Hunters & Looting

[Eonwind]

the only thing different from before and after is randomness, and always remember random numbers are pseudorandom numbers http://en.wikipedia.org/wiki/Pseudo-random_number

the only thing that made difference between us two Is that I'm player and I'm still playing this game a lot (maybe too much)...
And if You would play a bit more... you may have noticed the  big difference Eonwind...
I know you did a lot for PS and I would add that Its been a long time since we didnt have such an efficient team...
But don't get disconected with the reality, you know its what IRL we are reproaching to our gouvernement...

Ok Mouli, let's suppose you're right and I am wrong... now try to get in a dev's shoes and suppose you want to debug the problem to solve the issue, if any. Let's start from this facts:

• there is no runtime instrument any devs/GM system admin can use to affect player's good or bad luck, so this cannot be the cause of the problem.
• the ramdomizer code hasn't changed since 1 and 1/2 year and then it changed to fix a bug with some rare modifiers that were never loot-able because of a code bug, so regarding these changes they can only have improve looting. other than that if the problem appeared recently it cannot be related.
• the loot rules table has not changed regarding items and their probabilities, only new stuff was added for a recently introduced monster, plus 1 year ago dropped bows have been made randomizeable (means now you their enchanted version can be dropped while before only normal version were dropped).
• loot rule for monsters haven't changed except a few named NPCs at the camp banished (so you're either trying with them or it's not the problem and other than that the changed loot rule should be richer that the previous one).
• mathscripts regarding player level calculations (used to determine automatically XP, random-levels, etc.) haven't been touched at all.

Now considering a software and data don't change by themselves, tell me where you would logically start debugging the problem, because with the informations I have at hands I have really no idea where to start looking at.

[bilbous]

compiler optimizations? (slashdot) has the server been compiled for a different architecture? has the compiler stack been updated?

[Bonifarzia]

• mathscripts regarding player level calculations (used to determine automatically XP, random-levels, etc.) haven't been touched at all.

Oww, interesting... random-levels, etc? So the loot probabilities are adjusted in a similar manner as experience points? You don't have to answer if you don't want to

On topic: I was never really ambitious with looting, but my feeling is that the "high" loot rates that drive this discussion were observed during a time longer ago than the changes mentioned above. Or am I wrong? I dont see much of a change in the past 18 months or so.

[Mouli]

@Eonwind, I guess in theory you could use the black box approach which we were given to reproduce and compare the current and an year old PS to either support or prove our claims wrong. That, including OS, hardware, database engine, datasets, PS version and whatnot that has anything to do with random and floating point numbers.  But that is an impossible mission if you ask me. I’d rather go with rebalancing the system for a desired target.

@Boni, You're right the change are not so new... I dont really remember but it was almost suddently... like ON/OFF.

[Eonwind]

compiler optimizations? (slashdot) has the server been compiled for a different architecture? has the compiler stack been updated?

mhm... I'm not really the best person to answer this question... I can only say the server is compiled using gcc on a linux architecture.

• mathscripts regarding player level calculations (used to determine automatically XP, random-levels, etc.) haven't been touched at all.

Oww, interesting... random-levels, etc? So the loot probabilities are adjusted in a similar manner as experience points? You don't have to answer if you don't want to

I've no problem speaking about this stuff, also because you should find a working example in SVN. Of course I can't share the script itself but I can say there is a mathscript "LootModifierCostCap" (should be called by the code handling looting and random item with modifiers) responsible of calculating a character stats based cap cost for looted weapons. I guess the reason for its existence is limiting the probabilities for a newly created characters to loot the best weapon in game. This has never been changed as far as I can tell.

On topic: I was never really ambitious with looting, but my feeling is that the "high" loot rates that drive this discussion were observed during a time longer ago than the changes mentioned above. Or am I wrong? I dont see much of a change in the past 18 months or so.

If the problem has to do especially with weapons with modifiers and the changes are not very recent (older than 18 months) then there could be something wrong going on in the code. It's not very easy for me to debug it.

[bilbous]

Weird. Did a search on "LootModifierCostCap" Found a link containing a listing of math_scripts.sql from what appears to be an older version of the server code.

Code: [Select]

INSERT INTO math_scripts VALUES( "LootModifierCostCap", "ModCap = MaxHP*10;");
Want the best loot better train up your body development!

[Eonwind]

Weird. Did a search on "LootModifierCostCap" Found a link containing a listing of math_scripts.sql from what appears to be an older version of the server code.

Code: [Select]

INSERT INTO math_scripts VALUES( "LootModifierCostCap", "ModCap = MaxHP*10;");
Want the best loot better train up your body development!

Nope
you effectively found the test mathscript in SVN, its purpose is only to provide a working example for anyone willing to compile and run the server code, the version loaded in production is very different.

[Bonifarzia]

Thanks for your clarification, Eonwind. So that cap is only to limit the availability of the more expensive affixes to characters with low stats, I see.

I wonder, when large versions of beasties are going to be implemented, will the engine allow to bump up the chances for receiving rare affixes? I think that could have great potential to keep the veterans like Mouli busy

Hope to finally see some big bosses to fight!

[LigH]

Don't be too optimistic. Not everyone should own an "almost unique" item.

[Eonwind]

Thanks for your clarification, Eonwind. So that cap is only to limit the availability of the more expensive affixes to characters with low stats, I see.

I wonder, when large versions of beasties are going to be implemented, will the engine allow to bump up the chances for receiving rare affixes? I think that could have great potential to keep the veterans like Mouli busy

Hope to finally see some big bosses to fight!

mhm... hard to tell ... you know to answer this question I not only have to know that mathscript function very well but also gain a deeper knowlegde of that specific part of the code internals (like I had to do for fixing skill progression and practice point assignments, make tribes ... ...). May it's worth the time to take a look but it will take some time ... just to answer 
(but sometime once the internals have been understood improvements or fixes have been straightforward like for fixing the armor damage reduction bug)

[Bonifarzia]

Heheh, I dont want to bother you with silly questions, Eonwind, I was just wondering about the potential of Talad's ambitious announcement. Either way, it is great to see the project is making good progress (where fixing old issues is even more valuable than adding new content). Thanks again for all the replies.

[Rigwyn]

Might it be possible to see if there has been a change in what people have been looting over the period of time in question? I don't know if you keep this sort of data on the server or not. A quick sanity check might be done by checking a broad metric like the total value of items looted or something like that. Such data might reveal a flaw in the code/hardware/data or in the perception of players.

[Eonwind]

Might it be possible to see if there has been a change in what people have been looting over the period of time in question? I don't know if you keep this sort of data on the server or not. A quick sanity check might be done by checking a broad metric like the total value of items looted or something like that. Such data might reveal a flaw in the code/hardware/data or in the perception of players.

It's a good idea! I checked and it's possible but since the server change these logs seems to be unavailable yet. 

[tman]

That metric would have to be normalized though.  Obviously the total value of looted goods should be higher back when there were hundreds of players than this fall when there were ~40 on average.

[Denes]

How about when we recently (less than a year) started looting two hearts or two potions from the same beastie by chance? Could it be that those have something to do with each other?