7:48 AM
This question by @JeromyAnglim occupies a weird place in my mind. Although I upvoted it and think it is a fun question, I have also voted to close it as off-topic. I believe the question is only tangentially related to cogsci, and should be migrated to a broader site like academia.SE. Also, the backlink from the migration might draw some academic traffic our way :D. Comments?

3 hours later…
10:22 AM
@ArtemKaznatcheev We've had a similar issue relating to stats before. In my opinion, if it doesn't attract answers here, we could migrate yes, if it does, we can let it stay here.
Although, a simple rewrite of the title makes it off-topic here indeed.

11:15 AM
@StevenJeuris I think the stats question was actually on-topic, I believe this question might actually be off-topic. I can't see how it is specific to research in cognitive science. I am fine with it staying, but I just think it will be more consistent if we migrate it (plus we will draw attention to ourselves). We could ask some more io-psych question like "how effective are metrics at..." instead :P.

11:41 AM
@ArtemKaznatcheev I have asked for permission to send it over. It was granted and I have migrated it.

12:40 PM
Also, SE is double-counting votes :D. I cast an upvote for the question here on cogsci, then it was migrated to academia and kept the same number of upvotes (and increased Jeromy's academia rep by +30) but allowed me to cast another upvote! Heeheee :D

@ArtemKaznatcheev That... doesn't make sense
Oh, yes it does
Because the question here is closed but not deleted
so if it eventually goes away here, that rep will vanish. Don't get too attached to those specific 30 CogSci rep points ;-)

Haha, I was really surprised when I first saw this happen
when ML.SE folded
(without leaving private beta)
and all the questions were moved over to stats.SE along with all their upvotes
my rep on stats.SE jumped up like crazy

12:55 PM
0

I have to say, Rebecca confuses me greatly: Due to this, we want to see that MathJax is being used significantly on the site - otherwise, we are just needlessly slowing down the user experience on the site ...Uhhh, we don't have MathJax, so of course it's not being used! I'd like to ask th...

@JoshGitlin Well said.

@StevenJeuris I know. I don't know anything about MathJax really, but I do write RIAs for a living, so I feel they're just being... difficult...
@ArtemKaznatcheev and others, if you're in favor of MathJax, downvote this:
1

Enabling MathJax supports means that on every single page on this site, the MathJax scripts will be included, and they are not small. Due to this, we want to see that MathJax is being used significantly on the site - otherwise, we are just needlessly slowing down the user experience on the site....

1:35 PM
I'm introducing my girlfriend to Biology.SE. :)
She's been reading the site for the past 10 minutes or so. :)

@JoshGitlin downvote abilities exercised rests his downvoterevolver. Thanks for bringing this to my attention.

I don't think you're quite appreciating the significance of it
I've never seen a Q title that would have needed MathJax yet, and very few that need it compared to the other sites that have it. The MathJax include is a significant performance hit and at this point will needlessly slow the down the site

@BenBrocka I agree @StevenJeuris you need to be more careful with gfs and SE ;)

@BenBrocka Can't it be optionally enabled on questions which use it and others not?
So it's only a performance hit where it is used?
Otherwise really SE is badly designed. :)
If it has a performance hit, solve it as e.g. I suggested.

@StevenJeuris There's no good way to query whether a given page at this given second will require Mathjax or not
not without producing and processing the entire HTML structure every time before the page is loaded

1:45 PM
@BenBrocka SE knows when a page is edited, when a page is edited, it can be processed checking for Mathjax ... if it is used, set a flag to true, ... done
2

@BenBrocka if a user can't wait a fraction of a second for their question or answer to load, how much do they really care about that question or answer?
@StevenJeuris win

@ArtemKaznatcheev I think you seriously fail to understand user experience frankly
Load time does have an extremely significant effect on bounce rate on all audiences

@BenBrocka I'm sure I do fail... but I would rather have great content over a slight boost in user experience

If we reduce experience and thus users we're going to have less users to produce and curate good content

really? Are you going to stop returning to this site if the load time is a fraction of a second slower? I loaded stackexchange on something close to a 56k while on a bus between two canadian cities just to check on activity... I know that at least for me the slowdown of a fraction of a second won't kill me

1:49 PM
@ArtemKaznatcheev "Are you going to stop returning to this site if the load time is a fraction of a second slower?" .. he might not, others might.

especially if it allows me to typeset math
@StevenJeuris who out of the people that curate content on this site will stop returning?
Maybe we will get slightly fewer google hits
I agree

@ArtemKaznatcheev Potential new users ;p That was Ben's point I guess. But, you of course also have to evaluate whether properly formatted Math could attract new users. ;p

that would suck if we were relying on google hits, but at this point we don't have enough content to drive google our way
MathJax will give us an edge over other brain/mind forums
we will start asking questions they can't because of hardness of formatting
and giving answers they can't because of hardness of formatting
this will drive traffic to us

@ArtemKaznatcheev I agree. It's simply weighing advantages/disadvantages. If SE is designed as such that it has a performance hit across the entire site, that is a disadvantage (not saying it outweighs the advantage). My issue with that reasoning is it's an architectural problem which could be addressed.

1:51 PM
It's very significant

and maybe a small fraction of them will bounce from a slight loading delay (how much of a loading delay is there?) but I suspect we will gain more from the unique content

But of course since SE only started getting sciency over the past couple of months, they won't invest too much time in that yet.
@ArtemKaznatcheev According to Josh around 100ms in his tests.

@BenBrocka math.SE, cs.SE, cstheory.SE, stats.SE and the rest must survive by magic, eh?

... which to be fair seems like quite a lot.
It's noticeable.

with all the bounces they get from an extra 0.1s delay
@StevenJeuris another thing you could do is just disable MathJax for non-logged-in users
since if you already invested in making a cogsci.SE account, you are probably less likely to bounce

1:57 PM
@ArtemKaznatcheev Good idea, but that would mean you would need a way in which to convert MathJax syntax to non-mathjax syntax.

but anyways, I will let you experts figure this out. My two cents: the advantage having MathJax will bring to content will far outweigh potential delays in loading

... which probably exists.

@StevenJeuris I think non MathJax syntax just remains as $\frac{1}{2} \; \text{more fancy math stuff}$

@ArtemKaznatcheev ... for those questions that could use it. Agreed, but to be fair, I still haven't seen too many mathematical formulas around here. ;p

with the code written between \$s, but I might be wrong
@StevenJeuris why would someone try to write a formula if they know they can't?

1:58 PM
a + b = c! :)

that isn't the sort of stuff you need mathjax for :P
anyways, I have to run, that was 2 cents
cheers!

Great input. ;p

@ArtemKaznatcheev those sites make very heavy use of Mathjax
we have...what, maybe 1% of questions where it could be used?
@StevenJeuris 100ms is pretty significant if it's not the only load time (which it's certainly not)

@BenBrocka I have seen no evidence of said significant performance hit
@BenBrocka I would love to see some evidence that most people can even detect a 0.1 second delay, let alone be bothered by that
I may even ask a Cog Sci question about that
:-)
But right now I am seriously behind on work, have two major deadlines for today

2:14 PM
@JoshGitlin 100ms is pretty significant, 15% increase in page load times

I remain unconvinced
12 mins ago, by Josh Gitlin
@BenBrocka I would love to see some evidence that most people can even detect a 0.1 second delay, let alone be bothered by that
I will ask a question, probably on UX as there's not a wide enough audience here
In my experience, page load times <1 second are fantastic
Stats.SE has that with MathJax
As Nick said, the issue isn't download time it's render time
but by definition that only affects pages with MathJax equations on them!
because pages without MathJax don't need to be re-rendered
Sure, it takes CPU time to look for MathJax but I don't see how that's relevant

@JoshGitlin Of course people can, that's getting close to a very good human reaction time even
And it's absurd to say a 100ms difference is always imperceptable, you're getting into Zeno's paradox at that point
@JoshGitlin the render time is a significant (and ugly) issue but loading time is also a factor, and effects everyone
@JoshGitlin you also have a very fast internet connection

@BenBrocka I develop Rich Internet Applications for a living. I still stand by the fact that a <1 second load time is really, really good
@BenBrocka no, I have a fairly fast internet connection.
See my point about the target users of the site

@JoshGitlin it is, but it won't be <1 second for a lot of peopel

They are very likely middle- to high-income individuals who can afford a cable connection, or even more likely, researchers or students at a university who will have internet connections even better than mine

2:28 PM
Highest average US broadband speed I've seen is around 5Mbps
one sixth of your internet connection

@BenBrocka I'm actually paying for 100 Mbps but that's besides the point, lol

For the US that remains a high internet speed

5

commented: Not to be argumentative, but the comment was from over a year ago, from version 1.0 of MathJax which was much heavier. Painting that comment as a lie rather than out of date (which it is) is disingenuous - that answer simply needs updating. Also, it's a matter of perceived performance, keep in mind that we defer JS loading so of all the items listed, only all.css is actually needed to affect the page looking fully loaded. In the view of "what needs loading before the page is fully rendered correctly?", it's a much larger percentage (and HTTP connection count).

> Not to be argumentative, but the comment was from over a year ago, from version 1.0 of MathJax which was much heavier
The issue is no longer size
The new version is acceptably small
5

commented: To be clear: I'm not saying we shouldn't use MathJax, quite the contrary - it greatly enhances all the sites that benefit from it. Also, I'm not saying it's as heavy as most perceive; it certainly isn't and you've made huge improvements there as well. My only point is a simple size % isn't the only number that matters, load, rendering, and reflow order come into play here and there is a perceivable cost on that front (more so in the fresh-from-google cold-cache case). Because of that, we have to weigh that cost vs. benefit to the community for any JavaScript we add.

That's irrelevant, we're arguing over the figures you just found
that doesn't change what we're discussing at all

I disagree. Nick is saying:
> My only point is a simple size % isn't the only number that matters, load, rendering, and reflow order come into play here and there is a perceivable cost on that front (more so in the fresh-from-google cold-cache case)
but "there is a perceivable cost on that front" doesn't make sense because rendering only happens when MathJax is found
If someone can point me to empirical evidence / user complaints about the loading speed of Cross Validated, I'd like to see it
Otherwise I remain unconvinced
Also think about this: without MathJax one alternative would be posting images of equations. Not only would this be more trouble than it's worth for our users, and less accessible, but those images will likely take longer to load than MathJax itself

2:43 PM
@JoshGitlin he means there's a REALLY big differnce between content load and font rendering time
Which is pretty ugly, but unavoidable; cached JS doesn't stop it

@BenBrocka my point is: I fail to see how MathJax affects this

Affects what? The font thing? That basically is MathJAX

Right
So my point is, the fact that MathJax increases rendering time is irrelevant because that increase only occurs when there is MathJax to be rendered

That's not the case; the 100 ms loading time increase you noticed occurs for all pages
that's NOT the rendering time issue caused by font redrawing
5

commented: oh I agree, on non-math pages it's not even worth discussing, absolutely worth it. I think though that since the home page falls under the "has math" category now, it's usually the users first experience to the site that's affected. For example on stats.stackexchange.com homepage (quick test), a cold cache finish it at 836ms, then jax.js, then fontdata.js, then the fonts, finishing at 1390ms...it's the sequential nature of those requests having the biggest effect...any chance that can be improved? We load the same fonts every time, anything we can do to optimize there?

That rendering issue is more of a 550 ms difference

@BenBrocka For the first page load. Then those files are cached. And I still have not been convinced that 100ms makes any difference
30 mins ago, by Ben Brocka
@JoshGitlin Of course people can, that's getting close to a very good human reaction time even
So the most attune humans can barely notice a difference
I fail to see how this makes the site so slow users will leave. Again, still looking for empirical evidence
Plain discussion won't convince me

2:57 PM
reaction time is not "barely notice a difference"
reaction time is "notice something AND act on it"

3:09 PM
Whatever though, I'll leave you to try and convince the SE team

3:50 PM
I hope you know I mean no offense, this is just a friendly discussion @Ben :-)

Yeah, I understand

4:13 PM
@BenBrocka OK just making sure, I don't want to offend anyone!
Now back to work for me :-(

6 hours later…
10:06 PM
@JoshGitlin thanks for looking into MathJax so closely. You're my hero!