« first day (839 days earlier)      last day (3325 days later) » 

7:00 PM
Unless you're referring to actually winning the vote
 
yeah
 
not just passing the threshold
 
I'm referring to winning, yes
 
in which case carry on, ignore me
 
@ton.yeung They have too many regs.
There's a Goldilocks Zone is what I'm getting at.
 
7:01 PM
Oh my god, it's 1pm. I've lost like 3 hours to this.
 
@Eric I know. I needed to forget about this so long ago
 
2
A: Why am I suspended from chat?

Shog9My guess would be that movies and tv didn't want to be restarted.

^^ lol
 
@ton.yeung Too few regulars: can't muster minimum votes to be consistent with StackEgg (plus you're more vulnerable to raiding). Too many regulars: can't overcome general inertia of dumb.
 
@Yuuki Sounds like a lack to team coordination to me
 
7:03 PM
automation solves that
 
Hmm would be nice if the bot occasionally updated empty ones so it never gets to the warning level, but oh well
 
botters play twitch plays <game>
@Eric there might be a problem with that algo :p
 
@Hakase Yeah, that I made it. :P
 
lol
hmm it seems to be resolving itself
 
Hmm 6am. Spent way too long on this, so I might head off now :P
Nice knowing you all!
 
7:07 PM
@Sp3000 o/ come again~
 
see ya
visit us sometime again maybe
 
PPCG got another win, much faster.
Brought their average of 311 down to 304.
 
hey you lurker guys: we're not plotting anything~ we're just tryin' to live a dream
 
@ʞɹɐzǝɹ I believe it
Now I'm just hanging around because I secretly like anime
for no reason
 
well, maybe you should start loving it more openly
you could get recommendations that way
it's beneficial
 
7:12 PM
@John here's what other anime are doing for april fools: reddit.com/r/anime/comments/30xmfq/…
 
@ʞɹɐzǝɹ haha! :D
@Hakase Perhaps, but seeing as I don't actually know any Japanese, I should maybe wait until after April Fools day :)
 
why do you have to know japanese
 
@Hakase because they speak japanese in anime
and especially on twitter
 
@Hakase 's japanese is terrible and he still watches anime
 
oh dang I guess I never noticed
wtf is wrong with me
 
7:15 PM
@Hakase are you all funky again
 
did I have too little tea?
 
I have this one friend that is really really into anime
and he speaks japanese
 
he's hack a saying
 
so it's slightly intimidating
 
@Hakase have more cake
 
7:16 PM
@ʞɹɐzǝɹ perhaps, yes
@ton.yeung it showed content for a second and then some stupid message
second
first one shows everything fine
I live where I want to :p
 
Typical US Gov
 
sounds not cool
 
@Hakase don't worry the govnt like to be in your face
this is so you know that they're there.
 
7:20 PM
so this is what freedom's like
we do things more stealthily
 
@Hakase ppl call alot of things freedom
doesn't mean they're actually free
like the anime Free
 
Lol
 
they say freedom comes at a price
 
Communism is true freedom
 
@nhahtdh on paper, yes I would agree.
 
7:23 PM
but I'm tired and don't really care about any of that, and they also say a bunch of other stuff, so whatevs
time for some more tea and then sleep
 
@ʞɹɐzǝɹ Well, I could tell you what we are free to do, and what we are not
 
@Hakase what a coincidence tea makes me fall asleep too
 
I think there are 3 bots running now, one of which is laggy, or it might just be a very conforming user
 
@nhahtdh I am well acquainted with the duality of communism
 
@Hakase No, it's that the timing is weird.
I'm the "lagging" bot. But to me, the other two are lagging.
 
7:25 PM
lol
 
My vote goes in at 19 seconds and the other two at 9 seconds.
 
maybe reload the page so we get synced
 
@ʞɹɐzǝɹ Me, too, about the so called freedom in the 1st world countries
 
Well, unless it gets more desynced we shouldn't have a problem.
 
7:57 PM
Shirobako BD corrections (epi. 10-12) moeponzu.com/blog-entry-4861.html
spot the differences!
 
Oh wow we won an internet while I was gone
 
Yup. We are still between 130 and 140 days (+ 1 win) behind PPCG
 
Is PPCG using the same ordering as us?
 
I can't really tell.
They have quite a divided voting pattern.
It looks like they do it in order, though
Ask, Answer, Upvote, Downvote.
 
looks like it's different. I don't know if theirs is better though.
Should check their chatroom.
 
8:08 PM
No need. We're basically top down except quality over users ;)
 
Same as us, then.
 
Except for some that are voting out of sync, probably haven't been to chat.
 
Yeah. It definitely seems like sometimes Users takes priority over Quality.
 
@Madara where did you determine the ordering? Is it optimal?
 
Yea, our first couple wins were really high days, the last couple have been around 275.
 
8:09 PM
I think our last one was 283.
 
I don't know if PPCG knows optimal either, but they seem to be doing things somewhat differently.
 
Arqade got a 251 day win, but I don't know if that was a fluke or not. They have a post open about optimal strategy there.
 
Their top-voted answer is what we're doing.
 
I think I agree that beginning a phase with a close vote then a flag probably speeds things up.
 
8:14 PM
I dunno about the close vote, but doing one flag seems to be beneficial.
Instant boost to 1 star in each, and we have a flag to spare.
 
Close vote boosts the quality a lot initially. Then the flag gets rid of the negative side-effect of lowering the other stats.
 
Lowering the other stats?
They're at 0 after each phase starts, though.
 
If we're going to flag, we might as well close vote first.
@Eric The numbers aren't as discrete as they appear. 0 is the starting level, but it's definitely not the lowest.
 
Ah.
Well, I could add this:
if (close is an option and all hearts are empty)
    close
else if (close was cast last time or hearts are all empty)
    flag
else
    existing logic
 
0
Q: Need help finding this anime fast about a girl in debt

user13564This anime was the first anime I watched and i forgot the name of it D: It is about a girl who is in a lot of debt after her dad leaves and she stumbled upon a guy in glasses being chased by a dog she helps him and he kisses her on the forehead gives her a paper of his home and says you can sta...

 
8:17 PM
Do we know how many days it took last time?
 
About 283.
 
If so, I suggest doing 1 win with the modified logic and seeing if it makes a difference. I think the difference will be big (20+ days) so we'll notice.
 
Okay, I'll modify it now.
 
I don't think we should change the logic in the middle of this run, simply because we didn't flag at the start of public beta.
 
I think only 3 people are running the script so hopefully the two of us changing should override it.
Worst-case scenario, they win, and the strategy doesn't change.
 
8:19 PM
Go ahead and change it now, but I don't think we should switch strategies until next run.
That's just my opinion though.
 
I won't actually begin using it until next round.
 
ok, I'll do the same.
We should really get more than 3 people running this though. It's very nerve-racking.
 
We should have 5. I really don't know why there's only 3.
It should be Hakase, Unihedro, and nhahtdh in addition to you and me.
@LoganM Do we have any evidence that alternating during private beta is best?
 
Sorry, I don't have an account.
 
@LoganM Appears to be
 
8:28 PM
@Eric Took 18 days when we alternated and about 20 when we followed this strategy.
 
And from what I gather, codegolf reached the same copnclusion
 
@LoganM Ok, I'll change that too, then.
 
That could be random though.
Don't bother if it's only one or two days gained.
 
Alright.
I think PPCG is going to be around 260 this time.
@LoganM I've updated the script, so I'll swap it out on day 1 of the next Site.
But you can get the source now if you want.
 
Some other sites are winning in 270ish days. We should look into what they're doing.
I think it's at least partially random
 
8:37 PM
The flag/close will definitely help.
We're 5 days into graduation right now and we still have 0 users and 0 quality.
 
That isn't necessarily a problem.
 
Not a problem, per se, but it means we have to spend votes there when we might not have to with a flag.
 
Users drops rapidly when traffic is low, but at high traffic it's very stable.
traffic, in turn, is based on quality and answers (I think)
but answers drops quickly if questions is low, so that has to come first.
So our strategy is actually pretty good for an ad-hoc thing.
 
Yeah. PPCG has almost won.
My initial estimate of 260 was off.
274.
 
not sure but if every site's game lasts the same amount of time, but if it's true, we started earlier than other sites and our only opponent is golf
 
8:41 PM
I think Blender is fastest so far. 256
 
They use flags.
(Average 1.3 left after 3 Sites.)
 
oh wait flags means traffic also gets 1 heart
 
Their first run was terrible. Only the second one was good.
 
I bet it passively increases other stats a bit
 
PPCG alternates in round 1 also.
 
8:42 PM
They downvote when all stats are full as a graduated site.
 
That's probably better than Q/A
quality always seems to drop faster than the other stats.
@Hakase does it?
 
idk but increasing Q or A increases users
must be other rules
 
Arqade is definitely using flags.
@Eric Yeah I think alternating round 1 is at least better than always picking question.
 
Okay. Then I'll change that.
 
8:45 PM
probably wastes 3 or 4 days.
 
I'll just make it pick the lowest stat.
 
18
Q: What is the optimal strategy for the launched site in Stack Egg?

PearsonArtPhotoThe Stack Overflow instance in particular seems pretty poor. How should it be done to achieve the best performance? How is there such a difference between different sites?

maybe there's something useful in there
 
@Hakase Yeah, that's what we use already
 
well, it only really matters if both are at 4 I think.
 
Well, they fill them up evenly, too.
Instead of leaving Answers empty until Questions is 4.
 
8:47 PM
I don't think that makes a huge difference.
(in private beta)
we could time it though
 
So you think just alternating Q/A, when both are full, in private beta, makes a difference?
 
yes
 
Alright
FWIW, PPCG's latest round took 17 days to get to public beta.
I don't know what our previous ones were.
 
It looks like what we're seeing is quantized data. We want to boost both stats to a certain level, but sometimes we waste time doing that by boosting a stat that's already high enough. At least that's what my impression is.
I wish we could reverse engineer this, but I don't have the tools to gather the amount of data I'd need to accomplish it
@Eric did you change it so that if all stats are max in graduated site mode it downvotes? That seems to be a big time-save, possibly 10-20 days.
 
I didn't. I'll do that.
Graduated mode only?
 
8:51 PM
Yes. Public beta optimal strategy is more-or-less unchanged afaict
 
Okay, here is what has been added:
if private beta and all full:
    choose answer if chose ask previously
else if close is an option, and all stats are empty:
    close vote
else if there are flags and all stats are empty:
    flag
else:
    old logic
Oh, and default action is downvote if in graduated mode.
 
ok let's boot it up
 
We're going to switch after this round, I think
 
say when it's up
 
I'm going to keep looking into things until we're within a month or so of winning
I want to see what happens if the stats go red
I was able to ask 5 times in a row with answers in the red in private beta before it got shut down.
So we're probably loosing a bit of time with our current strategy, but I'd rather not change it now that I know we can lose that way.
Better to lose a couple days each run than to lose a run 200 days in.
 
8:58 PM
ATM, PPCG is focusing on Users over both Quality and Answers.
 
really?
 
Yes. They voted for users when it had 3, Answers had 3, and Quality had 1.
 
they're ahead by about 1.5 wins so they can afford anything now
 
Nothing in their chat suggests that to be optimal.
 
@Eric can you update the script so I can set it up and go sleep
 
9:01 PM
They do claim that downvoting when all stats are full isn't great, but Blender did just that and pulled off a 256.
 
@Hakase Sure.
 
ok we gud
 
The new version is uploaded.
 
it's already running lol
oh
even newer
 
9:03 PM
Yup.
 
ok runnin
 
PPCG claims that a flag is worth about 10 days if used optimally. I don't know but that seems possible.
 
it's not working
 
We should switch one at a time to block any trolls.
 
@Hakase ?
 
9:05 PM
answer is not being selected
 
oh fuck
Uploaded fix
That was my test version that doesn't click, lol
 
ok it's working
 
Good
 
we have a troll present. Luckily just 1 right now
 
some random guy there, it bothers me
we should get backup
 
9:07 PM
Yes, 3 people is not enough.
 
@Abyx just in time!
 
I'm inclined to think that in the late game, quality may be even more important than questions and answers.
It drops so quickly compared to everything else.
 
I'm goin to sleep, so gonna be out for about 8 hours, maybe I'll wake up at night, see script update and update mine too
 
kk
@LoganM Agreed.
 
Wait, I thought we were just testing the flag+close vote version.
Are you really going to run that overnight?
 
9:12 PM
@Hakase uhm for what?
 
Our StackEgg needs you
 
sounds scary >_<
 
@LoganM Are we not testing all the proposed changes next round?
 
@Eric yes, we're testing, but that'll be like a couple hours.
 
cuz I want to bring in more people like @Frosteeze on this but if we're going to be testing and modifying actively we should wait on that.
 
9:14 PM
If flags/CV turns out to be slow and wastes flags, we'll want to get rid of it.
Don't make modifications in the middle of a run, that'll only screw up our data collection.
 
i have a feeling we're gonna get raided soon, man the vote cannons
 
Some people are already using the modified script
facepalm
 
I think Hakase is.
It shouldn't make a difference in the end if it's just him.
 
At least 2 people downvoted one time.
 
One might not be a bot. Unihedro for sure is not here, and the only other person with the URL is nhahtdh.
 
9:21 PM
Won't be a big difference (probably not even a day).
quality drops so fast x.x
 
Yeah, it's brutal.
Hopefully the modified strategy will fix that.
 
ok, can someone explain the April Fools Joke for me?
 
@Memor-X It isn't a joke, it's serious business.
 
super srs bizness
@Memor-X it's like tamagotchi, keep the SE thing alive and win an internet
we just need to watch for raids
 
9:24 PM
Confirmed just one person using new script @LoganM
 
yep
Quality is dropping every single day. It's really terrible.
 
Yeah.
 
luckily users seems extremely stable.
 
We're not going to make 274. Probably around 278.
 
@Memor-X if you need more explanation just ask. I'm currently watching for raids atm
 
9:26 PM
My guess is low 280s
Everyone get ready to switch to the new one
 
What the hell, it's not casting the restart vote this time.
 
We only need one person on that, so you can fix that during the wait time
so it should flag at the start of public beta, right?
 
I'm thinking it might not. Public Beta starts with 1 heart in answers and 1 heart in questions, doesn't it?
 
I forget. You might be right.
 
That's fine for flagging. Because it only flags if it can, which is at the start of each section (since you get 1 flag for each).
I'm more worried about close voting. It needs to only be cast once.
 
9:34 PM
Is it possible for it to CV more than once?
 
I guess I'll make that the logic. That if it's graduated, cast a close vote, unless one has already been cast.
 
The alternating doesn't seem to be working.
 
hey what about 2 close votes and then a flag
 
@Frosteeze so do we just click on the buttons that highlight?
 
says "at expense of questions" does it work with questions at 0?
 
9:35 PM
Yeah, the flag in Public Beta didn't work.
 
2 CV in a row is potentially dangerous.
 
@Memor-X we click on the buttons that will fulfill the needs of the SE on the right
 
Right here is where a flag would have been good
 
@Memor-X so if Questions are down, click ask. If Answers are down, click answer. If users are down, upvote. If quality is down, downvote
 
3 stats at 0, with 2 in the red.
 
9:37 PM
@Frosteeze so just keep clicking buttons?
 
@Memor-X nah just install script and go sleep :p
 
@Memor-X click the right buttons
 
dont click restart or flag unless you know what you're doing
 
don't just do it randomly.
 
so how about use flag once per round (public beta, then graduated) at 3 stats at 0
 
9:39 PM
@Hakase Since you're still here, we'll bother you to upgrade to the debugged version at some point :P
 
uh I'm kinda getting into bed
 
I'm just going to tell if that when we start graduation, cast a close vote. After that, flag.
And in public beta, to flag.
The first one
 
Did you fix the vote alternation in private beta?
 
I think 2 votes against 1 will work just fine :p
 
It didn't seem to work.
 
9:40 PM
Working on figuring out why.
 
You can test on an inactive site like writers.stackexchange.com
 
Found the bug.
 
ping me when update is up
 
New version is uploaded @Hakase @Frosteeze @LoganM
 
9:45 PM
@Eric I just realized that we have enough sites to work with here that we could reasonably brute-force search for the optimal strategy in private beta, and the optimal Markov-type strategy in public beta, in a matter of a few hours.
 
Even though some sites are inactive atm?
Like Writers.
 
Based on anime.stackexchange.com/stackegg/leaderboard, it looks like there are tons of inactive sites.
I don't think that's the right approach here, but it's probably within the realm of possibility.
 
well it flagged
 
i updated mine
 
when did it flag?
 
9:47 PM
you didn't set up a check for whether it's the beginning of the new stage did ya
 
@LoganM As soon as the new script went in. We wasted a day, my bad. But it won't be flagging randumbly anymore.
@Hakase No, because we only get them at the start of each stage.
 
More importantly than a day, we wasted a flag.
 
Did we? Only one is useful after graduation.
 
But our average flags is reported on the scoreboard.
 
possibly affects final place
 
9:49 PM
So we want to keep that number down
 
Oh, yeah. That.
Regardless it won't happen again. (It was only because of installing it in the middle of a stage.)
 
actually, do we know if flags are the tiebreaker or days?
 
So far days, I think.
 
it's days
 
Flags might be 2ndary to that.
 
9:50 PM
wins ,days, flags(maybe)
so no future updates?
I mean you can do without me
 
I need to make one quick patch. After that, no.
 
I think it takes into account both. Otherwise Code Review is way too high.
 
Okay, final patch is up @Hakase @LoganM @Frosteeze.
 
More flags definitely helps the stats. If it doesn't gain us a lot of time, I say scrap the flags.
 
We'll have to see if it helps.
 
9:53 PM
@Eric updated
 
updated here as well
 
dun
ok I go slip 4 realz nao
 
Cool.
Night @Hakase
 
nyte
 
Arqade is currently my testing ground because they use all their flags.
 
9:56 PM
Yeah I'm keeping an eye on them
 
@ton.yeung you've beaten Lightning Returns. when should i return to Luxerion to continue Vanille's and Fang's story. so far i've only met Vanille and learned about the Chaos building within the Cathedral but from what i remember from screenshots posted on facebook before the game's release Fang is supposed to turn up at some point
 

« first day (839 days earlier)      last day (3325 days later) »