« first day (2487 days earlier)      last day (2751 days later) » 

user315433
00:17
Of course in 2017 it needs to be Machine Learning, would not sell if it was just Statistics.
user315433
I don't recommend that kind of high-dimensional stuff. As a first step, I'd come up with a measure of similarity of close voters, that fits in a matrix N*N where N is the number of close voters.
user315433
The simplest would be just the number of questions that both A and B closed together. A better one (probably) is (A and B closed)/sqrt((A closed) *(B closed))
I'm really fascinated by that pdf. It's just going to take me a while to process it since I'm missing a lot of fundamental background knowledge.
Yeah maybe something like that.
user315433
Kind of like cosine similarity, that. I don't know what kind of structures Matlab clustering accepts as input, though. Don't have that toolbox.
As simple as M(i,j) = % of times j appeared where i was. Clunky, maybe, but straightforward.
user315433
00:21
That's not symmetric, which may or may not be a problem.
Although hard to identify independent variables from that. Or something. I dunno.
user315433
The thing I suggested is actually sqrt(M(i,j)*M(j,i)). The root is optional.
Huh. Yeah. That's more informative.
But the inverse correlations are still invisible there. Like if !i implies j.
Although do I really care
No I probably don't
I definitely don't
I think I'll try that and see what comes out
Is it possible to determine the close reason from sede?
Oh yeah pendingflags
Oh no, it's in the comment field, right
user315433
00:41
> Q: Why is @glitch friendlier than @StackOverflow - @anildash says it's someone's job to monitor the community- u need humans! #DPLUSG -- Kathy Sandler at 5:12 PM - 25 May 2017
user315433
I thought humans were the source of the problem there.
Tempted to answer #moderatorsarehumansexceptwhentheyaredogs.
user202362
I can see dogs as moderators, I can't see cats or raccoons as moderators
00:57
oh, one of the CMs has a red panda ava.
Tho I don't think she went the mod -> CM route
user202362
red panda makes good mod too
user315433
@ShadowWizard Thanks for checking in, and sorry the ping languished here awhile. I'm very glad to see that folks found this valuable enough that they're eagerly awaiting the next one. The season has rolled around where the team is doing lots of planning as well as preparing for our annual company gathering. Since we're kinda jammed up, the 3rd 'Hall will be in early November. I'll be circulating a new post to gather suggested topics before then, so stay tuned. — Ana ♦ Sep 23 '16 at 17:22
user315433
Early November of what year?
user315433
We missed the opportunity to holds it right after the U.S. election.
user202362
01:19
Surely if A.I. is so powerful, Trump's relation with Russia should be obvious & self driving car technology should be mature?
@Gerry yeah. and dang, time flies
I missed that one cause, I was probably starting on job -1 and the timezones didn't match up
user202362
I need to mow the lawns again, 2 backyards, 1000 square meters of grass
user202362
but I enjoy physical work, just like I enjoy solving puzzles and work on software
01:44
thinking shorter chats more often might work better
Do a half-hour session every two weeks maybe
Less pressure if you miss it
@TimStone In SEDE, how does /query/run know what site to run on?
02:04
The first URL parameter is the site ID in the main SEDE database
Ah ha, thanks.
(You can find the appropriate ID via the /sites route)
Oh neat!
Sweet
user202362
02:33
an A.I. in need is an A.I. indeed - Telkitty™
user202362
please stop making useless things
user202362
(goes back to do something on the useless funny cat app)
user202362
have you realised that human is the only species that create useless things?
@SmokeDetector tpu-
user202362
Manure from livestock are fertilizers
user202362
02:43
various part of plants are eaten or become useful in ecosystem somewhere
user202362
but things human create - like used plastic bags, obsolete software, no rechargeable batteries, can be totally useless in ecosystem for thousands of years if not longer
user202362
since I am no genius, I couldn't be the only one who have thought of this
03:21
people are good at destroying things
03:45
read as: "elephant people are good at destroying things"
Truthy.
04:10
user202362
04:51
I bought another drone yesterday. My planned to use the last drone to stalk my chicken, but I lost it on the second day
user202362
I wonder if I start a channel/blog 'bushwalking with chicken and drone' will gain many followers
user202362
06:27
Stack Overflow is growing fat, and our infrastructure needs just keep getting bigger
user202362
makes sense
07:03
14 messages moved to Chimney
 
1 hour later…
08:24
SO IS DEAD!!!
OUR LIVES ARE OVER!
08:49
o shi, I suggested an edit without logging in
there goes my 2 rep...
09:25
!!/alive
10:04
!!/alive
@Elephant You doubt me?
@SmokeDetector kind of.
user202362
10:24
!!/alive
@Telkitty Of course
Apologies to users who saw home page errors today (8,372 hits). A single web server went off the rails and is being punished appropriately.
@SNAFU how to punish a server
 
3 hours later…
13:29
man, the spammers are busy today
user315433
Site birthday: Server Fault graduated 8 years ago today.
user315433
The most interesting thing about SF is whether it's the future of SO that played out at an accelerated pace.
user202362
I like your optimism
user202362
probably unthinkable, but the truth is that even companies like google and microsoft will be dead one day, just like that we will all be dead one day
13:43
@Gerry the social dynamic of SF users is... strange
there's a lot of people who are both disgruntled, yet care deeply for the site
user202362
but companies are like trees - it can be dead in less than a year or alive for more than 300 years
user202362
so, now, do you feel rather dumb?
user315433
14:10
Yay I can now approve tag wiki edits on SO (again). Which are the nearest thing to Documentation that the site had before Documentation. Reviews require 5000 rep for tag wiki vs 100 for Docs.
user202362
14:32
user202362
butter cream cakes, yum
user202362
the love of my life at this particular point in time
14:55
20K .net meta.stackexchange.com/questions/296439/… needs one more /cc @Bart @ShadowWizard
user315433
@FireAlarm Hi, do you speak any human languages?
15:22
@JasonC Yanno, I'm going through your meta samples...
there needs to be a too short option
@rene gone
I saw a question that was historically locked on crypto, but I think it should be moved to superuser. Is it possible to flag it? Is this the right place to ask?
Historical lock makes me think it's older than 60 days? In that case there can be no migration
3 years, 7 months ago
but it was just locked Mar 6
Which is still older than 60 days...
15:44
@BrianMinton you can bring it up on crypto's meta, and if there's a strong desire to migrate then I can do it. But first, make sure the topic isn't already well-covered on SU.
@BrianMinton Also make sure SuperUser want that question. Maybe @JourneymanGeek can advice on that.
Hi! I felt my ears burning ;p
You're welcome
@BrianMinton I can take a look to see if shog should take a look ;p
We'll give that question the dead stare
15:48
@JourneymanGeek Sure. Let's say write "s" in "too long" field for now.
Yay I can use the API again. Now I'm afraid, lol.
As I probably should be.
25 messages moved to Chimney
user315433
Watching backoff and quota_remaining fields usually helps
16:03
@Gerry Infinite loop yesterday and I made 8500-ish requests in about 15 seconds, so, yeah. Backoff time was 12 hours after the IP ban went away, heh.
I just permanently suck at JavaScript, it's weird.
It seems like it should be on SU because "If your question is about usage of a specific cryptographic software (not its cryptographic internals), Super User is the right site."
I'll have to dig a bit to see if it's already on SU.
16:19
superuser.com/search?q=s_client+curve doesn't show anything relevant.
user315433
16:53
Apparently, @AdamSmitht1 tweets all SO questions, linking to some ad-filled thing that might show the question by proxy if it works. 1.15M tweets so far.
17:03
@Gerry That's obnoxious.
I wonder if that counts as a "scraper site", conceptually.
17:27
@JasonC oh, I thought I was alone in that struggle ...
Nope. I assure you, it's constant fail over here.
@BrianMinton that question seems more to be about the protocol then it is about the tool, openssl in this case. Questions about crypto protocols are on-topic. Question that are on-topic doesn't need to be migrated. On top of that: I doubt if anyone on superuser could even answer that. In other words: On super user you'll find the wrong audience for that specific question.
17:47
Whoa
user315433
18:05
Can anyone work on the edit queue? by Kate Gregory on health.meta.stackexchange.com
user315433
s/on the edit queue/on anything at all/
What's the suggested edit review limit on beta sites? 20?
user315433
18:21
For one user, yes. But over there, the entire queue is full - nobody can suggest edits. There are 28 edits there, apparently...
user315433
I don't know if 28 is really the limit, or there's some caching as usual.
user315433
The last time someone reviewed suggested edits there was Lucky on May 15 (and before that, Fomite on May 4). The other reviews since then were by OP on own post.
Potentially bleak future, but manageable queue right now.
Also there's only 3 users with rep >= 900 and < 1000, i.e. near future potential new editors.
 
1 hour later…
20:08
@rene this time @Bart got to it!
@rene ouch! My eyes hurt..... it was bad enough for @Shog himself to delete it, lol ;)
user315433
20:34
Is this a coincidence, or a misunderstood joke? First, Nick Craver's tweet and then another one.
user315433
user315433
> More famous web applications build from micro services. Just confirmed Stack Overflow doing it. Obviously, it makes them more productive. -- Vignesh Kiswanth at 11:09 AM - 26 May 2017
user315433
Is there a new alignment issue with profile pages‌​?
user315433
user315433
I think the numbers+text used to be left-aligned here.
@Gerry could be. CSS got some love
21:51
@Shog9 Not sure if this is a known bug. But I went through this sequence of events today:
- Downvote an answer (-1).
- Remove downvote (+1).
- Hit the repcap.
- The next upvote gives +1. (+201 for the day)
- Force a rep-recalc by deleting/undeleting an answer. (Makes the extra +1 go away)
- Next upvote still gives +1. (still +201 for the day)
22:07
I think if the recalc does the right thing, it's pretty much by-design. (But that's a funny little edgecase.)
user315433
22:22
Was thinking of JavaScript addition today... you know, 37 + 23 = 3723. It sometimes happens that a+b = b+a, but all examples I can think of are boring, like a = 1212 and b = 121212, where both a and b are repetitions of the same string c. Are there any non-boring examples of a+b = b+a?
@Gerry No. In fact a+b=b+a isn't even possible unless a is 1 or more repeats of b, or vice versa. So they'll always look like that.
user315433
@JasonC I gave an example where neither is a repeat of the other.
Yeah but you said "non-boring".
And 37 + 23 is not equal to 23 + 37.
I know what you mean though and I meant the same thing but didn't phrase it formally enough.
a+b==b+a iff a and b are both repeats of some c.
assuming all strings
Yeah that. And the reason you can't think of other examples is because there aren't any.
user315433
22:28
Okay, I'd like a proof then...
user315433
Never mind, got it.
user315433
...or not, there was a gap. Grr, it must be on math.se or somewhere already.
Proof probably has something to do with decomposing it into a + (c + d) = (c + d) + a or something.
user315433
Search leads to Wikipedia: "Concatenation, the act of joining character strings together, is a noncommutative operation.
For example, EA+T=EAT ≠ TEA=T+EA
One thing you do know is that a[n] must equal b[n] for all n < min(len(a), len(b)). Because if a + b = b + a, both strings must start with the same thing.
user315433
22:32
Yes, the shorter must be a prefix and suffix of the longer one.
You could probably also derive a proof based on (a+b)[n] = (b+a)[n]
That's not enough, because 12 is a prefix and suffix of 123412, but 12 + 123412 != 123412 + 12
It must be a prefix, suffix, and middlefix :D
12 + 12312312 :P
It must cover all the middlefixes.
Youre a middlefix
22:35
This feels like a prolog type of problem.
I'm sure it's a simple proof
user315433
Say, a is longer than b. If we can write a=bcb, then from bcbb=bbcb it follows that cb=bc, and proof completes by induction on the length. Problem is, we don't necessarily have a=bcb because two copies of b inside of a may overlap (prefix and suffix).
It's probably easier to find proofs using whatever the correct math words are, like sequences and crap.
user315433
Concatenation theory, also called string theory, character-string theory, or theoretical syntax, studies character strings over finite alphabets of characters, signs, symbols, or marks. String theory is foundational for formal linguistics, computer science, logic, and metamathematics especially proof theory. A generative grammar can be seen as a recursive definition in string theory. The most basic operation on strings is concatenation; connect two strings to form a longer string whose length is the sum of the lengths of those two strings. ABCDE is the concatenation of AB with CDE, in symbols ABCDE...
user315433
"also called string theory". 8-/
22:38
juspreetsandhu.wordpress.com/2015/05/25/… #2 on that page, at least at a glance
I don't think that proof is very clearly stated though.
But something like that.
user315433
No, it simply says there exist a and b such that ab is not ba. Duh.
That's true
user315433
The closest I found was Levi's lemma. Looks like this sort of thing are more likely to be on Computer Science than Mathematics
I guess you have to set out to prove that if you let a+b=b+a, then a and b must therefore be repeats of some c. And that'd cover it.
user315433
Of course!
user315433
22:43
3
Q: Prove that a and b commute if and only if they are powers of the same word

ttkTwo words $a$ and $b$ commute if $ab=ba$. How to demonstrate that $a$ and $b$ commute if and only if they are powers of the same word ? $ab=ba \Leftrightarrow \exists \space \tau \in \Sigma^{*} \mid a=\tau^{k} and \space b=\tau^{k'}$ The demonstration in this direction is trivial: $\spac...

There you go
user315433
asked 12 days ago
Haha
The inductive version is there too
You know what's bumming me out here, though
I feel like 15 years ago I would've been able to immediately just blurt out a correct proof of that
user315433
Yes, it's almost what I tried, except I tried to do too much, use both prefix and suffix. Prefix is enough.
And now I'm sitting here fumbling around trying to make my brain work
user315433
22:46
Yuval's proof looks overcomplicated.
Agreed
Well I'm going to go buy cigarettes and sit on my front stoop feeling sorry for myself for awhile. Bbl.
user315433
Weird how CS formats the "active ... ago" line with special link color. On other sites it's same color as other lines above it.
user315433
user315433
Too much "click me" there.
user315433
JS-inspired arithmetical puzzles are sometimes fun. For example: 2, 3, 5, 7, 23, 37, 53, 73, 373 are the only numbers where every substring is prime.
user315433
23:00
Who suggests edits to Facebook pages?
user315433
user315433
Not a govt organization here. Pub? Oh yes, got several of those.
"Hmm" is an incredibly accurate description, lol.
user315433
@SmokeDetector notspam
user315433
23:10
Law is two years old today. And already graduated... Most viewed question (by far): What laws are there regarding being pulled over by an unmarked cop car/undercover cop?
I wonder if healthyalgorithms.com should be removed from the blacklist, it doesn't seem particularly spammy.
user315433
It's not blacklisted, it's just a pattern for healthysomething.com
user315433
Could add a negative lookahead there, but meh. There are probably other non-spam sites like that. Most of the time the rep=1 filter takes care of false positives.
Yeah. Meh.
23:20
@Gerry Ooh, these are nice.
I'd prefer to have it say sin(x)^2 since I use that less ambiguous notation.
23:51
@BrianMinton deleted :(
user315433
7
Q: With OpenSSL and ECDHE, how to show the actual curve being used?

oberstetUsing openssl s_client -host myserver.net -port 443 I can see the cipher negotiated is indeed using ECDHE for session key exchange: SSL handshake has read 5894 bytes and written 447 bytes --- New, TLSv1/SSLv3, Cipher is ECDHE-RSA-AES256-GCM-SHA384 Server public key is 2048 bit Secure Renegoti...

user315433
Broken link in the chat message
ah
hmm. I would guess its on topic, even if somewhat specialised @BrianMinton - if @Shog9's inclined, I wouldn't have any issues with a migration

« first day (2487 days earlier)      last day (2751 days later) »