Slipperyjim’s test has enough data to be interesting, but there are too many confounding factors and missing elements to the study to draw strong conclusions. Here are some ways it could be improved to give us much stronger evidence and rule out other causes. Do you want to poe buy items? If you do, you can visit U4GM to have more guides and methods about this.
1. Run all the maps white. Alch-and-go introduces a lot of variance in the test. If they aren’t run white, we need to know the iir, iiq, and packsize of every map so we have a chance to correct for those factors. Additionally, it’s important to record any mods that affect pack composition or externally effect packsize like nemesis, beyond, %more magic, %more rare.
2. No prophesies. I think he controlled for this.
3. Pick up ALL currency, including wisdoms and portals. They give a lot of data about the overall amount of currency dropping. It’s possible that IIQ isn’t working at all or isn’t working on white mobs and all we’re observing is RNG in which particular orbs drop.
4. Pick up ALL divination cards. They’re currency too.
5. Skip all breaches, abysses, and boxes. They may have different drop rules and can contaminate the examples. From the spreadsheet view, it looks like he may have encountered a lot more breaches in the IIR runs and breaches drop a ton of currency. Alternatively, record how many breaches and how many abysses were opened on every map.
6. Also test with 0 IIQ and 0 IIR. As mentioned before, it’s possible that IIQ simply isn’t working or isn’t working on white mobs. It’s also possible that IIR has been changed to give hidden quantity to currency drops (say at a 1/3 rate or so). This could easily be tested if we had a 0/0 baseline.
7. If vendoring items, don’t include the orbs you get from the vendor in the sample. I’m guessing he did this correctly, but am not sure. IIR should give a lot more uniques and rares that will vendor for alchs and that’ll contaminate the sample.
Also, if you’re doing alch-and-go, you need to record a bunch of mods that can have a big hidden effect on pack size and composition. Need to pick up every div card too. Skip boxes.
26 Apr 2018