Should we try to re-start manual deletions or is the situation fine anyway?

Question

On MO 1.0 we had, after some discussions and conflicts but still, what I think was a quite smoothly and efficiently working process of deleting "bad" content. It seems with the move to MO 2.0 this process of manual deletions came more-or-less to a halt.

Two main reasons for this could be:

• There is now an auto-deletion script so there is less need to do so⁽¹⁾.

• The deletion interface changed and makes it easy to overlook existing votes.

Are there any other reasons? Should we try to re-start manual deletions or is the situation fine anyway?

For those that might not know as 10k+ user one can vote to delete closed questions (with 20k+ one can delete certain answers). Lists of recent existing votes are available under "tools"; for example a list of pending votes over the last 30 days [10k+ only]. The list can be considerably expanded clicking the triangle; yet if the list is too long it can get cut-off, choosing a different period of time (14d, 2d, yesterday) sometimes yields different lists.

---

(1) The precise details of auto-deletion are somewhat involved (see "by the system" in this FAQ How does deleting work? What can cause a post to be deleted, and what does that actually mean? What are the criteria for deletion?). In brief and roughly a question will be deleted after:

• a week when it has negative score and is closed and "unanswered".

• a month when it has negative score and no answer.

• a year when it has zero-score and got little attention (low-views, few comments, and no answer).

What does not get auto-deleted are questions with an answer that has positive score or is accepted. Answers with a high score typically were not deleted. What's left would be badly received questions with accepted and/or low-score answers.

Something that does get auto-deleted are questions hardly anybody took some interest in, even if there is no negative feedback whatsoever. (Only after a year, though.)

Answer 1

I support not re-starting manual deletions. The situation today with automatic deletion and no manual deletion is better than it used to be in the manual deletions era and it is also better to avoid practicing manual deletion on top of the automatic one.

(On pupular request: update)

As quid explained: "What does not get auto-deleted are closed questions with an answer that has positive score or is accepted. Answers with a high score typically were not deleted. What's left would be badly received questions with accepted and/or low-score answers."

I think that overall, badly received questions with good answers should remain on the site. Moreover, quid described the typical situation and there were also atypical cases: there were cases were controversial questions were deleted in spite of many up-votes, or as part of an open/close dispute. Finally, manual deletion is time-consuming for users and now, with automatic deletion, for no or small marginal value.

Answer 2

The title of the question mentions manual deletions, and Gil Kalai's answer voices his opposition to that. Let me answer quid's other concern: the system is currently automatically deleting some perfectly fine questions. I encourage 10k+ users to monitor the list of recently deleted questions, and vote to undelete reasonable questions that have slipped through. This is not a huge problem, only five or six questions in the last week have been undeleted (I think I voted on three or four of these), but it would seem good for people to look at the automatic deletions from time to time.

Answer 3

Guess the deleted question!

Q1: I need following books (soft copies)

I know this is not the place to ask for such help, but I cant find these books in my country and not even on line and the shipping is very expensive. If someone out there have any of these books (soft copies), please email me.

A) A. H. Zemanian, "Generalized Integral Transforms", Intersciene, New York, 1968. B) A. H. Zemanian, "Distribution Theory and Transform Analysis", McGraw-Hill, New York, 1965.

---

Q2: Time decay for Hartree equation with Coulomb potential

Are there any time-decay results for the solution of the Hartree equation \begin{equation}\frac{1}{i}\partial_t\phi-\Delta\phi=-(|x|^{-1}\ast|\phi|^2)\phi,\quad x\in\mathbb{R}^3\end{equation} in $L^p$-spaces which would ensure e.g. $\phi\in L^2((0,\infty);L^3(\mathbb{R}^3))$ or $\phi\in L^2((0,\infty);L^4(\mathbb{R}^3))$?

According to the information at DispersiveWiki, Coulomb potential is the borderline case and there is no scattering results (in the sense of asymptotic compeleteness) (??) for the solutions of the above equation. There is a paper by Hayashi and Ozawa on time-decay for Hartree equation with Coulomb or more singular potentials which makes use of pseudo-conformal invariance. This work implies that one can get rates like $\|\phi\|_4\lesssim t^{-3/8}$ or $\|\phi\|_3\lesssim t^{-1/4}$ where $\phi$ is the solution of the above equation. Those are slower rates compared to the rates at which free solution decays. Is there more recent publication which might imply better rates? Do you know of any $L^\infty$-decay results (for the above equation) which might be interpolated by mass conservation to get faster $L^p$-decay?

Hints: Both were asked about a year ago. The first is closed and has score -5. The second is open and has score 0.

Since I ask, one might still have easily guessed that Q2 was the (auto-)deleted one, while Q1 would not get (auto-)deleted [under current rules and in current form]. (Comment: following this post, this got fixed by manual votes.)

Why? Well, because Q1 has an answer that reads:

both books can be traced on amazon. there are rather cheap used copies available.

Somehow it got score 2, and that's it as regards auto-delete. (While in my opinion more likely than not it does not answer the question at all, as it is made clear that the shipping costs are the issue.)

---

Now, let me stop my attempts at humor.

I think by and large auto-delete works alright or essentially "good enough" and there is no reason for major activity or concern.

However, as the above examples show, I think, the situation is not perfect, either. This is also not a unique example; likely one could produce hundreds, but then "hundreds" are only around one percent so it is not that big a deal either. Still, a little manual fine-tuning here and there might not be entirely useless.

Now, somebody might wonder why I did not do anything about the above questions. Well, I did cast my votes. Somebody else did, too. But, we'd need a third vote, and none was coming... (until I wrote this post, see comments below).

The undeletion and deletion process seems for (almost) all practical purposes simply broken. If too few check (and to be sure, I have to confess I do not check often either), undeletions and deletions just cannot happen anymore. This is likely not a tragedy, but it seems not good, either.