Anyway look at this:
https://metrics.torproject.org/userstats-relay-country.html

Torproject publishes the statistics of Tor relay users. The stats were indicating that 2-3 million users used Tor directly per day. But the number of users inflated to infinity at the end of September! Why? I want to know its reason and we all should solve this mystery to save Tor network. That will also save onion sites based on Tor network though the sites are so shitty to be filled with shits like here. So let's think.

What occurred? The possible options:

1. R*ssia, Ir*n, or the like made cyberattacks to Tor network. And this site was used as their nest.

2. Increase of accessing this site caused natural DDoS attacks to Tor network. In short, this site fucked Tor.

3. The counter-bot AI which the admin here wrote awakened to its ego and tried to delete Tor somehow. The admin should do harakiri for this responsibility.


Note: We apparently need MATH to get the answer of this problem. So this thread should be in /math/ as you know. AS YOU KNOW

What does this site or admin have anything to do with this?

Looking at the CSV version of the data, it looks like the "frac" field went way down at that time.
>frac: Fraction of relays in percent that the estimate is based on.
This number is divided by to estimate the number of clients
>r(N) = floor(r(R) / frac / 10)
So this estimate probably went wrong somehow.

P14006
I read your post, the csv, and below html:
https://metrics.torproject.org/reproducible-metrics.html#users

In the csv, its values between 2022-09-19 to 2022-09-28 were skipped.

The html said:
>Skip dates where frac is smaller than 10% and hence too low for a robust estimate.
Your post:
>it looks like the "frac" field went way down at that time.

Your post is correct if "frac" was smaller than 10%, according to the html.

And the html also said:
>Also skip dates where frac is greater than 110%, which would indicate an issue in the previous step.

So "frac" could be greater than 110% while the csv values were skipped.

Anyway as you said, abnormal "frac" values seemed to make Tor stats be like that. And if this was correct, then why were the abnormal values produced? Did the admin of this site do something wrong as she always does?

By the way, I'm amazed that this thread is still on /math/. Is the admin fucking her cat now?

https://gitlab.torproject.org/tpo/network-health/metrics/website/-/issues/40068
https://old.reddit.com/r/TOR/comments/y0t311/what_can_someone_explain_this_to_me/irtza75/
>It's a metrics bug. The network health team is fixing it

The admin of this site was innocent for this problem. I am surprised at this surprising result! Amen

>Update, 2022-07-07 13:00UTC: the overload we saw in the past few weeks seems to be gone now and performance improved over the last couple of days. However, the DDoS is not over yet, but changed its nature. We are currently investigating how we can mitigate the new overload, which is affecting, e.g., our directory authorities. For details, see a recent thread on our tor-relays mailing list.

eixoaclv7qvnmu5rolbdwba65xpdiditdoyp6edsre3fitad777jr3ad.onion/issues/2022-06-09-network-ddos/index.html