The Whiteboard

@Rachel good luck, odds are very low given the "profile" of the OP

I know the basis... or at least, I think I do. You try to circle the other person's pieces

Robert Harvey

I was actually kicking around some Othello algorithms while I was out of work. His stuff is very primitive, just a starting point.

is Othello and Go the same game?

lol

9:02 PM

nm, Yahoo answers to the rescue for my dumb questions :D

answers.yahoo.com/question/index?qid=20080112054042AAeKRgP

ahh yes, I've played an electronic version of both those at some point... hrm

I've heard the news lately of making a self-learning AI actually teach itself Go, and beat the #4 guy in the world

in a best of 5 match

was some talk about how much more advanced that was than AI chess because of the number of potential moves for Go players, and it wasn't something that you could just go "figure out all potential moves, and then pick the best one"

Steven Burnap

The Go AI is extremely impressive as it's the first time they've had to do something other than minmax with heuristics and lots of CPU to beat a human. I'd love to see the actual details.

Yep, I've always been interested in AI learning and stuff like that

just noticed the time though, I'm out of here :)

the articles about it always say the AI got better by playing itself zillions of times, so I assume it's doing some kind of machine learning technique and using itself as training data after a certain point

the big challenge with that of course is avoiding overfitting, how they managed that I dunno

Steven Burnap

What they used to beat Kasparov was still ultimately based on what they teach undergraduates in AI 101, with some bells and whistles and a huge library of canned responses.

They claim the technique they used for Go can ultimately learn and beat games purely by training. That's huge

I wrote a minmax checkers AI in comp sci 40, it's not even upper division content anymore

user114359

9:07 PM

someone (don't remember who) said the meatball wiki had a good article on quitting an online community, but I can't find it.

Yeah, I heard that chess is much easier to code AI for because there are limited number of moves, and you can just number crunch the best move to make

chess is at the level where you can't feasibly "just" crunch the numbers, but you can use that as your core strategy and make it work by identifying a bunch of cases where crunching can be skipped

Steven Burnap

It's just make a scored tree of positions to a certain depths of moves, assume both players picked the highest scored move on their turn, prune the tree based on heuristics.

checkers is the sort of thing where a computer can literally play all possible combinations and output the mathematically optimal game(s)

whatsisname

isn't checkers a solved game where you don't even need to do that

Q: 1, 2, 3...test. Let’s increase the number of reviews & close vote for science!

9:09 PM

there's a paper somewhere saying whether red or black wins, I can never remember which it is

@whatsisname I was trying to say what "solved game" means without assuming everyone knows what that means

in the informal sense, Go has been "solved"

user41796

@Snowman meatballwiki.org/wiki/GoodBye

gnat

68

For many, many years users have asked and asked and asked for changes to the number of close votes and the number of reviews per day. There's been a little bit of skepticism as to the efficacy of such changes, but so far that's been mostly speculative. So... Let's test it! Current Problem: We'v...

^^^ sounds familiar

user114359

@gnat Looks like they used Programmers as a guinea pig for SO. At least this time they have a better CM making the announcement.

if that's actually what happened...they really ought to have mentioned that when they announced the experiment here

user41796

@Ixrec I don't think the experiment here was run as formally

user41796

9:17 PM

more of a Jon & Shog had a conversation and said "You know, it would be interesting to see what happens when ..."

user114359

> Silent departures
In most cases, leaving a community is a process rather than an event. There is progressive disengagement as community involvement takes a lower priority than other activities. In time, an attitude shift takes place; as the departee internalizes a conclusion that the community no longer matters, or that it is beyond their control.

user41796

And the results weren't sufficiently interesting.

it's also a different test

user41796

@Snowman Yep, he followed the playbook to a T

they are increasing the CV queue size, not the number of CVs required to close questions

user114359

9:19 PM

@GlenH7 the only difference is that the MeatballWiki page says it is possible to reverse a silent departure: on SE, it is impossible to revert an account deletion.

user41796

Account was disassociated, not deleted.

I think that's permanent, though, isn't it?

user41796

@enderland No, I don't think so

user41796

I think it can be undone

@GlenH7 I suppose you're right. Still, I'm gonna miss him.

user41796

9:20 PM

At that point, you're already having a direct conversation with SE folk, including the dev team

user41796

As the disassociation breaks the DB link between tableOfPII and seSiteUsers

user41796

@Ampt Same here

user41796

And I do wish he'd said goodbye. But I certainly understand why he didn't.

user114359

Well I already lost ~~rep~~ imaginary internet points thanks to him. That is not supposed to happen with a high rep user, but it did.

user41796

@Snowman What site?

user114359

9:23 PM

At Movies.SE I lost 5, guessing he upvoted a question of mine. Here I had several -2s due to the user being removed.

@Snowman ooooh. that explains my random upvotes being lost on other sites

heh... that explains it for me too. I lost some points on Math.SE

user114359

Movies.SE makes sense, he had a few hundred rep like me

user41796

There's a rep level (300 or 500?) that affects how votes are handled

user114359

but here? He was a 40k user, and the page explaining that reason specifically states high rep users will not trigger reversals because it would affect too many people

9:24 PM

it's probably per site

...I only just realized that -10 "user was removed" I got on ELU.SE the other day might've been MichaelT

@Snowman probably only on the site you have high rep on then

user41796

Well, let me rephrase things. His progs account was disassociated. Other accounts may have been deleted

user41796

@Ixrec happened in the last 24 hours if I read comment times correctly

user114359

I don't get the -2 reversals on Programmers.SE - the only way to get exactly 2 rep is to have an edit approved. I can make my own edits except for tag wikis (almost to 20k...) but haven't done one in a while. But that rep shouldn't be reversed anyway.

9:25 PM

checks

by "the other day" I appear to have meant "yesterday"

user114359

As far as I can tell it was yesterday morning, US time.

user114359

is there a tag other than "feature request" I can use on meta to ping someone who can explain? It's not enough rep to get upset about, I'm curious more than anything.

that might be correlation than causation?

user41796

@Snowman it's been answered on MSE before

user114359

This removal occurs whenever a user is deleted, unless that user had a very high reputation score. Because high-reputation users have usually cast a great many votes, removing all of them could be that much more disruptive to other users. In such cases, the staff use a special deletion that preserves the votes, resulting in no reputation change for those who had been voted on by that user.

Robert Harvey

9:30 PM

@Snowman Accepting someone else's answer also gets you +2 rep.

user114359

@RobertHarvey so if I accepted his answer to my question eight times, that could explain it. I have asked a total of four questions, and he has posted answers to zero of them. So while you are correct, it does not explain this particular rep loss

also -2 just shows up for a downvote too >.>

or was it "user was removed" ?

mine was "User was removed"

user114359

in other news, @amon just got 10+15=25. I was going to accept his answer but forgot, looking through my questions I noticed one without an accepted answer.

Just read the meatspace wiki on leaving... I feel like we've been had!

what if @MichaelT was a bot the whole time simply following the Meatspace wiki's instructions and ended up on this article

The long con

Q: What does "user was removed" mean and why did my reputation change because of it?

9:35 PM

And what about all the people who won't be able to join the community because they're terrible at making helpful and constructive co-- ... oh.

someone succeeded then :-)

user114359

@enderland I thought that explained gnat

user114359

713

Message appearing when a user was removed. It seems I lost 5 reputation points on Stack Overflow yesterday. The reason given in my history is "removed", with the description of "User was removed (learn more)". I don't understand what that means. What caused this? Return to FAQ index

support faq reputation reputation-history

well it's 4:37 and our production author server is still down

how long after my bosses leave is long enough to go down to the bar to grab a beer while I wait on ops to load the tape

user41796

@Ampt 10, 15 mins tops

user41796

Just long enough to be sure they aren't going to turn around because they forgot something

A: How did this curious undownvote pattern come about?

9:39 PM

@Ampt why do you even care?

just leave?

user41796

@enderland he can't. He's stuck waiting on someone else so he can do the needful before heading home for the evening.

user41796

Must. Do. The. Needful.

user114359

36

This is a bug in our vote auto-invalidation task. When we delete highly active users upon their request (i.e. they no longer wish to participate in Stack Exchange), we preserve their up/down/accepted votes by moving them to our Community User. This is the first time we had removed a meta.SO use...

user41796

Community just picked up a helluva lot more votes...

user114359

After reading on Meta.SE I am still confused about those reversals, given we are/were both high-rep users