# Locating a deleted answer to my question

2 days ago I posted this question:

Suppose we have

$$F(z)=f(\phi(z))$$ where $$\phi$$ is a mobius transformation which maps points of the unit circle to points of the unit circle. Suppose also that there is an interval of length $$\pi$$ such that on that interval $$f(e^{it})-f(e^{-it})>0$$. How to prove then this is also true for $$F$$?

Here, I defined:

$$\phi(z):=\frac{z_0-z}{1-\bar{z_0}z}$$ with $$z_0$$ a fixed point of the unit disk (without the boundary) and $$f$$ a real harmonic continuous function defined on the unit disk. And $$F(z)=f(\phi(z))$$.

I had a good answer, but now it has disappeared and I really need it. The person that answered (and then I think I said: "a really big thank you!"), can you repeat your answer please? I would be very thankful

• Likely that answer was deleted because the ChatGPT detector claims it was AI/GPT generated (85%). Apr 10 at 18:38
• @BillDubuque Do we have a bot that automatically deletes ChatGPT posts? I had no idea that AI-generated content was such a big problem! I'm astonished that people are motivated to spam this site with AI-generated content. Apr 10 at 18:43
• Oh, can ChatGPT give answers? Apr 10 at 19:00
• Is there a possibility to show me the answer 1 time (after that you can delete it again but I need it...) Apr 10 at 19:01
• @BillDubuque Would you know whether any GPT detector is being run automatically on MSE, or otherwise whether a high score (say $\ge$ 90%) on a 3rd party detector is accepted as a legitimate reason to raise a mod flag?
– dxiv
Apr 11 at 2:15
• @anonymous123 You can try ChatGPT yourself. Ask it any question and it will give you an answer. Whether it's a good answer, however, is a whole different story. Worst of all in math matters.
– dxiv
Apr 11 at 2:20
• You can ask chatgpt the negation of your statement and they will still give you an answer. Apr 11 at 2:47
• It's worth noting the formatting of that post is pretty weird too May 20 at 7:59

I read through this deleted answer, and I must say I really do think it was generated by ChatGPT.

Take this passage, for example.

... if a Möbius transformation maps a circle or a line to another circle or line, then the transformation will preserve distances between points on those circles or lines. This is because circles and lines in the complex plane can be thought of as "geodesics," or paths that follow the shortest distance between two points.

This is plainly wrong. And yet, it is stated with confidence and assertiveness.

Now look at this calculation.

\begin{align*} \operatorname{Re}\left(\frac{w_2}{w_1}\right) &= \operatorname{Re}\left(\frac{\phi(z_2)}{\phi(z_1)}\right) \ \\ &= \operatorname{Re}(\phi(z_2)\cdot \overline{\phi(z_1)}) \ \\ &= \operatorname{Re}(f(\phi(z_2))\cdot \overline{f(\phi(z_1))}) \ \\ &= \operatorname{Re}(F(z_2)\cdot \overline{F(z_1)}) \ \\ &= \operatorname{Re}(f(\phi(z_2))\cdot \overline{f(\phi(z_1))}) \ \\ &= \operatorname{Re}\left(\frac{f(\phi(z_2))}{f(\phi(z_1))}\right)\cdot |f(\phi(z_1))|^2 \ \\ &> 0 \end{align*}

It's interesting how the expression $$\operatorname{Re}(f(\phi(z_2))\cdot \overline{f(\phi(z_1))})$$ appears on line 3, then again on line 5. It's as if the calculation is doubling back on itself, rather than aiming towards a purpose.

Overall, I couldn't make head or tail of this answer. I think it is a hoax.

• Your analysis is helpful. I'm agnostic about its connection to ChatGPT, but I did look at the history of the Answer, and it appears that the initial post was MathJax-free, something I associate with the swarm of ChatGPT posts network-wide. The edit, however misguided, contains math notation. A better approach would of course be to give a counterexample to the original problem, and then explore patching up the claims. Apr 11 at 13:33

I presume the question was this one. The answer that disappeared was deleted by a moderator. I don't know the reason why it was deleted, and I'm not prepared to undelete it myself.

Underneath your question you'll see a link labelled "flag". You can use this to flag the question as needing attention from a moderator by choosing the last of the options listed ("In need of moderator intervention"). If you provide a reasonable explanation of why you would like the answer to be undeleted in the window provided, a moderator might be prepared to undelete it, or at least provide an explanation of why it's not possible to fulfil your request.

• Thank you, I really hope I will get it back. Apr 9 at 19:28
• FYI, regarding not being prepared to "undelete it myself", note the "How can I undelete one of my posts?" section of this answer to the network Meta FAQ post How does deleting work? What can cause a post to be deleted, and what does that actually mean? What are the criteria for deletion?, states if the post was deleted by "a moderator: you cannot undelete it, and high-reputation users cannot vote to undelete; the only ones who can undelete them are moderators". Apr 10 at 14:57
• Thank you for the information. Apr 10 at 15:08
• Is there a possibility it will be undeleted? Apr 10 at 18:59
• @anonymous123 Not unless you can provide a more satisfactory justification for that to be done. I had not tried reading through the deleted answer until I saw Kenny Wong's answer to this question. Having now read through it, I agree with Kenny Wong that it's nonsense. It might not be exactly a hoax, but whatever it's purpose, it's clearly not in line with that for which Mathematics Stack Exchange was set up. Apr 11 at 1:40
• Oh, it was deleted because it was not a good answer? Really sad, but thanks Apr 12 at 6:58
• It's a little more serious than merely not being a good answer. From the nature of its errors, it seems likely that it wasn't even a good-faith attempt at an answer. Apr 12 at 8:50
• @anoniem Only very low quality answers get deleted. This one is low quality on grounds on being (suspected) non-human generated May 20 at 8:02