Looking for goblin archeologist in Permafrost

[cd288]

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

So what do you think is an accurate sample, what's an acceptable margin of error and do you think anyone in their sane mind would ever kill that many?

I guess it depends on what the original alleged % is, but in general I think it's tough to call any personal experience in EQ "accurate" regardless of how many mobs or cycles you've gone through.

Anyone who has played this game for a long time knows that whatever your experience will be with a camp, mob, etc. is impossible to predict since the RNG can be simply insane. Use the goblin ghazi ring as an easy example: I've had an experience where I killed like 20 PHs, had 4 couriers spawn, and one of them had the ring. I've also had another experience where I killed a few hundred PHs, probably saw 10 couriers, and none had the ring. Because of RNG every anecdotal experience is of very limited value when assessing percentages.

There was another post a few months ago where there was a great, detailed discussion on this. I'll see if I can find it.

[Goregasmic]

Story time: before I did the sarnak courier last year I tallied all the player submitted results on its page and ended up with an average of ~8 hours and 24 couriers for 1 ring. Took me 8.5h and 26 couriers heh, pretty close. People claim it is a 1% deal, seems more like 3% (but I didn't change it!). Granted, not many pages have that many player entries, if any.

A 100 coin flip will have about 4% variance either way, in the worst case. Most mobs are not 50% but more like 20-25% so more room for mistakes but I don't think anyone would sweat a 5% off spawn percentage, still gives you a good idea. Most camps I've done 40-80 mobs that I kept track of were within 3% of the wiki number so it was interesting to see the wiki being mostly accurate on that front. Of course, mobs with <10% spawn rate will require a lot more data to make an accurate statement.

I respect the wiki enough not to change anything willy nilly but that being said, a lot of the damage ranges are off, same for most AC values from what I understand. I've seen some mob HPs off and a lot of bogus loot tables so I don't view the wiki as a temple of accuracy that must be preserved in its pristine form either.

[loramin]

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

So what do you think is an accurate sample, what's an acceptable margin of error and do you think anyone in their sane mind would ever kill that many?

I'm not a stats geek, but as I understand it, the rule of thumb for a die is to have at least five times as many rolls as the die has sides. In other words, if you wanted to check that a six-sided die rolled "6" one out of six times, you'd need to roll it at least thirty times (and see five "6" results).

If your theory is that A Sarnak Courier drops the ring 2% of the time, that equates to 1 out of 50 rolls of a fifty-sided die. Thus, you'd need to do about 250 (50 * 5) rolls (ie. 250 Courier kills), and see five rings drop, to be reasonably certain that the drop rate was 2%.

Obviously, no one is going to do 250 Courier kills, because no one needs five goblin rings. For cases like that, I think the best anyone can do is start a list on the wiki page, saying something like "Bob: 47 Courier kills until a ring dropped". Then Fred can add "Fred: 55 kills", and eventually, over time, the wiki will collect enough cases (250 for the courier) to be confident in changing a number.

[maddysinko]

It can definitely be frustrating when the spawn rate feels super low, but 9 cycles isn’t too bad when it comes to these kinds of drop chances. I haven’t heard of any bugs with the spawn, so it's likely just bad luck.

[Goregasmic]

Originally Posted by loramin [You must be logged in to view images. Log in or Register.]

I'm not a stats geek, but as I understand it, the rule of thumb for a die is to have at least five times as many rolls as the die has sides. In other words, if you wanted to check that a six-sided die rolled "6" one out of six times, you'd need to roll it at least thirty times (and see five "6" results).

If your theory is that A Sarnak Courier drops the ring 2% of the time, that equates to 1 out of 50 rolls of a fifty-sided die. Thus, you'd need to do about 250 (50 * 5) rolls (ie. 250 Courier kills), and see five rings drop, to be reasonably certain that the drop rate was 2%.

Obviously, no one is going to do 250 Courier kills, because no one needs five goblin rings. For cases like that, I think the best anyone can do is start a list on the wiki page, saying something like "Bob: 47 Courier kills until a ring dropped". Then Fred can add "Fred: 55 kills", and eventually, over time, the wiki will collect enough cases (250 for the courier) to be confident in changing a number.

I was wondering if there was something like that, interesting read thank you!

Yeah I don't think most mobs are interesting enough to be worthy of such a list but there should be more. Real quick courier is the only one I can think of but I would have thought people doing pained soul/rotting skeleton or other such mobs would have kept a log.

Originally Posted by maddysinko [You must be logged in to view images. Log in or Register.]

It can definitely be frustrating when the spawn rate feels super low, but 9 cycles isn’t too bad when it comes to these kinds of drop chances. I haven’t heard of any bugs with the spawn, so it's likely just bad luck.

Yeah if 20% is accurate you have a 13% chance to not see it for 9 spawns, not incredibly wild.

[cd288]

What's the 13% from?

[loramin]

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

Yeah I don't think most mobs are interesting enough to be worthy of such a list but there should be more. Real quick courier is the only one I can think of but I would have thought people doing pained soul/rotting skeleton or other such mobs would have kept a log.

I just had to redo this page (someone deleted my previous report), but it gave me a chance to remake it as an example of how I think people should do "spawn reports" on the wiki:

https://wiki.project1999.com/Driftwood_Treasure_Chest

P.S. And yes, you read that right: I killed that stupid fish 700 times; I have the chests to prove it [You must be logged in to view images. Log in or Register.]

[Goregasmic]

Originally Posted by cd288 [You must be logged in to view images. Log in or Register.]

What's the 13% from?

0.8^9, which would be the probability of not getting one 9 times in a row at 20% spawn rate.

Originally Posted by loramin [You must be logged in to view images. Log in or Register.]

I just had to redo this page (someone deleted my previous report), but it gave me a chance to remake it as an example of how I think people should do "spawn reports" on the wiki:

https://wiki.project1999.com/Driftwood_Treasure_Chest

P.S. And yes, you read that right: I killed that stupid fish 700 times; I have the chests to prove it [You must be logged in to view images. Log in or Register.]

Dayum that's some dedication, even afk farming that's a long time and still semi-involved as that shark can chew you up and he's not alone in the room. Your driftwood sample section is nice but on mobile it is a bit messed up due to the auction tracker taking so much real estate.

I looked it up again and the page with the courier data is actually on the Goblin Gazughi Ring page. That one is also clean, i should crunch the numbers again.

The cleanest example I've seen (could be a bit more minimalist) is the tables Frantz256255 did at the bottom of some mob pages. If everyone did a table it would be chaos but as a tally it pretty complete. A nice table for stuff like the goblin ring submissions would probably be the cleanest way to implement this. Might work on it next time I'm bored at work.

I think a forum admin worked a script to scrape logs for all the drop data as a project for the wiki but not sure what happened to that.

[WarpathEQ]

Quite frankly spawn percentage is about as useless of a metric to the person trying to consume it as it is to the person trying to determine it so. Its an RNG game so true results will vary massively for the same camp, exponentially so for rarer spawns.

IMO the only real valuable data around spawn frequency would be people recording the time it took between named spawns without leaving the camp and killing the PH consistently on spawn without gaps. Being able to triangulate a rough average of how many hours a camp takes for named to spawn would be actually useful.

[loramin]

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

Your driftwood sample section is nice but on mobile it is a bit messed up due to the auction tracker taking so much real estate.

Unfortunately, that section is inside a wiki template, and a table would require using another template ... but our wiki doesn't let you have a template inside a template (in this case, a table inside an "Itembox" template).

Also, different screen sizes can position things differently (eg. on my phone everything looks fine). However, I changed the styling and text a bit, so it should look better now.

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

The cleanest example I've seen (could be a bit more minimalist) is the tables Frantz256255 did at the bottom of some mob pages. If everyone did a table it would be chaos but as a tally it pretty complete. A nice table for stuff like the goblin ring submissions would probably be the cleanest way to implement this.

Yeah, he was able to use a table because he put it outside of the template, at the bottom of the page. The Driftwood Chest page (or any other) could be redone that way, but it might not look as good at the bottom (and people might miss it).

In theory, I think the best solution would be to add a bunch of new fields to the "Itembox" template, and then if those fields are present have the template add a table itself. It probably wouldn't be that hard (the existing template is very simple) ... but I have other wiki projects at the moment, so I'll leave that as a project for someone else.

Originally Posted by Goregasmic [You must be logged in to view images. Log in or Register.]

I think a forum admin worked a script to scrape logs for all the drop data as a project for the wiki but not sure what happened to that.

I thought I heard about a log parser awhile back, but it was just from a regular player, for personal log parsing. I hadn't heard of any effort to collect people's data for the wiki (although that'd be cool!)