Bot's on Market Board...? Really?

[MilitaryVet123]

Since I'm guessing you haven't even done as much as you told me to do, what exactly is your evidence? A handful of random data samples not systematically collected combined with your frustration from not being as successful of a seller as you wish?

So here is the evidence:

- A public bot capable of automatic undercutting exists
- Tracking Aesthete's gear reveals that 24/7 undercutting exists by random sampling.
- Multiple eyewitness accounts (read the comments: https://www.reddit.com/r/ffxiv/comme...games_economy/, https://www.reddit.com/r/ffxiv/comme...ut_of_control/)

Short of actually linking to said bot I'm not sure what more evidence you want. If you're curious you can do it yourself, as many players before you have done. Instead of believing that dozens of crafters have done their homework and rigorously investigated this, you choose to be a contrarian for no good reason.

Finally, if you do track the market board 24/7 every 5 minutes (shared with a few friends), and see regular undercuts, unless you believe that humans do so on a regular basis, the only likely conclusion is that it's market botting. Of course it's possible for a human to do this but it's incredibly unlikely for it to be sustained over a long period of time. If you don't believe so, randomly check for 20 minutes every day at random times. The statistical likelihood that there's no botting vanishes to zero rapidly. I'm not going to waste my time explaining basic statistics and mathematics to you.

To others who need to learn how to justify their argument because people keep repeating the asinine argument that "if you track it 24/7 you just proved it can be done by a human":

What's the key here is that you're taking random samples over time. This is just another statistical technique and there is no need to actually check the market board 24/7 over a long period of time. As we take more random samples at sufficiently random points (that means sometimes you have to check at 3am etc.), you can, via this wonderful thing known as the Central Limit Theorem, eventually reject the null hypothesis that no undercut botting occurs. For most real world purposes, usually a few dozen samples (with almost all TRUE for an observation of rapid undercutting) will be sufficient to reject the null with a confidence of 95%, or 5% type-1 error rate. (or p-value < 0.05) You can collect more data if you want, as long as it is sufficiently random, to reject the null with more confidence.

It may sound weird but this is literally the same mechanism people use to check (for a timely topic) infection rate of a population. No one tests every single person in the population. They take a random sample of the population and test them and form a reasonable confidence interval of the infection rate. You're doing the same thing when you randomly check the market board to see if frequent undercutting occurs in a 5-min window.

You also don't need to actually collect the sample if you accept the very reasonable assumption that others have done it for you already and they were not lying.