing my concentration
and doing something wrong, as I had already been working on the
problem for an hour or so at that point, but on the other hand, maybe
I transcribed them just fine, as you in essence state your model
reflects the data set. I went ahead and sold off the "extra"
appearance-modifying items I used for testing at Leggot's after
completing the experimentation, so I won't be able to double-check for
the moment (sorry! Didn't think of double-checking originally, and I
in any case I was bored out of my skull and itching to play by the
time I reached Ap 1 anyway) unless I can find some in Dwarftown.
However, I'm a bit pessimistic on this, as I generally avoid selling
stuff (except rocks) in Dwarftown due to the note in the Guidebook
that the shop prices (which presumably includes Sell prices) at
Leggot's are better.
If the group is willing to accept a bit of save-scumming in the name
of science, I could create a branch reality where this character heads
back to Leggot's and double-checks the Buy prices for Ap 12 and lower,
and perhaps more importantly, I can transcribe Sell prices for
multi-item stacks. For the moment, all of the conceivably important
parameters (the ones mentioned above) should be the same as during the
experiment.
Laszlo, I'd like to repeat the note I placed in a separate thread
yesterday -- the odd Sell Price behavior shift at Ap 20 only occurred
when selling stacks of 1 item. For stacks of 2 or more items, the Sell
Price continued to rise after Ap 20 as one would expect. For selling
via Extended Drop, the behavior varied stack-by-stack according to the
above. Thus by design or mistake, there seem to be different
algorithms applied to the Sell Price (only!) for 1-item stacks and
multi-item stacks.
By the way Laszlo -- all in all, I do assume that you really DO know
what you are saying below, which leads me to state -- GREAT quick
conversion of raw data to formulae. If that was all via blood sweat
and tears, I bow down to you. Even if it was with the help of a
mathematical modelling program, I bow down anyway.
Laszlo, assuming that I reperform the experiment and confirm the Buy
Price results I got at Ap 13 and lower, does the math below really
reflect that? Although I confess that I haven't verified your
algorithms yet, they seem too simple at first glance to reflect such
intense weirdness. Maybe I am merely a man of little faith.
> Excellent work. What the data seems to suggest is that Appearance
> modifies the _buy price_ of items downward. It also modifies the
> sales price of items upward. HOWEVER, there is an overriding function
> that says the sales price of items can't be less than a certain number,
> based on the buy price. (This appears to be 35% of the buy price).
>
And perhaps also that they can't be more than a certain number?
Perhaps Thomas tried to implement a sales-price ceiling with reduced
returns after a certain point, but accidentally created one producing
negative returns after Ap 20 instead? (And meanwhile no major loss in
returns for multi-item stacks?!?!)
> Let's try to set up an exact function (pseudocode):
>
> Buy price: The item has a base price.
>
> If Apprearance is 10 or less:
> Buy_Price = ( (30-App)/30 ) * Base_Price
> If Appearane is higher than 10:
> Buy_Price = (2/3) * ( (85-App) / 75 ) * Base_Price
>
> If this seems a little abstruse, here's what it means in English.
> Each item has a theoretical price, that would be its Buy Price for
> a PC with Appearance 0. This one's is 2805, by the way. Every point
> of Appearance up to 10 knocks 1/30 off the price, and every point
> of App above 10 knocks another 1/75 off.
>
> This holds true until an App of 31, at least; I'm pretty sure
> there's another break in the function somewhere, and since I'm
> starting to understand TB's thinking, I'd bet that it's at App
> 35.
>
You say you're starting to understand TB's thinking. Do you think he
really intended to produce negative returns for Sale prices around Ap
20?
To comment on myself: It really IS too bad I didn't try that from the
start, as I would have immediately uncovered the "aberration in the
aberration" regarding multi-item stacks.
>
> If anyone gives me more data to work with, I'll gladly revise my
> equations as needed. I still don't know whether Charisma works in
> exactly the same way for males, for example, or what effect being a
> merchant has on these functions.
Not to mention Haggling, restocks, and the "Shop ever invisirobbed?"
flag, if any. (Not sure why I think there might be such a flag, but I
do.) This whole subject really intrigues me, so perhaps the best would
be for me to simply get my hands on the newest testing-friendly
version -- either the pre-release 1.0.0 with the wizard mode that John
Rowat made famous, or 1.1.1 if somebody will kindly point me to the
wand of wishing trick. I can't promise to do exhaustive testing, but
I'd be happy to do at least "baseline" testing -- i.e. with a
character who has never haggled and never invisirobbed, has N=
alignment, and is visiting a never-restocked shop. However, I can
provide testing with my current character much faster than this
more-valid testing.
Laszlo
Assuming you consider that character (the one restored from backup, not
the current one) inelegible for a win, whatever happens, I don't think
anyone could possibly have any problem with it.
> Laszlo, I'd like to repeat the note I placed in a separate thread
> yesterday -- the odd Sell Price behavior shift at Ap 20 only occurred
> when selling stacks of 1 item. For stacks of 2 or more items, the Sell
> Price continued to rise after Ap 20 as one would expect. For selling
> via Extended Drop, the behavior varied stack-by-stack according to the
> above. Thus by design or mistake, there seem to be different
> algorithms applied to the Sell Price (only!) for 1-item stacks and
> multi-item stacks.
That's kinda strange. If you can gimme data, I'll see what comes o't.
It sounds a bit like a bug, actually, and it _might_ possibly have
contributed to the reason why one of my recent 1.1.1 merchants was
able to sell a 4-pack of iron rations to Munxip for about 100 more
than he could buy them (also obviously a bug).
> By the way Laszlo -- all in all, I do assume that you really DO know
> what you are saying below, which leads me to state -- GREAT quick
> conversion of raw data to formulae. If that was all via blood sweat
> and tears, I bow down to you. Even if it was with the help of a
> mathematical modelling program, I bow down anyway.
I'm a mathematician, it really wasn't difficult :) Since it was
obvious from a casual glance at the data that we were talking linear
functions, a mathematical modelling program wasn't needed. Just a
calculator :)
It would reflect it perfectly, except, as I said, for rounding errors.
Within 1-2 gold pieces.
> > Excellent work. What the data seems to suggest is that Appearance
> > modifies the _buy price_ of items downward. It also modifies the
> > sales price of items upward. HOWEVER, there is an overriding function
> > that says the sales price of items can't be less than a certain number,
> > based on the buy price. (This appears to be 35% of the buy price).
> >
> And perhaps also that they can't be more than a certain number?
> Perhaps Thomas tried to implement a sales-price ceiling with reduced
> returns after a certain point, but accidentally created one producing
> negative returns after Ap 20 instead? (And meanwhile no major loss in
> returns for multi-item stacks?!?!)
This seems unlikely. One can sell artifacts for far more than the
values in your table.
No, I don't think there's a bug... just a (probably) unintended side-
effect of some fairly reasonable control parameters.
> > Let's try to set up an exact function (pseudocode):
> >
> > Buy price: The item has a base price.
> >
> > If Apprearance is 10 or less:
> > Buy_Price = ( (30-App)/30 ) * Base_Price
> > If Appearane is higher than 10:
> > Buy_Price = (2/3) * ( (85-App) / 75 ) * Base_Price
> >
> > If this seems a little abstruse, here's what it means in English.
> > Each item has a theoretical price, that would be its Buy Price for
> > a PC with Appearance 0. This one's is 2805, by the way. Every point
> > of Appearance up to 10 knocks 1/30 off the price, and every point
> > of App above 10 knocks another 1/75 off.
> >
> > This holds true until an App of 31, at least; I'm pretty sure
> > there's another break in the function somewhere, and since I'm
> > starting to understand TB's thinking, I'd bet that it's at App
> > 35.
> >
> You say you're starting to understand TB's thinking. Do you think he
> really intended to produce negative returns for Sale prices around Ap
> 20?
He had three choices.
1) Have buy and sell prices be _very_ far apart for characters with
low Appearance scores.
2) Have Appearance only have a minor effect on buy and sell prices.
3) The current solution.
He went with number 3. If he hadn't put that 35% limit in there, the
sell and buy prices would have met at high Appearance scores, which
would have been highly abusable, of course.
I'm not saying it would be impossible to come up with a better system,
but this one is "close enough for government work", as they say. He
might have made a bit of a msitake with item stacks, though. We'll
see :)
Laszlo
Erik Piper
> er...@sky.cz (Erik Piper) wrote in message news:<c605db35.04021...@posting.google.com>...
> > laszlo_...@freemail.hu (Laszlo) wrote in message news:<36c239a1.04021...@posting.google.com>...
> > > er...@sky.cz (Erik Piper) wrote in message news:<c605db35.04021...@posting.google.com>...
[sorry for the lack of snippage in the below; I am in a hurry since I
am at work, and am only posting now because of the Google Groups
3-to-9-hour delay]
Oh, absolutely. Actually, at the moment I wrote that, I was
considering any trip to the east at this stage as an "anti-winning
move," although as luck would have it, I just lost my teleport control
intrinsic and my nearest blink dog corpse is at Leggot's, so I may
simply head back there in the main reality. I'm in the money so much
at this point that investing a couple thousand gold in buy/loss losses
on appearance-altering items again is no real loss.
> > Laszlo, I'd like to repeat the note I placed in a separate thread
> > yesterday -- the odd Sell Price behavior shift at Ap 20 only occurred
> > when selling stacks of 1 item. For stacks of 2 or more items, the Sell
> > Price continued to rise after Ap 20 as one would expect. For selling
> > via Extended Drop, the behavior varied stack-by-stack according to the
> > above. Thus by design or mistake, there seem to be different
> > algorithms applied to the Sell Price (only!) for 1-item stacks and
> > multi-item stacks.
>
> That's kinda strange. If you can gimme data, I'll see what comes o't.
Will do. (Note to self: be sure to also test the selling of a single
item out of a multi-item stack.)
I guess I expressed myself confusingly. What I mean is that they can't
be more than a certain number for a given item, i.e. for each "base
price" there is supposed to be (and is) a sell-price ceiling.
> No, I don't think there's a bug... just a (probably) unintended side-
> effect of some fairly reasonable control parameters.
>
Or as we call it in the software business, a "feature." In any case,
in (at least) ADOM 1.1.1 it seems to be unwise to have an appearance
over 20 when SELLING single-item stacks, with the unwiseness rising
the further one rises above 20, which makes this a definite and quite
non-trivial YAGBU. Whether there should be a bug report, and what
should be claimed is a bug, is another matter. Inconsistent behavior
for the two stack types is pretty obviously a bug. I think negative
returns on appearances above 20 is a bug, but I see the issue is
debatable, so I think I'll wait for more opinions to come in.
> > > Let's try to set up an exact function (pseudocode):
> > >
> > > Buy price: The item has a base price.
> > >
> > > If Apprearance is 10 or less:
> > > Buy_Price = ( (30-App)/30 ) * Base_Price
> > > If Appearane is higher than 10:
> > > Buy_Price = (2/3) * ( (85-App) / 75 ) * Base_Price
> > >
> > > If this seems a little abstruse, here's what it means in English.
> > > Each item has a theoretical price, that would be its Buy Price for
> > > a PC with Appearance 0. This one's is 2805, by the way. Every point
> > > of Appearance up to 10 knocks 1/30 off the price, and every point
> > > of App above 10 knocks another 1/75 off.
> > >
> > > This holds true until an App of 31, at least; I'm pretty sure
> > > there's another break in the function somewhere, and since I'm
> > > starting to understand TB's thinking, I'd bet that it's at App
> > > 35.
> > >
> > You say you're starting to understand TB's thinking. Do you think he
> > really intended to produce negative returns for Sale prices around Ap
> > 20?
>
> He had three choices.
>
> 1) Have buy and sell prices be _very_ far apart for characters with
> low Appearance scores.
>
> 2) Have Appearance only have a minor effect on buy and sell prices.
>
> 3) The current solution.
>
Couldn't he have created a solution where the Sell prices simply hit a
ceiling at a certain point and then bounce around it, as with the
floor for the Buy prices? If so, what would it look like (in
pseudocode)?
> He went with number 3. If he hadn't put that 35% limit in there, the
> sell and buy prices would have met at high Appearance scores, which
> would have been highly abusable, of course.
>
> I'm not saying it would be impossible to come up with a better system,
> but this one is "close enough for government work", as they say. He
> might have made a bit of a msitake with item stacks, though. We'll
> see :)
>
> Laszlo
By the way, where did you get your English? Nothing seems Hungarian
about your posts except your name and e-mail!
Laszlo
> laszlo_...@freemail.hu (Laszlo) wrote in message news:<36c239a1.0402...@posting.google.com>...
> > er...@sky.cz (Erik Piper) wrote in message news:<c605db35.04021...@posting.google.com>...
> > > laszlo_...@freemail.hu (Laszlo) wrote in message news:<36c239a1.04021...@posting.google.com>...
> > > > er...@sky.cz (Erik Piper) wrote in message news:<c605db35.04021...@posting.google.com>...
>
> [sorry for the lack of snippage in the below; I am in a hurry since I
> am at work, and am only posting now because of the Google Groups
> 3-to-9-hour delay]
>
> > > > > ARGUABLY
> > > > > SLIGHTLY
> > > > > SPOILY
> > > > >
> > > > > ARGUABLY
> > > > > SLIGHTLY
> > > > > SPOILY
> > > > >
> > > > > ARGUABLY
> > > > > SLIGHTLY
> > > > > SPOILY
> > > > >
> > > > > ARGUABLY
> > > > > SLIGHTLY
> > > > > SPOILY
> > > > >
> > > > > ARGUABLY
> > > > > SLIGHTLY
> > > > > SPOILY
<snip>
> > > > Excellent work. What the data seems to suggest is that Appearance
> > > > modifies the _buy price_ of items downward. It also modifies the
> > > > sales price of items upward. HOWEVER, there is an overriding function
> > > > that says the sales price of items can't be less than a certain number,
> > > > based on the buy price. (This appears to be 35% of the buy price).
> > > >
> > > And perhaps also that they can't be more than a certain number?
> > > Perhaps Thomas tried to implement a sales-price ceiling with reduced
> > > returns after a certain point, but accidentally created one producing
> > > negative returns after Ap 20 instead? (And meanwhile no major loss in
> > > returns for multi-item stacks?!?!)
> >
> > This seems unlikely. One can sell artifacts for far more than the
> > values in your table.
> >
> I guess I expressed myself confusingly. What I mean is that they can't
> be more than a certain number for a given item, i.e. for each "base
> price" there is supposed to be (and is) a sell-price ceiling.
Ah. Right. No, I don't think that's what TB wanted to do; if there was
a static limit, then as the buy price goes down at really high App
scores, there's a danger of the buy price dipping below the limit.
Of course, the problem _could_ have been solved without this strange
phenomenon where the sell prices go _down_ at really high App scores.
I don't think TB fully thought this through when he coded it, to be
frank. Looks like a "quick fix" to me.
> > No, I don't think there's a bug... just a (probably) unintended side-
> > effect of some fairly reasonable control parameters.
> >
> Or as we call it in the software business, a "feature." In any case,
> in (at least) ADOM 1.1.1 it seems to be unwise to have an appearance
> over 20 when SELLING single-item stacks, with the unwiseness rising
> the further one rises above 20, which makes this a definite and quite
> non-trivial YAGBU. Whether there should be a bug report, and what
> should be claimed is a bug, is another matter. Inconsistent behavior
> for the two stack types is pretty obviously a bug. I think negative
> returns on appearances above 20 is a bug, but I see the issue is
> debatable, so I think I'll wait for more opinions to come in.
It's certainly an "undesirable feature", IMO, and illogical to boot.
Worth an RFE, I think.
> > > You say you're starting to understand TB's thinking. Do you think he
> > > really intended to produce negative returns for Sale prices around Ap
> > > 20?
> >
> > He had three choices.
> >
> > 1) Have buy and sell prices be _very_ far apart for characters with
> > low Appearance scores.
> >
> > 2) Have Appearance only have a minor effect on buy and sell prices.
> >
> > 3) The current solution.
>
> Couldn't he have created a solution where the Sell prices simply hit a
> ceiling at a certain point and then bounce around it, as with the
> floor for the Buy prices? If so, what would it look like (in
> pseudocode)?
He could have; I'll write pseudocode for it later (very tired right
now). It would have taken a lot of work to balance properly, more than
the current solution. TB probably just wanted a quick fix to the bug
that Merchants could sometimes sell items for more than they could buy
them, and this is what he implemented as a quick fix.
> > He went with number 3. If he hadn't put that 35% limit in there, the
> > sell and buy prices would have met at high Appearance scores, which
> > would have been highly abusable, of course.
> >
> > I'm not saying it would be impossible to come up with a better system,
> > but this one is "close enough for government work", as they say. He
> > might have made a bit of a msitake with item stacks, though. We'll
> > see :)
> >
> > Laszlo
>
> By the way, where did you get your English? Nothing seems Hungarian
> about your posts except your name and e-mail!
Kind of you to say so :) I spent four years abroad as a kid.
Laszlo