« first day (1534 days earlier)      last day (533 days later) » 

user351417
4:06 AM
@rob sure, they're not going to care that we don't like those questions. But still, how do they find us? There's got to be a couple of major inlets which we can try to block.
 
4:47 AM
The vast majority of our traffic comes from Google searches, and we can't really block those off, nor do we want to.
Among traffic which comes from links, most of it comes from other SE sites, and I don't think there's anywhere in the network that is recommending us as a place to get homework problems solved.
 
user351417
Ouch. I guess it'll be to hard and unreliable to set up something which allows google hits only for specific physics questions and not for general 'I want physics help'-type google searches?
 
Yeah, hard and unreliable at best, more likely impossible. I don't think Google even passes on the search terms in many cases.
 
user351417
5:03 AM
Update: I've searched for 'physics homework', 'physics questions', 'physics', 'physics q&a', and a couple of other similar nonsense. PSE didn't turn up even once.
 
Huh, that's not good. What do you find for those searches?
 
user351417
Khan academy, chegg, physicsq&aexchange (not SE), study.com, and some university links like cornell askastro and illinois.
 
user351417
and askphysics.com
 
user351417
Those repeated several times. The others were unique for each search
 
This might be something to talk to SE about. They have put a fair amount of work into optimizing our position in search results, although they've focused more on making sure questions show up high in the results for searches relevant to that specific question, less so on the ranking of the site as a whole.
I or one of the other mods can bring it up with them.
 
user351417
5:10 AM
This leads me to believe that people actually search online for their question before they actually ask that on physics SE. Because when I search for specific physics concepts, SE frequently turns up near the top (but that may be due to my personalized search options, since my history shows that I frequently visit PSE).
 
user351417
Maybe near-immediate deletion of older homework questions, which probably closely resemble the new ones people ask, would help?
 
It's possible, although we do want to be wary about deleting anything too quickly. We don't want to shut off the community's chance to review it.
 
user351417
I would say that our track record shows that we don't make too many mistakes with homework questions' deletion. Looking at meta, I see several requests for reopening questions, out of which practically none are homework. Of course, it could be because the askers don't have the reputation for meta, but still...
 
user351417
On the other hand, we could end up putting off someone who's actually good, just by an accident, and if that happens to even a couple of people, it would be pretty terrible for the community.
 
5:27 AM
Yeah... that's part of the reason the mods are reluctant to delete questions manually, except when they're spam or some such thing.
In any case, questions which are downvoted and don't have upvoted answers (plus some other minor criteria) are automatically deleted by the system.
 
 
8 hours later…
user351417
1:15 PM
Another update, because I'm kind of jobless. I tried to paste a few problems from a textbook in the Google search box. No appearances by Physics SE. Apparently I was wrong, and people don't see old homework questions and get drawn here. But when I search for legitimate concepts which I don't know much about, PSE turns up reliably. (This is only considering page 1 of search results)
 
user351417
Does anybody have any idea how we attract so much traffic from homework-help-seekers?
 
user351417
I'm stumped. It doesn't show up in general searches for help. I've never heard anything by word-of-mouth. It doesn't show up when you search for specific problems.
 
1:28 PM
According to some posts on Meta Stack Exchange, in site analytics there is (or at least was) moderator-only tab which contains search keywords used to arrive to the site.
So perhaps mods could help you out @Chair. Although as you see from my link, some people have complained about quality of the output shown in this tab.
 
1:42 PM
BTW for me physics homeworks exercises returns as the first hits some posts from this site and from its meta. (Which is rather unsurprising, since I used phrase very similar to name of a specific tag. In fact, I expected to see more results from here.)
 
user351417
1:53 PM
@MartinSleziak I dunno, the likelihood of a random lazy person using the exact name of the tag is pretty low.
 
user351417
I don't have the rep to view site analytics myself, but if there's something interesting/useful there, someone'll point it out soon enough
 
@Chair Well, from the meta posts is seems that this feature (keywords) is mod-only. Not the "regular" analytics accessible to users above 25k.
 
user351417
Maybe a mod'll mention something... We'll see.
 
rob
7:52 PM
@MartinSleziak The search keywords analytics are pretty useless. Out of something like 10M search engine referrals, there are about 9.5M for which the search terms were not provided. For rare search terms, the analytics page gives "samples" of search phrases, which are mostly obvious singletons.
For example, the page currently says that 1750 people came to the site using the search phrase "how do you defeat that monster vampire in the sewers in witcher3 when trying to help triss and the mages escape novigrad"; that 1750 people came to the site using the search phrase "what is a curl in electromagnetic"; and that 3501 people came using the search phrase "derivative sqrt(-g) metric".
I interpret that to mean that the first two search phrases were each used once, that the person using the third phrase visited two pages, and that the list of search phrases I can see on the moderator page represents about 0.06% of the total search terms.
 
8:11 PM
@Chair Some of what I told you above was based on the mod analytics.
 
@rob The SE network: "Come for Witcher 3 advice, stay for the physics!"
 
I've speculated in the past that it probably stores fractions, not absolute hit counts, and rounds to some number of decimal places that corresponds to e.g. 1750.4 hits.
 
rob
8:29 PM
@DavidZ I found the sampling explanation in some actual documentation a couple of years ago.
 
Oh, interesting, I never knew that was described anywhere (outside of the company)
 

« first day (1534 days earlier)      last day (533 days later) »