« first day (839 days earlier)      last day (3313 days later) » 

10:06 PM
Code Golf's average is now significantly lower than ours.
We should probably borrow their strategy.
 
Arqade just scored 267.
 
Did our flag work as planned?
@Eric That isn't appreciably better than what PPCG has been getting.
 
Sort of. A red stat trumped it and shouldn't have. Other than that, yes. The close went as planned, and the flag was used.
@LoganM What was PPCG's last?
 
230ish :D
 
The issue here is I haven't noticed a consistent strategy. It seems very follow-the-leader-ish.
What did you guys change @Geobits?
 
10:12 PM
We figured Users out (mostly). See our chat for more, it's a bunch of back and forth.
 
235, and they didn't even have max stats.
 
@ton.yeung what about the Dead Dunes?
 
The strategy they are using is much more complex.
It's based on previous happenings and predictions on when things will happen next.
 
It's not that much more complex. I was already thinking in much the same way.
 
10:22 PM
We can outline a new strategy, then.
 
Okay, first, we need to rethink the order. Our current order is not very efficient.
 
Well, Users is by far the least important stat.
And it's the slowest decay.
It can be completely low priority.
 
Pretty much.
 
I don't know the relation between Q&A though.
Whether they're more important than quality or what.
And how important compared to each other.
 
One thing I am sure of is that it's extremely inefficient for our strategy when all are maxed to be just to pick one and always upvote that.
 
10:25 PM
Yeah, we need a way to know what's going to go down next.
But I don't know what causes that.
 
Just watch. You get a good feel for it after a dozen or so turns.
Between Questions, Answers, and Downvotes, I'd say to pick the one that we haven't picked for the longest.
Maybe one of them is more important than the others though.
Downvotes feels a bit more important in public beta, but I can't say.
 
Except for this new decay idea, I think our strategy is optimal for public beta already.
It has no traffic influence, which seems to be important.
 
Users is definitely not important at high traffic levels.
 
Indeed. It's more than possible to let it decay several times.
 
Questions also doesn't decay much.
@Eric In public beta we can avoid upvoting except when we hit a warning.
 
10:30 PM
That means we can just leave users at 0 or warning?
 
And then, wait 2 turns before actually upvoting.
@Eric It'll go up to 1 once we upvote.
 
Yeah, but we don't go up any higher?
Even if other things are full?
 
Eventually, we probably want to. I don't know when though.
 
Okay. I'm going to start writing a second script that won't do any selecting yet.
And start tracking decays and such.
 
The other thing is that voting for the same thing multiple times seems to lower the effect.
 
10:36 PM
The decay effect?
 
0
Q: Looking for an anime around this description

HelpMeMasterGuy is in hospital and needs a girl to kiss him or else he is forever a ghost. He will become one when his entire body is glowing, but he slowly starts to glow from the feet up.

 
Okay, Questions seems to decay at a consistent rate of 1 every 2 rounds.
 
You're using decay to mean 2 different things
 
I'm using it to mean decreases in stats that aren't caused by raising others.
 
I was talking about the effect on the progress bar.
 
10:46 PM
Ah, that's why I asked.
@LoganM Our current strategy is working well.
It's not as good as 235, but it's much better than our previous one.
I think if we neglect users more, we will have it solidly.
 
I think we can get rid of the flag
definitely the close vote
 
Why? I think the flag helped. Close vote, yeah.
 
did nothing helpful in the long term
Our immediate goal has to be to boost traffic. Once traffic is boosted, questions and users are stable.
 
One thing we can do, though
Is after questions goes down, wait a turn to vote it up
Because it's not going anywhere in the meantime
 
10:49 PM
Yes, that's fine
 
Answers is a bit more volatile, but seems to be about every 3 rounds.
 
Don't vote for a stat that just dropped, except near the end if it's quality.
 
0: 2
1: 1
2: 2
3: 3
4: 3
5: 3
6: 2
7: 3
8: 3
9: 3
10: 4
11: 4
12: 3
13: 3
14: 4
15: 4
16: 4
17: 3
18: 3
19: 3
20: 4
21: 4
22: 3
23: 4
24: 4
25: 4
26: 4
27: 3
28: 3
29: 4
30: 4
31: 3
 
we have a restart troll
 
PPCG's Answers history ^
 
10:50 PM
We don't seem to be downvoting
(when all stats maxed)
 
Eh? I think we are. I'll watch more closely.
 
2 people were, but 2 were still asking
In the graduated phase, we should initially focus on Q + A I think. Once those are high (3+ for each), start boosting quality. At that point, traffic will rise, and then users and questions will stabilize.
 
Without any votes to users?
 
Upvote users once the other stats are all high
We won't need to do it a lot
 
It's still okay to not go above 1?
 
10:55 PM
Right now, we just had 2 DV/ 2 ask
some people are still using the old version or a bugged version of the code.
 
Hmm
That's really weird.
 
The question is, does users even do anything?
 
I know the previous version had a bug that would keep using Ask instead of Downvote.
But everyone said they updated.
@LoganM I don't think it does.
That was a 3/1 split.
 
now a close vote?
 
The hell?
 
10:57 PM
The 3/1 is because I manually closed my script and started voting on my own.
It isn't working for me.
Just upvoted. Almost 0 progress to the bar.
It seems to be pretty useless most of the time.
 
kk. Let's leave users at <= 1 always.
 
I think if all the other stats have been upvoted in succession since reaching 4, users is probably worth clicking.
 
Wouldn't it be better to upvote something like Quality at that point?
 
I think we do need users, but probably only once.
Maybe once we hit full traffic, upvote users. Not before then.
that way it will stick
Do you have a new version ready? We're going to win soon
flag did almost nothing
 
Not really, only with minor improvements.
 
11:02 PM
One person is still using an old version
Does it still not click restart?
 
It doesn't. I have no idea why. It picks a blank string as its upvote target.
there, found that bug.
 
We should really go to the better strategy asap. We're wasting over 10% of the days each run.
 
I know, I'm working as fast as I can. But I have to make sure it's perfect because it's hard to test.
Arqade also uses only their first flag.
I don't see much point as you enter public beta with 1Q/1A anyway.
 
You can test it on literally any site
 
@LoganM Yeah but there's different logic per stage and it also takes 20 seconds per round. :P
 
11:05 PM
At the very least, get rid of the flag and CV. They didn't do anything.
 
kk, both are removed
 
@ton.yeung ahhh ok....makes sense if she was tied with the main quest there since so far every other main quest has had a returning character involved. probably also a stroke of luck that the Dead Dunes is the last on my list. was kinda worried i'd miss out on saving Fang and Vanille given that nthe first meeting of Vanille was a side quest
 
I also make it allow stats below 0 for 3 turns
i.e. "warnings"
 
3 turns is pushing it. I'd say 2 turns is the right number of times to ignore
 
Apparently it's a consistent 5 that fails you.
I can set it to 2 though.
 
11:07 PM
That's only in private beta
No one has tested public beta afaik
Also, does it handle simultaneous warnings accurately?
 
kk. I'll leave it as 3 for private beta and 2 for public.
@LoganM What do you mean?
Define "accurately"
 
@Eric If there are two warnings that appear in the same turn, you can't wait both of them out the maximum length
 
True enough.
So the "maximum length" would be 3 minus the number of warnings.
 
ppl flagging x.x
Did no one even update?
 
11:12 PM
PPCG is getting heavily raided right now.
 
Literally 1 vote from a restart
 
Yup.
 
This is way more amusing to watch than ours
 
Indeed. Distracting as hell.
 
One Piece Pirate Warriors 3 Limited Edition is avliable for pre-order, released in August. Has anyone played the series?
 
11:23 PM
We're starting to get someone voting restart here
Not sure if it's the same folks spamming PPCG or PPCG users trying to retaliate on us for something we have nothing to do with, but either way, don't let it through.
 
It won't get through. Even if it gets to the 2nd vote, we'll all vote against.
 
Did you test your script for that?
 
Yeah.
 
ok good
 
I love what's happened to Nero for Type-Moon's April Fools day
"In Type-Moon's April Fools' Day 2015, she is the most popular artist in the Imperial Roma talent agency. She debuted in the 22nd July 2010 and she is known as the Red Saber or Nero. Her class is Idol Emperor and she is from Rome. She says she can do anything as her Skills and Hobbies. Her Tmitter user account is @tmtt15_nero"
 
11:28 PM
Looks like PPCG will survive the restarts and finish in under 250
 
250, yeah.
They came really close to dying.
 
Yeah, on the last day it was tied.
 
Yup.
Well, they got their restart on day 1.
 
Much better than on day 250
 
Yeah.
Worth noting is that at least one is a bot.
 
11:31 PM
day 1 restarts are hilarious
In the meantime ours is going strong.
@Eric did you update to the one that doesn't CV and flag?
 
Not yet, since we're only at day 93.
I've been tweaking it further.
I'm running the previously uploaded version though.
 
Can you do that before we graduate?
Wasting flags is really bad for our leaderboard position.
 
I'll make a quick version that changes nothing except removes that.
 
ok thx
 
@LoganM Done and uploaded.
 
11:39 PM
okay anyone else using the old version should update (c.c. @Frosteeze)
 
Hopefully we break the tie if he doesn't update in time.
I don't know how ties are determined.
 
we have 5 users voting right now, so I don't think we will :(
@Eric random
(afaict)
Is it just me or is our beta lasting longer than usual?
120 seems really long
 
It is a bit long.
Should be around 110.
 
Yeah the flag does nothing initially
 
I think I'll make that the final version, though. I need to get to the stuff I was supposed to be doing all day.
 
11:45 PM
you guys have some kind of bot?
 
How is this not the most important thing all day?
 
@LoganM It was the most important. That's why I did it instead of everything else. So now I need to cram the other things.
The important thing is we're not going to end up changing positions.
 
We're wasting like 40 days every run with this approach though.
Someone is still using a horribly out-of-date version and flagging
I guess it will fix itself so long as we have enough other people.
@Gerret Yes. It isn't perfect though.
 
@lo
@LoganM where do i find it?
 
11:49 PM
The problem is that modifying the approach will take a lot of time. Probably an hour or so to get all the logic of the decays and warnings.
 
Give @Eric an email to send it to. We aren't posting the url here because other sites could use it.
 
@Eric would be nice if i could get it -> gerret.shoemaker@gmail.com
 
Sent.
Do you know how to run JS in your browser?
 
Now we have a flag-bot :(
 
@Eric ill find out in a secound im a programmer
 
11:52 PM
Yeah. As long as there are flags available someone will try. :\
5 hours ago, by Eric
I have no idea if that would work. I use CJS. (Click toolbar button, insert script, click the checkbox at the top right, and click save.)
 
Looks like PPCG abandoned their thing. But Arqade is going faster than we are based on recent runs.
Maybe we should dump the bots and do it manually.
(at least until none of us can stay online)
 
If we have a sound strategy, sure.
 
Wow, for being so big, SO's Egg is really undermanaged.
 

« first day (839 days earlier)      last day (3313 days later) »