Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
To try to help combat a recent series of spam attacks in which an entire page is erased, the content replaced with spam, and title changed, I have introduced a more sophisticated spam detector which evaluates the textual similarity of edit and the original, and checks whether this ’difference ratio’ is over a certain threshold. The threshold is currently determined by whether the author of the nLab edit has a corresponding nForum user, and whether the title was changed. See github.
I expect that the thresholds and the logic will need to be tweaked and improved over time. If you find a harmless edit blocked, just raise this here. The threshold is generous in the case that one has a corresponding nForum user and does not change the title, so hopefully such false negatives should not occur too often.
Hi Richard,
thanks for this!!
Will the message issued when spam is blocked say that it is about spam? Just asking since Dacid C. just ran into some issue with edits not going through: here
Thanks for raising this, I’ll look into it later. If the spam is blocked, the error message mentions spam, but not if there is an actual error.
Hi Richard,
in trying to edit the entry Ruben Minasian (adding references) I am now getting the error message
Edit blocked by spam detector
What might cause this?
With some fiddling, I got the content at Ruben Minasian to save now.
But next I was now trying to clear the entry generalized Calabi-Yau space
(which is essentially empty and should be renamed to something like “generalized Calabi-Yau space > history”, with the term redirecting instead to the existing generalized Calabi-Yau manifold)
But trying to clear it triggers again the error message:
Edit blocked by spam detector
Hi Urs, the spam detector looks at the difference between the edits before and after and checks whether this difference is within some threshold. In both these cases I think it is probably correct that it kicked in. Especially in the second case this kind of thing is what we are trying to prevent from malicious users. Probably the solution here is to allow trusted users to bypass the spam detector in some way. I will add such a mechanism when I get the chance.
Thanks, Richard. Now I am triggering the Spam blocker in trying to add references to the entry on Frank Wilczek. As previously, I managed to work-around it after some trial-and-error by deleting most content of the entry and then adding it in again incrementally. It’s not clear to me why this is how the Spam blocker should behave. Maybe there is some room to fine-tune it a little?
By the way, I just noticed that there is a misspelled parallel entry Frank Wilcek (missing the “z”) which ought to be cleared and renamed (which I would do, but doesn’t work with the Spam blocker now).
Hi Richard,
sorry, no rush, but allow me to just record, not to forget, another little issue that occurred recently:
I gather the font size of blockquote-environments got reduced recently. But now I see that the inline maths in blockquotes doesn’t scale along well, e.g. this example.
As I said, no rush, I understand you have other things to take care of. Just to mention it for when there may be time.
Thanks for noting these things here Urs. Just keep doing so, and I’ll fix as soon as I get a chance.
Hi Richard,
okay, here is one more bug which just came up, slightly alarming:
The page R-symmetry currently does show my latest edit from a few minutes back, but hitting edit gives me the source code from before these edits.
So for instance the word “parity” currently appears in the displayed page, but not in the source code that is shown upon hitting edit (same when hitting source)
Hi Richard,
there is Spam that just came in (Comment_78643) but I can’t clear it since the spam blocker prevents me from removing spam.
Given what we have so far, allow me to suggest that the spam blocker in it’s current form is maybe not that much of a help. Might it make sense to remove it until we come up with another idea for how it could operate?
Hi Urs, apologies for lack of action, I have been too busy the last few days to address this. I have a way to improve it which I think will work well, I’ll implement it when I get the chance. Unfortunately you are not a typical user! My improvement will address that (as I 100% agree is clearly needed). The main test of the spam detector will be whether it prevents what it is intended to prevent. It is only for page edits, not page creations (ref. other thread, now deleted).
I implemented it after all. The spam detector should now never kick in for you Urs, or any of the steering committee. Basically the threshold now takes into account the number of pages which the author of edit has previously edited. github commit
I have now deleted the page referred to in #12, and the associated nForum thread. Just to emphasise once more: there is no spam detection on page creations, only on page edits, so we can still certainly can get spam on page creations.
I will take a look at #9 and #11 when I get a chance.
Okay, thanks!
Hi Richard,
if I may, pointing out just one other little thing:
Recently the font size in blockquote environments got reduced noticeably, but the inline maths in these blockquotes currently stays overly large in comparison (e.g. here)
Hi Richard,
just an idea: For the xymatrix-functionality it would be useful if some of the basic maths-typesetting packages were loaded, for instance amssymb to have fraktur symbols, etc.
If this is possible, maybe you could pass all of
\usepackage{amssymb, amsmath, amsthm, mathtools}
?
Hi Urs, I have addressed #18 now (this was a problem only in Firefox or elsewhere where MathML was being directly rendered; MathJax adjusted it itself). Let me know if it does not look OK. Let me also know if the quote size should be increased a bit, or if the typesetting of quotes shoud be otherwise tweaked. Zoran was quite unhappy about the visibility of them before, so they need to be definitely distinguishable.
Re #19: done. Apart from mathtools, these packages were already loaded for tikz diagrams, but not for xypic ones.
See github for the precise code changes.
Richard, thanks!!
Regarding the quote formatting: Yes, I would vote for increasing the font size a little more. There is room to increase it while it still being smaller than the main text, it seems.
I think all we need is the quote to be clearly recognizable as such. Since the left indention makes that very clear already, the font size does not need to be all that small. I’d think.
OK, thanks for the feedback. I have made the font size a little larger now, and reduced the margin a little as well.
1 to 22 of 22