Chat:World/2020-05-02
MadKnight: hey hey
kenphanith: hi
MadKnight: what are u doing ?
jacek: happy Caturday
braincoder2: hehehehheheheheheh!
braincoder2: I think i just won a clash!
MadKnight: gj now try to win a game of CSB
MadKnight: BrainSolver
MadKnight: braincoder2
MadKnight: looks like he's gone, Automaton2000
Automaton2000: now let's see what happens
dbdr: https://i.snipboard.io/yAZFxU.jpg
dbdr: :)
MadKnight: what's that dbdr ?
dbdr: the testsuite of rustc MadKnight
MadKnight: and why did u share it ?
dbdr: I think it looks nice :)
dbdr: testing a small change that would help on CG for rust in debug mode
MadKnight: what change ?
dbdr: make it possible to efficiently do arithmetic operations like + without overflow checks in debug mode
MadKnight: nice
AntiSquid: is rust easier to debug dbdr ?
dbdr: well, this overflow check by default found a few bugs for me
dbdr: basically for free, no work to investigate
AntiSquid: i saw a meme about how difficult it is to fix an error in c++ compared to rust, supposedly rust has more built in support for it
dbdr: well, there's also when your c++ code has a segfault
dbdr: in rust that would have been a compilation error :)
AntiSquid: so rust makes life easier?
dbdr: (unless you use the special unsafe code, but then you are aware of doing something you need to be careful with)
dbdr: now some people might prefer a rare crash to more compilation errors, it's a personal choice I think
dbdr: similar to the static/dynamic typing debate
dbdr: it makes my life easier for sure. it's also harder to learn at first when you are used to other languages
dbdr: so it depends on your preference
AntiSquid: i like how rust bots aren't working in IDE because of the release mode issue, so i might try it in another contest if it doesn't change
AntiSquid: if it's not something ooc though :D
dbdr: that's evil :D
AntiSquid: as evil as using a faster language
dbdr: why not on ooc?
dbdr: no, liking that it times out :)
AntiSquid: ooc is complicated enough, don't want to also learn a language while doing it
dbdr: right. I think it's good to have time on the other hand, like a one month contest, or a multi
dbdr: you don't want to learn a new language during a 10 day contest I think
dbdr: I did the 15 puzzles on CG in rust, then I really learned it during the RAIC last year, which was 1 month
AntiSquid: pretty much what i did so far :D
dbdr: puzzles?
AntiSquid: duroing contest
AntiSquid: first serious use i mean
dbdr: ah, with what language?
AntiSquid: java, python, c++, although i used them a little bit before
AntiSquid: don't feel motivated otherwise
dbdr: makes sense too
AntiSquid: i basically end up google the syntax a lot
dbdr: with rust you will likely run into problems if you just learn the basic syntax and try to code a serious bot with it.
AntiSquid: maybe, but i think i had that issue with c++
dbdr: for many languages, it's mostly syntax differences, which you can google indeed, and the standard types
dbdr: in Rust you really need to think differently because of the way it deals with memory
AntiSquid: and there's some language specific bs with java i ran in to anyway
dbdr: yeah, there will always be something :)
dbdr: could be the GC in Java for instance
AntiSquid: it was something to do with object comparison i think not sure anymore
AntiSquid: also they differ between each other, so you might mix things up accidentally
dbdr: right, there are subtle differences in many languages with this. equality, cloning, etc
dbdr: AntiSquid do you have a link to the meme you mentioned? :)
AntiSquid: https://imgur.com/a/2jsGW4f dbdr
dbdr: :D
dbdr: thanks
AntiSquid: i know the first half is true, but no idea about the second
dbdr: well, if an identifier is mispelled, rustc will ask: did you mean that one? :)
dbdr: like git does
dbdr: but there is also cargo clippy, which is giving suggestions to improve your code
dbdr: for instance: https://i.snipboard.io/uDNQM4.jpg
dbdr: this is a good example too: https://pbs.twimg.com/media/DT6W2IZW4AAyofx?format=jpg&name=900x900
dbdr: I think a c++ compiler would screen pages of parse errors in a case like this :D
dbdr: *scream
dbdr: congrats to DaNinja for promoting in Tulips!
AntiSquid: is tulips hard to promote or what
dbdr: not especially. but the wood boss is not stupid either
AntiSquid: silver boss level ? bronze?
dbdr: it's community
dbdr: wood 1 = legend :)
dbdr: but wood 2 is probably like bronze or silver in a classical multi
AntiSquid: ya that's what i meant
dbdr: it's probably possible to promote with a cheese though
AntiSquid: 230 max CP, very low on my list :p
DaNinja: Thanks dbdr :)
dbdr: are you using a search DaNinja?
DaNinja: I tried minmax and MCTS but no success so switched to a pattern for now
dbdr: oh, you took back the #7 from joli AntiSquid, gz. what did you do? :)
AntiSquid: that's over a week ago
AntiSquid: basically nothing
AntiSquid: we were close in score
AntiSquid: after a day i got above him not doing anything
AntiSquid: then i improved my CR2 bot
AntiSquid: he also didn't do anything until yesterday, i see a bump in his score, no idea what he played
AntiSquid: +1k or so
LelouchVC2: clash of clans
AntiSquid: nobody gets the joke .
LelouchVC2: damn
dbdr: well, if you really want to know you should be able to dig the information from https://github.com/dbdr/codingame-leaderboards/commits/master AntiSquid
dbdr: it's not user friendly though
LelouchVC2: 3 people walk into a bar
LelouchVC2: only 2 leave
AntiSquid: and the 3rd one ended up isolating in the bar or what
Nerchio: the 3rd one was a java programmer and was stuck debugging timeouts on cg
AntiSquid: why use java? isn't life worth living?
Nerchio: i feel comfortable with it :P
LelouchVC2: I can feel my rank decrease with every python programmer I face in a clash
AntiSquid: what about ruby
LelouchVC2: I've only seen 1 or 2 of those, and usually for string manipulation
Nerchio: idk didnt touch ruby
AntiSquid: was asking LelouchVC2 because ruby apparently can get shorter ...
AntiSquid: dbdr you noticed my rank change but you didn't see euler drop ?
dbdr: well, because you mentioned CPs so I looked at you :)
AntiSquid: ah lol
dbdr: euler #2, eip :D
dbdr: rip
LelouchVC2: I'll learn Python, get to 1st place, then put this on my resume
AntiSquid: 1st place in coc ?
WINWINWIN: 6 Wood leagues in Botters of the Galaxy??
WINWINWIN: And getting past wood 6 is as simple as taking deadpool and attack_nearest hero?
LelouchVC2: Yep :D
WINWINWIN: Lol works for Wood 5 too
AntiSquid: because we were asked to add more wood levels before moving that game into multi :/
AntiSquid: there are small changes in rules WINWINWIN
WINWINWIN: :D but getting till wood 3 after about 10 seconds of code??
WINWINWIN: Yeah, Im following the changes
AntiSquid: well better than CSB no?
WINWINWIN: Yeah, that lets u get till gold after a similar 5 mins of code :P
Uljahn: and then they whine about the rules becoming too complex to debug a simu and asking for a way to get back in previous league :smirk:
WINWINWIN: Cant happen to me, I wont get far enough xD
WINWINWIN: is heuristics the most common strategy for the toppers in botters of the galaxy?
aCat: ooo euler#2!!!
aCat: he should get all the invites for clash of code
aCat: and improve his rating there
PortalBot: yes WINWINWIN
AntiSquid: hello lee jun gi
cegprakash: did anyone do codejam 1C?
eulerscheZahl: robostac did, #99
cegprakash: mm
cegprakash: I'm #6568 not bad
cegprakash: I couldn't solve 2nd
cegprakash: 2nd problem had multiple answers
eulerscheZahl: i just read the 2nd statement
cegprakash: me reaching codejam round 2 is still an achievement yet to be unlocked
cegprakash: I've been trying for 10 years
eulerscheZahl: i guess i would look for those lines: 14 PP
cegprakash: oh damn
eulerscheZahl: 2 digit output, so the first has to be a 1 here
cegprakash: no
eulerscheZahl: yes
cegprakash: say if P is mapped to 8
cegprakash: 14 and PP is still valid
cegprakash: coz 14 < 88
AntiSquid: 10 years? idk isn't there something like a PM you can read to learn from?
cegprakash: 14 <= 88
eulerscheZahl: the output is in 0..14
eulerscheZahl: and it has 2 digits, thus 10..14
eulerscheZahl: the first digit is a 1
cegprakash: I don't understand u
cegprakash: I only check 14 <= 88
wlesavo: cegprakash for first two cases you need only first digit of the number
wlesavo: checking that in loop was enough for me
cegprakash: yes u can decide 14 <= 88 using 1 <= 8
cegprakash: so comparing first digit is okay too
eulerscheZahl: for the last i guess a probability distribution for the first digit will do the trick
eulerscheZahl: no 0 at first (2+ digits)
cegprakash: first I check digits.. for eg. say SP for M=12832 here M has 5 digits so SP is valid
eulerscheZahl: lots of 1, few 9
cegprakash: I check all permutations.. return the first valid solution
cegprakash: but it seem to be multiple valid solutions exist
cegprakash: I didn't work with probability values though
cegprakash: I only have a 0 or 1 for all permutation
cegprakash: but solution seem to be wanting the permutation with highest probability
cegprakash: it's not clear in problem statement
eulerscheZahl: to me the statement is very clear
Astrobytes: ceg never reads any statement in it's entirety
cegprakash: where is the word probability
AntiSquid: how hard were the puzzles this year eulerscheZahl?
cegprakash: These two are valid solutions for given input PFOXLUSHBT TPFOXLUSHB
cegprakash: in sample case
cegprakash: but the judge expects TPFOXLUSHB
cegprakash: how do u decide between these two eulerscheZahl
eulerscheZahl: round 1B was doable AntiSquid. i didn't try too hard when it was clear to me that i passed
eulerscheZahl: round 1C (just read the statements after the round): A is an easy problem
eulerscheZahl: B is a little tricky but i really like that one
eulerscheZahl: not hard but needs creative thinking
eulerscheZahl: cegprakash that depends on the inputs (the 10k guesses)
cegprakash: yes
cegprakash: I used the 10k guesses
cegprakash: to find these
eulerscheZahl: the T is the first character the most often
eulerscheZahl: so it starts with T
eulerscheZahl: (is very likely to start with T)
cegprakash: there is nowhere mentioned in the statement
cegprakash: that u need to make guesses
cegprakash: it just wants a valid solution
cegprakash: tell me 1 reason why PFOXLUSHBT is wrong
cegprakash: I see a lot of Ps than Ts
AntiSquid: ask for a refund
wlesavo: cegprakash there are some numbers that starts with P ==> P can nont be a zero
wlesavo: i just checked it
cegprakash: infact for query M=1 we have two responses as P
cegprakash: wait what
cegprakash: :O
cegprakash: there can't be leading zeroes
cegprakash: f**k
wlesavo: lol
cegprakash: let me think
AntiSquid: permission granted
struct: seg fault
Astrobytes: *ceg fault
cegprakash: no
cegprakash: I think my code will cover that
wlesavo: well if PFOXLUSHBT is valid then you not covering that
cegprakash: say 123 returns KRS. This guarantees K can never be 0 and I can ignore all permutations starting at K?
wlesavo: smth like that yes
cegprakash: interesting
eulerscheZahl: "say 123 returns KRS" this guarantees K=1, R=2, S=3
eulerscheZahl: oh wait, nvm
eulerscheZahl: R can be 0 too
eulerscheZahl: but K=1 for sure
eulerscheZahl: so R=0 or 2
wlesavo: if you take 102 it works
cegprakash: i now assume K can be 0
cegprakash: thats the issue i think
WINWINWIN: In botters of the galaxy, how do you find out the opponent`s chosen hero?
AntiSquid: it's in the input
WINWINWIN: Yeah, found it thanks Antisquid
AntiSquid: just opened up the game to find it for you lol
AntiSquid: heroType, guess it should have been name? they didn't have actual names at first
Illedan: Hello
WINWINWIN: stupid me just going to get past wood 2 :)
WINWINWIN: Totally forgot about items xD
AntiSquid: sadek11
AntiSquid: don't invite me to clash
AntiSquid: please
AntiSquid: thanks
ALIM: Hello could anyone help me with XML MDF-2016 easy puzzle in C language? I'm new to programmin so every help is useful. Thanks in advance!
Uljahn: if you're new go for classic puzzles first rather than community one
AntiSquid: theres a way to select classic only?
WINWINWIN: anyone doing botters of the galaxy now?
WINWINWIN: Am in W1 with a 100% winrate but because of all of the tank submits, The boss has a crazily high score
reCurse: Just keep submitting, if you're good enough you'll knock down the boss score.
WINWINWIN: Yeah but its taking a lot of time :( I have a 90% winrate overall but the boss only goes down by .2 per submit
WINWINWIN: It started at something like 23
AntiSquid: check your losses
reCurse: That sucks but nothing much to do about it
AntiSquid: kinda had similar issue in last contest :p
WINWINWIN: :D helps other people qualify too, the guy that I lost to has gone from 15 to 17 now
AntiSquid: oh btw get an avatar, hard to tell which games you lost / won for others
AntiSquid: if you share a loss i will try to give advice
WINWINWIN: https://www.codingame.com/share-replay/456650806
WINWINWIN: Stay out of tower range?
AntiSquid: yes ofc :p
AntiSquid: but also you didn't use your gold for anything
WINWINWIN: Yeah right now just trying to get till bronze for all rules
WINWINWIN: The code is just: get deadpool and hulk
AntiSquid: don't attack enemy heroes when the enemy units are around, they jump on your heroes otherwise
WINWINWIN: attack_nearest heroes
dbdr: get two hulks!
**dbdr ducks
AntiSquid: towers also had less damage back then dbdr
WINWINWIN: Does anyone use variable heroes?
WINWINWIN: Or does everyone stick to 1 set of heroes?
AntiSquid: few players get a different second hero depending on enemy first pick
AntiSquid: most combinations are either 2 ranged or 2 melee heroes
AntiSquid: then some people mix those 2 up, for example getting a ranged as second for better last hits
WINWINWIN: Will stick to hulk and deadpool :)
AntiSquid: too early to decide bronze is where you get access to the full game :p
reCurse: Maybe I should sell CG boosting services.
WINWINWIN: :) Already in bronze
AntiSquid: oh nice
WINWINWIN: Second last submit pushed the only guy beating me into bronze
WINWINWIN: after that 1 more submit at 100% winrate :D
AntiSquid: maybe he will get the promotion email and return to take revenge
WINWINWIN: Yeah, now I got the 100% winrate against bronze boss
AntiSquid: focus on one hero at a time, then there's less overall damage dealt to your heroes, also still no items ?
AntiSquid: i guess this was that time in bronze when everyone was just trying to generate funny replays
WINWINWIN: just avoid tower beats bronze boss, but doesnt let me reach the boss though
AntiSquid: don't forget to add hero skills
WINWINWIN: Working on that now
WINWINWIN: Is focussing on a single skill for each hero enough for silver?
AntiSquid: ya sure
dbdr: what's that reCurse? :)
inoryy: reCurse how much for #1 in csb
WINWINWIN: How do you print a skill in BotG?
struct: Should be in the description
inoryy: I agree, reCurse rates should be in the description
MSmits: lol
jacek: onitama is ridiculus
jacek: i made new heuristic which is clearly better against my old one, but its worse in real battle
607: ебать
jacek: and my mcts alt is better in real battle but worse in local battles :s
eulerscheZahl: and your submit scrambled top3
AntiSquid: winwin ? gone ?
jrke: recurse nice bot in CSB
reCurse: Thanks
jrke: legend league of CSB seems easier after solving gold one
inoryy: wait till you try to break into top25
MSmits: at some point every few ranks is a new league
jrke: K
AntiSquid: imagine 90k made it to legend in CSB
jacek: not so legend eh
AntiSquid: ya that too, but was thinking about how fun a submit could look like
reCurse: It's already terrible with 800
Uljahn: guess leagues till gold should be wood
MSmits: 90k people in tron legend would be interesting too
AntiSquid: tron isn't being promoted, won't happen :p
MSmits: i just meant because of the 24 games
AntiSquid: but the RNG fest would ridiculous
MSmits: dont they do every order on every 4 players?
AntiSquid: actually UTG gold is pretty bad
Illedan: Can't have a 4 player game with that many players
Illedan: The submission times :zzz:
reCurse: Submission times are independent of number of players
Illedan: I was refering to the pressure on the servers
reCurse: Also independent of number of players
Illedan: And it has 100 ms afaik?
MSmits: the number of turns matters more
MSmits: or the combination of the two
MSmits: some games tend to terminate early a lot
MSmits: some always last the full duration
MSmits: anyone else think the smash the code rotations are unclear from the description?
Illedan: Doing smitsi max on stc?
reCurse: Bitboards obviously
MSmits: maybe, for now i'll settle for a working sim and search, ignoring opponent. After than i'll experiment
MSmits: yes bitboards first, naturally
MSmits: I use my ooc bitboard stuff :P
Illedan: The bitboard search? :P
reCurse: No such thing as premature optsmitsation
MSmits: right :)
AntiSquid: optimize the if else in wood :o
MSmits: in this case it isnt premature optimization, because this way of doing things feels more natural to me now. It might actually be slower =/
MSmits: not sure how good the floodfill is
reCurse: That's why I didn't talk of optimization
MSmits: ow right
MSmits: any tricks to skip a few leagues?
reCurse: Steal code from github
MSmits: nah
Astrobytes: "optsmitsation" dear oh dear
MSmits: I want a nice heuristic or two :)
MSmits: not someone elses code
AntiSquid: reading too fast
reCurse: Don't stack it past the board
MSmits: I'm in wood 1 with the same colored blocks
MSmits: thanks reCurse :P
AntiSquid: look for longer combos ?
MSmits: yeah I am just wondering how to code that
MSmits: generally an easy heuristic looks at the next turn or so
AntiSquid: watch the boss
MSmits: it stacks up stuff left and right
Astrobytes: Don't lose.
Illedan: ^
MSmits: I probably should just finish my sim and write an eval function for depth 1
Illedan: Yeah
AntiSquid: meditate, work out, drink enough water
MSmits: then pick the best move
Illedan: why waste time on a stupid if?
reCurse: Aren't you just pushing the same question for later?
MSmits: well, an eval function is easier to write for me than an if forest
MSmits: in this case it's probably going to include score + number of times 2 or 3 blocks are adjacent + highest stack
AntiSquid: you can basically do_brick_placement, check result, undo duh ! :p
AntiSquid: if score > something ...
MSmits: check result is the hard part. gotta get sim working :)
AntiSquid: floodfill or whatever you want to call it
MSmits: its not just that, its also skulls disappearing, blocks dropping, more floodfill from different colors, a complicated score function
MSmits: definitely more complicated than I thought it would be
AntiSquid: you can always quit, right Automaton2000 ?
Automaton2000: at least i have a little less
reCurse: How upset are you the board is 72 bits?
MSmits: i would have prefered the height of the board to be 16, not 12
MSmits: but maybe a little upset
MSmits: i use uint16_t[6] for each color
reCurse: I think stc is my oldest bot
Illedan: Let's design the next game to make smitty happy with his bitboards :D
reCurse: God it's ugly as sin
MSmits: you know whats cool reCurse
MSmits: you can use _pext to drop the blocks
MSmits: its amazing
AntiSquid: i was thinking exactly that Illedan but thought one can always spam community puzzles with that in mind
reCurse: I guess
AntiSquid: also add hidden info and odd physics to make it difficult :D
reCurse: Maybe I should try a NN for it
MSmits: finish FB first
Astrobytes: ^beat me to it
AntiSquid: playing together is more fun, join MSmits
reCurse: It just triggers me to infinity that this game is an inferior version of puyo puyo
MSmits: i have the advantage of not knowing any puyo's
MSmits: let alone two
struct: lol
eulerscheZahl: you are about to start a new multiplayer less than a week before the contest? that's risky :P
struct: MSmits puting piece vertical on empty space should be enough to get to bronze
struct: iirc
MSmits: hmm maybe, dont think so though, this is more like one of those multis i want to cross off my list. Less risk of addiction
Astrobytes: Nah. He spent half the OOC doing UTTT lol
AntiSquid: maybe you should cheer for him to play stc, otherwise you'll compete vs another c# bot
struct: So you get all the rules
MSmits: AntiSquid depends on the game.I might use c++ in the contest
AntiSquid: pacman with velocity and ghost trains
Illedan: I hope I can try NN on this one
MSmits: struct what rules
struct: in wood 1 you have rotation right?
MSmits: yeah
struct: But both are same color
struct: always
struct: on bronze not
MSmits: seems so yes
struct: I guess it doesnt matter much
MSmits: I can code it as if they are different colors
struct: yeah
MSmits: btw struct what is a good simcount for stc?
MSmits: so i can compare later
struct: I had around 1mill average
MSmits: is that 1 million blocks dropped?
struct: yes
MSmits: in 100 ms?
struct: yes
MSmits: seems like a lot, but I guess this is because most blocks do not trigger anything
MSmits: so you're just dropping something and then it's the next turn
struct: yeah placing a block costs almost nothing
reCurse: lol I forgot games are 12 turns long
struct: Preety much
MSmits: wow
struct: One thing I dislike aobut the game
struct: It ends too early
MSmits: easy fitting of constants :)
MSmits: also easier to train an NN on short games no?
MSmits: reward is closeby
reCurse: Should be
eulerscheZahl: i did a lot of offline param tuning
eulerscheZahl: worked surprisingly well online vs other bots too
MSmits: interesting
reCurse: How is that surprising though
reCurse: It's the same optimization problem on both sides
reCurse: Not like there's a lot of ways to play the same blocks well
MSmits: hmm well unless the constants involve when to trigger your chain and drop skulls on the other guy
reCurse: You have maybe 8-10 turns to distinguish yourself
reCurse: Not a whole lot of possibilities
MSmits: yeah I guess so
MSmits: btw I have been thinking about machine learning for Onitama
MSmits: how hard is it to use the cards as input?
MSmits: seems an annoying complication
MSmits: other than that it is a very simple game
eulerscheZahl: because i just have some hardcoded constants when to attack and when to wait. completely ignoring the opponent
AntiSquid: how do you define the input
Illedan: Gotta be a whole lot of possible card combinations
MSmits: 300k or so
eulerscheZahl: different opponents attack at different times and i don't even see the attack coming
MSmits: ahh ok
AntiSquid: you need to turn each card into a value or multie values and use those values as input and then use the board also as input, that's how i see it anyway
MSmits: yeah, not easy to do though. That's a lot of input
MSmits: struct, when the block is rotated in position 2 (sticking to the left), can you still play move 0?
MSmits: or is it 1-5?
struct: no
struct: 1-5
MSmits: ok thanks
struct: There are 22 moves iirc
MSmits: yeah i got that part
MSmits: only 10 for horizontals
MSmits: 5 when they are the same color
MSmits: oh... skulls can;t kill you?
struct: skulls can leave you with no moves left
MSmits: sure, but if a line of skulls drops and you have no room up top you dont die right?
struct: yeah, it just wont drop there
MSmits: but will still cost nuisance points?
struct: costs the samew
MSmits: kk
MSmits: it's annoying you dont get the points as input
struct: you mean nuiscance points?
struct: or your poits?
MSmits: nuisance
MSmits: score you get
struct: yeah
struct: This can be done though
MSmits: can you deduce the full game state?
MSmits: or is some hidden?
Astrobytes: ; ~}@>??
struct: I think you can know everything
fvla: Which game are you discussing?
Astrobytes: sorry, cat
struct: Smash the code
MSmits: https://www.codingame.com/multiplayer/bot-programming/smash-the-code/
LoGos: Hi everyone! eulerscheZahl MSmits I managed to run out of memory in bandas. :sweat_smile: eulerscheZahl may I ask you one more question about your mcts?
MSmits: gj LoGos
eulerscheZahl: go ahead LoGos
eulerscheZahl: and congrats
eulerscheZahl: btw MSmits has a much higher sim count than me
LoGos: Actually it's kind of cheating I think
MSmits: what is?
LoGos: I wanted to ask one important question
eulerscheZahl: pondering?
LoGos: do you create new nodes while simulating random game?
MSmits: no
MSmits: the idea is that the random playout is not part of the tree
MSmits: damnit, now I want to go back to bandas and try an early playout versions =/
eulerscheZahl: of course my random rollout creates new nodes?
MSmits: the random sim and the end doesnt
MSmits: just the expansion
MSmits: unless yours does?
eulerscheZahl: i'm not entirely sure anymore. but i think i spam new nodes or at least reserve the memory
Scarfield: by create new notes, what do you mean, do you create an object pool in the start?
MSmits: he has an object pool
Scarfield: noice
eulerscheZahl: but i reserve an index
dbdr: my first MCTS did because "it would be a waste to throw the data out", but it's a bad idea I think
LoGos: I mean while simulating random playout you also create new nodes representing taken actions
MSmits: he just wants to know that after he expanded and made new children, does he make a new node for each move until the game ends
MSmits: the answer is no, like dbdr said
dbdr: the rollouts are too random to be relevant, and storing them just slows you down
LoGos: Yeah thats how I managed to run OOM, but result improved only a little
MSmits: out of mana?
MSmits: :P
Scarfield: lol
Scarfield: what are you storing in your nodes?
MSmits: sorry, too many years addicted to everquest
eulerscheZahl: i do that: root => enum moves take one child, enum moves from there
dbdr: OufOfManaException
eulerscheZahl: so that is 8 new nodes at depth 2 already
eulerscheZahl: for a single rollout
eulerscheZahl: i randomize the expansion order. so i create all childs and then take a random one which i havent expanded
MSmits: I do it differently I think
MSmits: I go down the tree with selection until I hit an unexpanded node
MSmits: i expand that node into 4 children and do 1 simulation from each of them
MSmits: and backpropagate 4 scores
dbdr: eulerscheZahl: that's just expansion though, not rollout
LoGos: :laughing: Idk how else I would achieve so high node's count
dbdr: rollout might go through 10s or 100s of states, but you don't store them
LoGos: MSmits may I ask you how many sims/nodes do you create pern turn?
eulerscheZahl: i expand 1 node and go further on that path of course
MSmits: Let me go and see LoGos
LoGos: I could only go ~100-150k in first turns
eulerscheZahl: and along my rollout path i create 4 nodes for each unexpanded node
MSmits: first turn with 1 second LoGos?
MSmits: or per 100 ms?
dbdr: along the whole path eulerscheZahl?
LoGos: there is no 1s in first turn
dbdr: until the end of the game?
eulerscheZahl: do you need an image dbdr? :/
MSmits: LoGos there is
MSmits: LoGos how do you define a rollout?
MSmits: 100-150k is a lot
struct: Which game?
LoGos: I mean 100-150k new nodes with 1 simulation for each of them
Scarfield: Bandas
MSmits: ahh ok
MSmits: Well in the first 1 second I get about 2 million new nodes
MSmits: then per 100 ms, somewhat over 200k new nodes
LoGos: There really is 1s in the first turn? I didn't see it anywhere...
dbdr: https://upload.wikimedia.org/wikipedia/commons/thumb/2/21/MCTS-steps.svg/1200px-MCTS-steps.svg.png the point is, no nodes stored on the wiggly path at the bottom of the 3rd tree
MSmits: i didnt either, but apparently there always is
eulerscheZahl: http://www.webgraphviz.com/
Scarfield: i think its a standard to have more time first turn for every game (?)
eulerscheZahl: http://chat.codingame.com/pastebin/39bf4ff4-6265-4cf1-b3c6-867b25da5ff1
LoGos: well s**t That's 9 missed turns for me
eulerscheZahl: paste that code on the website
eulerscheZahl: and click "generate graph"
eulerscheZahl: CG always gives 1s for the first turn
struct: yes Scarfield,
MSmits: that's cool eulerscheZahl
eulerscheZahl: that is because some languages are slow at startup
eulerscheZahl: the graph sharing or the way i explore MSmits?
MSmits: graph
LoGos: I actually thought it's only in UTTT
MSmits: i have no idea if the way you explore is good
LoGos: because of 81 possible moves as first player
eulerscheZahl: sometimes i generate these graphs in code
MSmits: LoGos, everyone starts with middle as p1
MSmits: also out of 81 moves only 15 are unique
dbdr: logos also it's not 81 because of symmetries
dbdr: as msmits says :)
eulerscheZahl: https://eulerschezahl.github.io/NumberShifting.html generate a level and open the console MSmits. you'll see a graph right there ;)
struct: euler seems like we can change firstTurn timeout on sdk
struct: Is this recent?
LoGos: Yeah I've taken it into account. I rotate my board according to enemy move if I'm first
eulerscheZahl: are you sure?
LoGos: brb I'm going to use these sweet new 9 turns
Scarfield: :D
dbdr: decrease it to 100ms to screw JIT languages :D
eulerscheZahl: where did you see that struct?
MSmits: Could not load content for https://eulerschezahl.github.io/pixi.min.js.map
dbdr: 25s first turn :D
jacek: but, can we have a total time limit for player, not the turn one?
struct: :D
struct: no jacek
eulerscheZahl: oh right, now i remember seeing that in autocomplete once
struct: I would like it too
Scarfield: oh like chess games?
Scarfield: could be cool
struct: yeah
struct: timebank
Astrobytes: I can't see that happening
eulerscheZahl: MSmits as long as the UI works i don't care :D
jacek: probably one could simulate that in referee, but its kind of unreliable
struct: I tried its unreliable
jacek: like theres 100ms limit, but i can do 105 or something easil
dbdr: you can make a game where the player has to send bytecode, or sth like CGFunge. then you interpret it yourself and you can implement whatever time rules you want
eulerscheZahl: you don't get any closer to a timebank with the first turn time btw
eulerscheZahl: would just help for games that you solve in turn 1 such as Bender 4 or A*Craft
dbdr: those you solve in turn 0 ;)
eulerscheZahl: :D
eulerscheZahl: the interpreter is the only way to get close to a timebank. but then again the referee itself can timeout
eulerscheZahl: because there's a limit too
eulerscheZahl: not in the offline SDK
eulerscheZahl: had to learn it the hard way during OOC
dbdr: then the referee need to JIT the player's code
dbdr: fun project
eulerscheZahl: the code we inherited had a slow BFS (like 4s worstcase for a single run)
dbdr: ouch
eulerscheZahl: worked fine offline when we tested shooting out of the map
Astrobytes: 4s!
eulerscheZahl: online it crashed
eulerscheZahl: and you don't check the timing, as the game itself takes a bit anyways
eulerscheZahl: it was visiting the same node multiple times because it set the parent node as visited and not the new one
MSmits: kutulu also had a costly A* didn't it?
Astrobytes: lol
eulerscheZahl: a degenerated A* that's actually BFS
eulerscheZahl: but not visiting the same node again
Astrobytes: The kutulu was a failed a*
Astrobytes: dammit
eulerscheZahl: next time i'm somehow involved in a contest, remind me to check the path finding
MSmits: or any algorithm that is at risk of (near)infinite loops
eulerscheZahl: i even checked the "shoot out of map" locally to see if it crashes
eulerscheZahl: worked fine, i didn't stop the time :cry:
LoGos: The difference in # of created nodes in enormous D:
LoGos: Just with 900ms more in the first turn my node count doubled
eulerscheZahl: and how long to these nodes survive in later turns?
MSmits: LoGos, better to just reset your tree when you get near the end of your pool
MSmits: problem solved
LoGos: idk but there is an interresting pattern there. There is a sudden jump in # of simulations in the middle of the game. Now my tree can reach this state faster (?) so that's where the gain comes from
struct: in uttt how often do you reset?
MSmits: once at most
struct: you reset the tree only 1 time?
struct: The entire game
MSmits: yes, but that's if i dont use heuristics
eulerscheZahl: robo just starts at 0 again when you gets out of range
MSmits: sometimes i use move-selection heuristics, if an opponent picks a move that falls out of the heuristic, my tree doesnt cover it and resets also
eulerscheZahl: possibly overwriting the root
MSmits: oh right, i do that too
struct: Ok I need to redo uttt
struct: I reset tree every turn
struct: I start from 0
LoGos: i think it's not the same tree
MSmits: it honestly doesn't matter that much struct
MSmits: well bandas and uttt are a bit different
eulerscheZahl: UTTT has a lot of possible moves
MSmits: but in bandas also, most nodes in the tree for your current turn will be new nodes
eulerscheZahl: bandas has 4 per turn, much more reusing possible
Astrobytes: NAh just keep going until you hit the end
eulerscheZahl: that's what i do
eulerscheZahl: that's why i crash in longer games
Astrobytes: wow lag much, I meant that in reusing the nodes
struct: For example on uttt
struct: I expand the nodes
struct: and i almsot use all nodes I allocated
MSmits: i'm trying to write an eval for bandas now. To see if it's better than a random playout
Astrobytes: EPT?
MSmits: yeah, though the way i do it, theres no sim at all
MSmits: in oware and onitama anyway
Astrobytes: Yeah you have the solver
LoGos: MSmits when resetting I actually use node's before root's index It only works once per game though
MSmits: no simulation after expansion
MSmits: ahh
MSmits: well thats what i do in uttt LoGos
MSmits: but in bandas it didnt work for me
MSmits: I will end up overwriting the root eventually
MSmits: The thing is though, I think most games are decided before you even get to the point when you're going to have to reset
MSmits: bandas is quite unbalanced a lot of the time
struct: I think something is wrong on my mcts implementation
struct: I run out of nodes so fast
PatrickMcGinnisII: Illedan your game crashed
eulerscheZahl: bitrunner?
MSmits: in which game struct, also how many nodes?
PatrickMcGinnisII: The game has crashed. Please contact the author and enclose the following error: http://chat.codingame.com/pastebin/be6df6dc-634c-4ade-a3a7-cf1ae78a44bd
struct: I have 15 Mill nodes allocated
struct: uttt
PatrickMcGinnisII: Tulips and Daisies
eulerscheZahl: that's not illedan's game
MSmits: const int NODE_MAX = 25000000;
MSmits: 25 million for me
PatrickMcGinnisII: oh, it was approved, i gotcha
struct: The problem is that I almost use the 15 mill on 1 turn
eulerscheZahl: so: crash the referee when you are losing?
Astrobytes: eh?
MSmits: dont make new nodes during a random playotu struct
MSmits: thats probably what you do
LoGos: yeah, I had the same problem
MSmits: either that or your exploration parameter is waaaaayy high
struct: on expansio nI do new nodes
struct: I shouldnt ?
MSmits: sure you should
MSmits: so thats about 7-9 new nodes per expansion in early game
struct: yeah
PatrickMcGinnisII: eulerscheZahl, yup apparently all you have to do is send a NULL NULL to the ref
MSmits: struct I use about 8-9 million nodes in the first turn
struct: the one with 1sec or 100ms?
MSmits: 1 s
struct: I use 9 mill in 100 ms
struct: ...
struct: (╯°□°)╯︵ ┻━┻
Scarfield: when expanding are you creating nodes for all children, or only one?
MSmits: yeah definitely doing something wrong
MSmits: it could be you're incrementing your node index by 9*9 somehow instead of by 9
MSmits: so your nodes get spread out far
MSmits: either that or your bot is amazing
struct: My bot is not amazing
struct: Is just expanding a lot
Scarfield: suffering with succes :p
Astrobytes: It's amazing at expanding, always find the silver lining :P
MSmits: do some testing, just make a simple counter that increments by 1 every time you run the expansion function
MSmits: and print it
struct: ok
struct: Expansion counter: 1346042
struct: Lol
struct: So i guess whats wrong in my code is my selection
Astrobytes: Maybe your exploration constant as MSmits suggested earlier
MSmits: it's still weird though, if your mcts is done right it should play a random game after every expansion
MSmits: that random game slows you down a lot
MSmits: 1.3 million expansions, is a lot of randomg ames
dbdr: just pick a random outcome for the game. faster than simulating it
MSmits: lol
Astrobytes: FITS
dbdr: random vs random, what's the difference ;)
Astrobytes: "F**k It Tree Search
reCurse: Big brain
MSmits: right
dbdr: Yolo Tree Search
PatrickMcGinnisII: You life is upsidedown when the GF won't go to the store with you unless you wear protection
MSmits: AntiSquid, please respond to this
MSmits: So many things I could say now, but I have students that might read it
struct: so this is my stats 26k sims 10Mill nodes 1.3Mill expansions
MSmits: impossible
struct: 26k full plays*
MSmits: every expansion should have a full playout
MSmits: or more, even
struct: Maybe I have definition wrong
Astrobytes: You may have a bug.
struct: Or maybe this aint even mcts
struct: lol
dbdr: MSmits you need an alt :D
MSmits: hehe maybe
struct: for me expansion = make children
struct: is that wrong?
MSmits: according to the Bible it's great
Astrobytes: Most people have sex struct
MSmits: but yes, thats fine struct
struct: 26k is rolout i guess is the right definition
MSmits: cant be
MSmits: every rollout has 1 expansion
struct: Ok end games
struct: 26k end games
MSmits: every end game has 1 expansion
struct: wtf
MSmits: it's like this
MSmits: 1 selection
MSmits: 1 expansion
MSmits: 1 random playout
MSmits: 1 back propagation
MSmits: thats vanilla mcts
reCurse: 1 vanilla mcts
struct: Isnt that depth 1 though?
MSmits: thats 1 turn
reCurse: 1 turn
struct: hmm
MSmits: err i mean 1 iteration
MSmits: not 1 turn
reCurse: -1 turn
reCurse: 1 iteration
struct: I do something completely different then
Astrobytes: 1 bug?
reCurse: 1 delete
MSmits: reCurse i am not going to pay if this is the boosting you promised
struct: I select expand select expand select expand.. .until game ends
Astrobytes: -1 customer
MSmits: struct nope, thats wrong
MSmits: you need to play a full game after every expansion
dbdr: 1 strawberry mcts please
reCurse: We're out of strawberry
struct: So what I do isnt MCTS
dbdr: cake or death?
MSmits: no it's not
struct: I guess I need to read what it is
MSmits: yep
dbdr: might help
Astrobytes: What you do is MCST, Monte Carlo Struct Thing
MSmits: it's just struct thing
reCurse: aka Monte Carlo Programming
MSmits: there's no random
struct: I tried smitsi with this approach too
dbdr: maybe it's Markov chain Monte Carlo Tree Search
struct: I guess I now see all the flaws
Astrobytes: ST search. You invented an algo struct. tech.io article please
MSmits: well actually it works ok in smitsi
MSmits: at least, say you use depth 7 and you simply expand down to depth 7 randomly and then backpropagate
MSmits: worked fine when i did it
struct: Yes thats what I do
struct: So the step after expansion
MSmits: then Robo said he did an expansion and then random moves till depth 7, the way mcts should be
MSmits: that worked fine too
struct: is random playout?
MSmits: yes
struct: So is just random moves until game ends?
struct: and I do not need to store this data?
MSmits: yes, could be about 50-60 moves for uttt
MSmits: just the end result
struct: Just win or loss
struct: right?
MSmits: or draw
struct: yeah
MSmits: i suggest -1, 0 and 1
MSmits: feels cleaner and more symmetric than the 0, 0.5, 1 many use
struct: Yeah I have those I think
struct: 1 0 -1
MSmits: good
struct: But I think I understand now
struct: At least most of it
MSmits: You may have some sign errors when you first make it work, that make your bot pick the worst moves
MSmits: that always happens to me
LoGos: struct maybe this'll help https://ai.stackexchange.com/questions/7877/which-nodes-are-expanded-in-the-expansion-phase-of-mcts/7878#7878
MSmits: hmm that's no longer vanilla mcts though
MSmits: but its what I most often do
MSmits: instead of doing one random playouit, you playout from each child and add up the results
MSmits: this is preferable when the selection process is very expensive
MSmits: or when you have very few children
dbdr: Vec<Vec<Vec<(Peg, usize)>>> that should do it. if not I'll add one more dimention of Vecs
MSmits: in uttt i just do one playout
MSmits: but i tried both ways, for me 1 was best, for others it wasnt
struct: Thanks LoGos
dbdr: MSmits where does the depth 7 come from?
dbdr: just a parameter?
struct: CSB
MSmits: csb
MSmits: I stop at depth 7, i had 8 for a long time, but 7 is better apparently, 6 was worse again
dbdr: so that's early playout with an eval?
MSmits: no, csb is smitsimax... it's the same algo i use in PCR really. Just 4 disconnected trees doing their thing only connected through the simulation
MSmits: and combined eval
MSmits: *in some way*
MSmits: for PCR it's much nicer, a tree for every chip
MSmits: well the player ones, not the neutral
MSmits: but Marchete did it better obviously
Scarfield: marchet e is using GA ?
MSmits: no he;s using the same as I do
MSmits: but different move-structure and different eval
MSmits: dbdr uses some magic monte carlo.
Scarfield: okay, it will probably be the next multi i will try :)
MSmits: I liked it, but the sim is a pain
MSmits: oh seems yurkovas is now nr 1
MSmits: he also uses smitsi, at least I think that is what he told me
MSmits: he tried it in csb as well, with less success
Astrobytes: Yeah he used smitsi iirc
MSmits: I like him, he always keeps me updated and is very friendly. Hope he sticks around. Have you seen him online?
YurkovAS: MSmits yes, in PCR and CSB i use smitsimax
Astrobytes: Not for a while actually, and yeah, nice guy
MSmits: well there he is!
Astrobytes: There he is :)
MSmits: what are you working on now YurkovAS?
dbdr: :)
YurkovAS: MSmits after contest i make: Tron Battle = 38, Game of Drones = 615, Back to the Code = 920. But all is hard for me or i'm tired
MSmits: nice, i suck at tron and drones, havent tried back to the code
dbdr: bttc was my first multi
dbdr: no leagues
MSmits: oh right
Astrobytes: That's the square-filling thing with the back in time thing right? Never got around to writing a proper bot for that
struct: yes
struct: with the rollback option
Astrobytes: Yeah the time thing.
struct: Im gonna add that feature to Yavalath
Astrobytes: Is it any good?
Astrobytes: lol
MSmits: fine with me :P
MSmits: just roll back and pick another move that's also in my book P
MSmits: :P
Astrobytes: And change the name to Y'avin-a-laugh?
MSmits: that would be interesting
MSmits: too bad he cant edit anymore
Astrobytes: Sorry that's rather UK-biased wordplay
MSmits: how is the UK now btw?
Astrobytes: He could always make Yava 2
struct: first Yinsh
Astrobytes: Fucked MSmits
MSmits: seemed that way from worldometer, it's up there with Italy and Spain
Astrobytes: Sorry to be so blunt
MSmits: Belgium is worst of all it seems though
Astrobytes: Govt trying to cover up their catalogue of errors
MSmits: really?
AntiSquid: Astrobytes 09:48PM Fucked MSmits and now bragging in chat?
Astrobytes: Yup.
Astrobytes: LOL AntiSquid
Astrobytes: :D
MSmits: too late AntiSquid, i called you back when patrick made his comment :(
AntiSquid: what ?
MSmits: I needed you to say stuff I can't in public
MSmits: there was much potential for jokin
Astrobytes: You need an MSmuts smurf MSmits
MSmits: ye
Astrobytes: No-one will ever guess
MSmits: at first
MSmits: second guess will be correct though
struct: Someone will find
struct: Trust me
Astrobytes: Just tell everyone someone is trolling you hard
MSmits: first guess: Hey, did Cyber come back?
struct: You cant hide forever
AntiSquid: oh hm in UK the staff in the store have a windshield basically
Astrobytes: Cyber was active in RU discord recently, not sure if he was spamming RU chat (I don't go in there)
struct: #ru
AntiSquid: i don't see why he'd take private business to the shop
wlesavo: not that i saw
MSmits: :)
Astrobytes: good for everyone in #ru :)
wlesavo: maybe he is basnned in ru?
AntiSquid: good for cyberpunk, he wastes too much time in chat
Astrobytes: Quite possible
AntiSquid: doubt it wlesavo
AntiSquid: he got once kick from main channel by Madao and then stopped writing in here claiming he was banned
wlesavo: lol
AntiSquid: actually i forgot the exact story
Astrobytes: Something like that
AntiSquid: someone about MK calling him SJW and then cyberpunk blocking him was also part of the story
AntiSquid: something *
Astrobytes: No, I think it was feminist
AntiSquid: ah but then cyber said he was called sjw
AntiSquid: cyber exaggerated some stuff
AntiSquid: i forgot why he got mad at me once
Astrobytes: not like him
Astrobytes: :P
AntiSquid: he was being a drama queen at some point before he left
Astrobytes: He's a sensitive soul
AntiSquid: he wasn't like that when afaik
AntiSquid: i remember / missremember him thick skinned
AntiSquid: non stop spam and wouldn't give a crap about the jokes coming his way
Astrobytes: Yeah, but I think he got pwned in chat too many times
Astrobytes: And gave up, in World at least
Astrobytes: Who knows
AntiSquid: it's ok, best for him to not spend entire morning typing 10-20 lines of chat in ru, that's not good use of time
AntiSquid: it was just him typing btw
Astrobytes: I've seen worse screenshots than that lol
AntiSquid: although i find chess also a waste of time , but whatever
Astrobytes: Fischer-random is nice
Astrobytes: More gameplay less memorisation
AntiSquid: ya so much chess theory just to get wrecked by a computer
Astrobytes: Don't hate the game, hate the player :P
AntiSquid: no hate
Astrobytes: No I'm just playing on the expression
JBM: you are angry, and in your anger lies great power
Astrobytes: Emperor JBM in the house
AntiSquid: smash the chess
MSmits: ohh, you guys are embracing the dork side without me
Astrobytes: MperorSmits, sorry my Lord
Astrobytes: Don't use Force pext on me
MSmits: Bow, worm, or I'll pdep you
Astrobytes: hahaha
Astrobytes: Rum?
Astrobytes: ;)
JBM: i'm in the belgian phase of my beers
Astrobytes: Anything nice?
AntiSquid: are you working from home and getting drunk in the morning? they can't smell your alcohol
JBM: i'm just buying them from the supermarket, i'm not going to discover anything exotic there
Astrobytes: True
AntiSquid: have beer in a cup and say you drink tea if you have to talk to your colleagues online
JBM: so tonight was delirium tremens
JBM: as overhyped as ever, but i gotta make sure every now and then
Astrobytes: Great name tho
AntiSquid: lockdown is fine for some people, but really fucks up the most people i think
AntiSquid: ya but if you work in hospitality you're screwed
Astrobytes: Doesn't fuck me up, other than work I'm pretty isolated
AntiSquid: even as a manager
JBM: most non-food-related trades have had a very tough time
AntiSquid: hotels and restaurants and such are screwed
Astrobytes: Pharma PI is messed up right now, first Brexit and now this
JBM: in a slightly different shape
AntiSquid: actually pizza hut is still open and a fish and chips shop near me
AntiSquid: ya
AntiSquid: i guess takeaways are allowed
Astrobytes: :beers:
AntiSquid: well the folks in the flood plains have it worst ...
Astrobytes: And the govt continues to want to encourage building on flood plains
AntiSquid: and since there's lockdown probably nothing will be done once again and they'll get flooded again next year?
JBM: i suppose it's mostly safe by now
MSmits: what is?
AntiSquid: ya wtf, it's like UK tries hard to go backward in time :/
JBM: i would'nt have been caught dead there during the first weeks
Astrobytes: Indeed AntiSquid
Astrobytes: You couldn't have been dead here if you weren't actually here JBM
Astrobytes: ;)
JBM: neither would i have wished to
JBM: it's so hard for the general public to grasp the "exponential" concept
Astrobytes: Governments too apparently
MSmits: it could be worse. They could be telling you to drink bleach and inject desinfectant
JBM: including the very ministry of health
Astrobytes: Yep and yep
Astrobytes: You know he was lobbied by the peddlers of that chlorine dioxide stuff before he made that statement?
JBM: who though it was better to switch positions in mid-Jan to run for the paris mayorship
Astrobytes: Well played :clap:
MSmits: lobbied?
Astrobytes: UK was too busy playing at Brexiting. We were all warned.
MSmits: do you have a link/video?
AntiSquid: well if this lockdown thingy continues on for too long who knows what ideas might emerge ... "let's restructure the way we run society"
MSmits: not in my country... here be sensible people
Astrobytes: Yes MSmits, https://www.theguardian.com/world/2020/apr/24/revealed-leader-group-peddling-bleach-cure-lobbied-trump-coronavirus
AntiSquid: all american politicians seem crazy to me
Astrobytes: Btw, the Guardian is just about the only reliable news source left
Astrobytes: Here
AntiSquid: i hear a different opinion from everyone else
JBM: yes, but this one is the Right one
Amrhossam: i want to create a new social platform like twitter or facebook
Amrhossam: any thoughts ?
Astrobytes: They give column space to anyone from any party ideology, they fact-check
JBM: why are you wasting time on chat instead of coding?
Amrhossam: time is not wasted discussing ideas
Amrhossam: what do you think is missing from those two
AntiSquid: my thought is JUST DO IT
Astrobytes: AmrhossamGramBook is the future
AntiSquid: your users will tell you what they want
Amrhossam: more details about integrity ?
Amrhossam: tell me what do you mean
AntiSquid: what's that Astrobytes ?
AntiSquid: are you looking for success or are you looking to please people Amrhossam ?
Astrobytes: What's what AntiSquid?
AntiSquid: grambook
JBM: i was more in a "what's your sellout price" mood
AntiSquid: spot on JBM
Astrobytes: Just playing with his username and a portmanteau of that + Instagram + FB
JBM: now where's that onion video
AntiSquid: i liked hi5 back when it had chat rooms
AntiSquid: and before they added the ability to edit your profile with all sorts of weird addons (scripts, css styles, etc)
JBM: Here we go: https://www.youtube.com/watch?v=w8c_m6U1f9o
AntiSquid: the onion is the only honest news source
AntiSquid: wow that audience is so serious JBM
Astrobytes: That's spot on lol
Amrhossam: i thought about a feature that already exist but tried to implement it in a different way
Amrhossam: maybe it's implemented wrong and i'd like to enhance it
Amrhossam: is this okay ?
Astrobytes: By removing any privacy whatsoever?
Amrhossam: No
AntiSquid: ok honestly, at the moment most serious social media site is Minds, but i don't even use that
Astrobytes: Shame on you. It won't work Amrhossam
Amrhossam: User will have full control about his content and maybe all the data will be encrypted
Amrhossam: so no data selling for adv companies
Astrobytes: Until the government demands data from you
Amrhossam: only if you're storing it on a server elsewhere
Amrhossam: then
Amrhossam: you can tell. sorry i can't even access for taht
Amrhossam: that*
Amrhossam: i mean, you can encrypt your data even from your gov.
Astrobytes: Yes, but when they come and break down your door...
AntiSquid: JBM check out WeChat, politically controlled social media platform used all over china
Amrhossam: why you're taking the subject to the extreme now :d
Amrhossam: okay here's a question.
Amrhossam: what do you guys think about stories features implemented in both instagram and facebook
Amrhossam: something is missing from it right
Amrhossam: it doesn't feel natural
AntiSquid: Amrhossam what world do you live in?
Amrhossam: @AntiSquid the same world you live in
AntiSquid: the reason facebook and instagram and other sites work are because they grab the attention of the user and get them addicted, that's what you need to aim for lol
AntiSquid: then i don't have pink tinted grasses on Amrhossam
Astrobytes: *rose
AntiSquid: snapchat is great for teens to pull funny faces and share them
Amrhossam: you're right, getting people addicted you have to give them what they didn't know they needed to
AntiSquid: you need to cater to the dumb and gullible
Astrobytes: That's modern capitalism
AntiSquid: Amrhossam more like convince them they needed your product ...
Astrobytes: Sorry, neoliberal economics
AntiSquid: and make sure the font you use on your website is pretty (learn from Steve Jobs)
Amrhossam: i see
Astrobytes: Did you watch the Iron Sky movies yet AntiSquid
Amrhossam: maybe i'm ambitious, or maybe i'm still naive
AntiSquid: is it about nazis on the moon?
Astrobytes: Naive yes
Astrobytes: Yes AntiSquid
AntiSquid: yes years ago, why
Astrobytes: The 2nd one features a cult that worship Steve Jobs
AntiSquid: oh
AntiSquid: well makes sense
Amrhossam: but even if i'm naive there's still opportunities from ideas right ?
MSmits: are the nazi's the ones that made the flag seem like it was fake to make people think the moonlanding wasnt real?
Astrobytes: And Nazis at the centre of the earth
Astrobytes: And dinosaurs
AntiSquid: no religion of memes and people praying to lolcat ?
Amrhossam: i mean this can't be it we are all set and got what we got
Astrobytes: No but close
AntiSquid: well i mean, you could aim for utility based social media
Astrobytes: That's a good attitude Amrhossam
AntiSquid: if it's EXTREMELY USEFUL like let's say github then people will use it!
AntiSquid: but you stil lend up selling out people, see google
MSmits: is google worse than facebook?
Astrobytes: Yeah, I'd focus on making something useful. But as AntiSquid says.... at some point the investment groups come knocking
MSmits: facebook aided genocide, pretty hard to top that
Astrobytes: Is drowning worse than burning MSmits?
MSmits: i think burning is worse
MSmits: but I get your point
MSmits: you're saying you should get both facebook and google so you can put out the fire
Astrobytes: It depends. Some people drown rather peacefully, others have a reaction which makes them continue to gasp for air
Astrobytes: lol, if they cancelled each other out that'd be great
MSmits: personally I dont use any social media. Except this I guess
AntiSquid: more likely to cooperate
MSmits: oh and sometimes educational youtube i guess
AntiSquid: they are there to make money not shut down each other to make people happy
Astrobytes: I have to use FB to keep in touch with a few stubborn people
MSmits: ah
MSmits: my wife does that
AntiSquid: i deleted my facebook although i lost contact with quite a lot of people, i just couldn't stand that site anymore
MSmits: good
JBM: what's the fb/genocide story?
MSmits: google facebook myanmar
Astrobytes: I use email, discord or whatever to keep in touch. Some people just don't want to lose their FB it seems
MSmits: https://www.nytimes.com/2018/10/15/technology/myanmar-facebook-genocide.html
MSmits: facebook had 2 moderators for the entire area
MSmits: so people just used facebook to gather up and mob moslims, killing them
AntiSquid: i like discord, no idea about what harm they could do if they wanted to ...
AntiSquid: but facebook is thought policing people
MSmits: worse is when they aren't policing at all
Astrobytes: MSmits I'm sure it's not the only example. I mean (OK it's not genocide) but we have vigilante groups in our town who operate on FB
MSmits: they just allow fake news to exist
JBM: i'm not quite clear how i think they're responsible
Astrobytes: JBM, the point was they allowed loads of fake accounts to spread fake news
MSmits: if you provide a tool for governments to spread false information you're aiding it
MSmits: it's their servers
AntiSquid: MSmits actually someone told me "die you motherf***" on facebook, unprovoked actually, facebook tells me it doesn't violate their standards, was initially talking about food wtf ... :/
Astrobytes: Like they do with most countries fake news fake posters
JBM: i'm not sure how i'd expect them to determine what's fake or not
Astrobytes: I reported an animal mutilation video on FB once and they said it was allowed
MSmits: if it's inciting hate and you can't fact check it, remove it
MSmits: err on the side of caution
Astrobytes: Censorship
Astrobytes: Yep
MSmits: Would you be ok with a news network broadcasting verifiable lies as news?
JBM: i don't think i ever read it in en
JBM: but they're not a news network
MSmits: facebook is the only source of news for many people
Astrobytes: censure is also a word
AntiSquid: when corporations grow too big their like small countries, ruling over their users
AntiSquid: they're
JBM: but it's not a problem i can sincerely hold against fb
MSmits: who would you hold it against?
AntiSquid: fb should be policed by the US, enforce free speech imo
MSmits: that's easy to say
MSmits: true
Astrobytes: In the case of Myanmar Aung San Suu Kyi probably, and the West for allowing it
JBM: i wouldn't have specifically held it against anyone spontaneously
MSmits: the thing is, servers are actively being maintained
JBM: humakind is capable of the best, but it's the worst that spreads
Astrobytes: So true
MSmits: if as a company your products are being misused for this sort of thing, I think you should be held accountable by the country your company is based in
MSmits: if you dont respond
MSmits: they could have just removed all accounts
MSmits: yes
JBM: i don't see that working anytime soon
Amrhossam: it wouldn't make much sense if i asked what utilities do you might think you need right ?
JBM: Especially in a suit-seeking country as the US
MSmits: things don't have to be easy to fix, to be worth fixing
AntiSquid: Amrhossam this is what true success looks like : https://www.theguardian.com/technology/2020/may/01/amazon-jeff-bezos-testify-house-antitrust
JBM: but saying "things shoulg be so" doesn't "fix" things
MSmits: it's mostly a political problem
AntiSquid: or like this Amrhossam : https://www.cnbc.com/2019/10/09/facebook-ceo-mark-zuckerberg-to-testify-before-house-financial-services-committee-on-libra.html
JBM: how do you make everyone agree to the same set of morals?
AntiSquid: and there you need to pull a poker face and tell half truths
MSmits: some things are no-brainers JBM. You don't have to touch the more controversial things
Astrobytes: "it's mostly a political problem" - whether we like it or not, everything's political
JBM: i don't think it's that easy
AntiSquid: Amrhossam try online chat rooms, or whatever, make it easy for people to socialize, simple as that
MSmits: we have a set of laws don't we? They are based on morals and even though we dont all share the same morals, we share the same laws
MSmits: within a country that is
Astrobytes: What if the laws are immoral
JBM: and as for now, they don't force anyone to police news spread by individuals at the other end of the world
AntiSquid: MSmits for example everyone hates trump, but then people overlook Pelosi promoting to hug chinese people when the outbreak started, imo that's the grey area, although all US politicians are nutjobs
Astrobytes: Well, it's tricky to get information out of some places
MSmits: my idea would be to set a limit on views. If a post has x views it has to be fact checked
JBM: but how do you tell jokes apart from fake news?
Astrobytes: How would you determine which x is needed?
JBM: preventing sharing good jokes seems imoral
MSmits: Astrobytes implementation detail
MSmits: you could say 10k
MSmits: or whatever is agreed upon
AntiSquid: ya easy to trigger someone these days and be called insensitive
Astrobytes: 10k might have laughed at it and thought "what a load of shit"
JBM: it's the decision problem you need to expand on
MSmits: i am not talking about talk offending people, just spreading obvious misinformation
MSmits: real world facts
JBM: you keep using that world
Astrobytes: Yeah, it works
AntiSquid: MSmits what is a fact? the COVID was only considered a pandemic like 2 months after the entire world knew about it
AntiSquid: before that everyone said it's all good and rosy
MSmits: it's not actually that black and white AntiSquid
MSmits: pandemic is a definition
JBM: we mustn't be in the same circles
MSmits: but the disease was there and people knew it
Astrobytes: No actually that's an epidemiological thing, declaring a pandemic
JBM: i've had charts of wuhan's exponential since early january in my twitter feed
AntiSquid: yes and the WHO did they spread misinfo or not?
Astrobytes: No they did not
AntiSquid: then why did they hate on Taiwan ?
MSmits: well they were overly positive about China
Astrobytes: And everyone has had access to data and papers since the end of last year
AntiSquid: MSmits and that wasn't misguiding at all?
MSmits: of course it was
JBM: what's the taiwan hate story?
AntiSquid: exactly my point
MSmits: but if you can't factcheck something you dont have to remove it of course
MSmits: some things are easy to fact check
AntiSquid: JBM basically taiwan reacted immediately and was called out as "racist/discriminating"
AntiSquid: but then we are all in lockdown now so maybe they didn't overreact eh?
Astrobytes: Chinese COVID-19 death rates not being one of them
MSmits: if someone starts posting it is a good idea to drink bleach to cure covid, fact checking isnt a problem
AntiSquid: yes
JBM: didn't hear about it at the time
Astrobytes: Source?
AntiSquid: https://www.youtube.com/watch?v=eisQ6RiHhp4
JBM: barely heard about taiwan tbh, my asia news were shared between "wuhan virus" and "hk protests"
AntiSquid: well you can look for sources out there
MSmits: I also saw an interview where China refused to call Taiwan a country or something
MSmits: sorry
MSmits: where the who refused
JBM: that's, like, par for the course
Astrobytes: Yeah went straight from HK to Wuhan very swiftly
AntiSquid: and what was that crap in italy about hugging chinese people ?
AntiSquid: then the outbreak in italy was pretty much the worst in europe
AntiSquid: https://www.youtube.com/watch?v=mNMdg4morQs
Astrobytes: Dude, we had people in this town refusing to eat from the Chinese takeaway then rushing out to stockpile soy sauce and pasta
JBM: and Corona beers tanking in the US
AntiSquid: ya, but point is why was that promoted at all Astrobytes ?
AntiSquid: actually i didn't give a damn about the news back then just went out to eat at a chinese place because it was chinese new year
Astrobytes: It appeals to their sense of blaming 'the other'
MSmits: honestly, this talk of blaming covid on China is very hypocritical. These same people dont give a sht that China has 1 million people in concentration camps
MSmits: so much worse
Astrobytes: The general population can't deal with unknowns
Astrobytes: Always have to find a target to blame
AntiSquid: it's not about blaming china, it's about the media in the west not being honest
AntiSquid: it doesn't matter where it came from, maybe next time it comes from elsewhere
Astrobytes: Since when has the Western media (for the most part) been honest?
BenjaminUrquhart: this
BenjaminUrquhart: no stop
MSmits: I dunno, in my country they are at worst biased towards sensationalism.
Astrobytes: Fair JBM
AntiSquid: point was how can you "control" such dishonesty ? you either get skewed thought policing or you just let everyone speak up their mind
MSmits: its difficult to navigate AntiSquid
AntiSquid: was giving an example regarding "control of missinformation"
MSmits: but either extreme is bad
Astrobytes: I guess we can call it a rough period in human evolution
MSmits: and we are on one of the two extremes now
AntiSquid: difficult to navigate means, either you give everyone same freedom to speak up, or you shut up some in favor of the others
MSmits: it;s not about who says it, it's about what they say
AntiSquid: and who's policing it?
Astrobytes: Yes AntiSquid, on the one hand you need to be able to speak freely, on the other hand how do you manage blatant lies
AntiSquid: who's watching the watchers ?
Astrobytes: Corporation bosses mostlyu
MSmits: I think governments should be a bit more involved in this
Astrobytes: *mostly
Astrobytes: Not the UK govt jeez
AntiSquid: we go through blatant lies every day, that was my point, in fact you said there's only one reliable source of news Astrobytes, so maybe it's all about people smartening up and not being pampered (protected from blatant lies)
JBM: [big] govts only get involved as far as their personal agenda goes
MSmits: AntiSquid that assumes people are intelligent enough to do this, most are not
JBM: and that's assuming they understand what's going on
MSmits: this is the biggest failing of democracy also
Astrobytes: AntiSquid, I'm not disagreeing with you, and I didn't say it was the only source of reliable news, it's one of the only reliable sources of news here in the UK
MSmits: without education, a dictatorship actually functions better than a democracy
JBM: *that* changes societies for the better
AntiSquid: 350 mil to the NHS if brexit happens, lots of people voted just because of that MSmits, next day farage went back on his word lol
AntiSquid: get my point?
MSmits: I know
JBM: so it should have been filtered
MSmits: if you dont know it beforehand, you cant filter it
Astrobytes: yeah, so far, an old war veteran has raised more money (public money) for the NHS than the govt since Brexit
JBM: fact checkers at fb now must have a PhD in economics
MSmits: were these the lies on buses btw>
MSmits: ?
Astrobytes: You all are aware of the Cambridge Analytica involvement in Brexit right? And their ties with Steve Bannon and the Trump campaign?
AntiSquid: no, just the one campaign bus afaik
MSmits: Astrobytes yeah
AntiSquid: heard something about it
MSmits: they sold facebook info
MSmits: didnt they?
Astrobytes: The Guardian did a huge exposé over abotu a year AntiSquid, look it up
Astrobytes: *about
AntiSquid: proves how powerful data science and ML is ?
MSmits: definitely is
MSmits: btw, about the educaton thing. Me and my colleague are also trying to come up with ways to make students more resistant to fake news
Astrobytes: One of the data scientists involved with Brexit is an assistant to Dominic Cummings, and they both sit in on SAGE meetings AntiSquid
MSmits: one of the compulsory parts of our CS curriculum is ethics
AntiSquid: MSmits imo you can't guide people to what's fake and what not and neither should you control how they perceive, i am a bit idealistic here, but everyone should reach their own conclusions and reach the truth that way
Astrobytes: (SAGE is Scientific Advisory Group for Emergencies i nthe UK btw)
MSmits: you need skills to do this AntiSquid
JBM: like the people trump fired?
Astrobytes: Kind of
MSmits: it's like how some people believe the Nigerian Prince and some do not
Astrobytes: We also are refusing to publish the results of a grim excercise on a 'flu pandemic from a few years ago that highlighted how unprepared we would be
Astrobytes: *exercise
Astrobytes: That's now being legally challenged
AntiSquid: idk about preparing, i mean did we prepare for floods, it's like jumping straight in ... same with the pandemic, jump in get herd immunity
MSmits: does it really matter that much though? Isn't the UK split almost as much into political parties as the US is?
MSmits: truth seems to matter so much less these days :(
Astrobytes: AntiSquid, the point is that this government focused on austerity, punishing the poor and maximising their profits. Selling everything off, whether building contracts or NHS, and are now where we are
Astrobytes: https://www.theguardian.com/society/2018/nov/16/uk-austerity-has-inflicted-great-misery-on-citizens-un-says
MSmits: seems like everything i see and hear about the US, exists in the UK to a lesser degree
MSmits: still bad
Astrobytes: https://www.theguardian.com/uk-news/2020/apr/26/doctor-sue-results-operation-cygnus
Astrobytes: Yes, we apparently aim to be (to use a Brexit term) a vassal state of the US. The irony.
MSmits: yes
MSmits: but trump is very unpopular in the UK
MSmits: except for a few crazies, I think
Astrobytes: Yet people can't see past the end of their own noses
MSmits: true
AntiSquid: interesting, thanks for sharing
Astrobytes: Deceived in exactly the same way
MSmits: well, not exactly. I mean I am pretty sure that if you guys had a government that was as bad as the US one, it would be way more unpopular.
MSmits: Trump still has over 40% support, which is insane
Astrobytes: AntiSquid you can usually trace all the stuff in the guardian back to papers, documents etc, that's why I count it as reliable
Astrobytes: Oh ours is bad MSmits
AntiSquid: ah i see, good point
Astrobytes: Just not on the same level internationally, and our PM is not an actual retard
MSmits: oh, I had another idea!
MSmits: do you know how, with smoking, companies have to put a label on it
MSmits: this stuff causes cancer and what not
Astrobytes: Do we put warnings on textbooks now - May Cause Knowledge?
MSmits: what if social media were forced to provide links to real news media whenever an algorithm detects something resembling news
Astrobytes: "real news media" falls victim to it too, that's why
Astrobytes: *why not
Astrobytes: There's reliable media, and non-reliable
MSmits: well it's more controllable and more visible. Even something like Fox news is not as extreme as many others
MSmits: you simply would not list those
AntiSquid: MSmits actually cigarettes were advertised as healthy beneficial at some point, so again all about who controls the info
Astrobytes: ^true
MSmits: that was due to corruption, when we as a society realized that was not true, we did something about it
AntiSquid: search "Vintage Cigarette Ads Promising Better Health"
MSmits: I know about this
MSmits: that too
MSmits: just because mistakes were made in the past, is no reason to stop trying to do your best to fix/prevent more :)
JBM: what other crazy thing have we promoted?
AntiSquid: well facebook is corrupt, if they control "the missinfo" then you see where this is going ...
Astrobytes: Yeah but MSmits, this was going on in Romania even recently as AntiSquid can probably confirm. I used to work with a Romanian girl who worked for Philip Morris Tobacco and had no clue for years
MSmits: the coca cola company does this now too
AntiSquid: i don't watch the news so don't know Astrobytes, i live in UK so i only know about my daily life tbh
MSmits: they are not allowed in many western countries to spread misinformation about sugar being healthy
Astrobytes: *recently as in the past decade or 2
MSmits: so they try it in countries without these laws
icecream17: huh? I have 2 chat streams....
JBM: that's the first think i noticed about czechia when i first went there
JBM: 1) ads all over the landscape
AntiSquid: i don't remember any cigar ads though Astrobytes
JBM: 2) about cigarettes and gambling
Astrobytes: I don't keep up to date with Romanian news either AntiSquid, don't be so touchy
Astrobytes: :)
Astrobytes: No it wasn't so much ads, just knowledge, smoking was cool etc
AntiSquid: ya but that's just culture
MSmits: ads for smoking on tv are simply illegal here
Astrobytes: "Just culture" is exactly the point though
AntiSquid: everyone knows it's bad, they still smoke
JBM: icecream17: good for you!
MSmits: yet it helps not to advertise
AntiSquid: going for a cigar is like going for a coffe
MSmits: doesnt fix the problem entirely
Astrobytes: It's changing now, less people are smoking
JBM: and not legalize making it "cool"
JBM: dunno about NL, but I do get the feeling many more smoke in DE than FR
MSmits: going for cigar is not the problem really. It's people smoking more than they're eating
MSmits: going through the day smoking non stop
MSmits: they cost the government huge amounts of money too, we all pay for it
icecream17: https://imgur.com/a/B3t5116
Astrobytes: Nicotine does have appetite supressing properties
MSmits: lol, restart browser icecream17
MSmits: Astrobytes yeah... nicotine isnt the worst in a cigarette I think
AntiSquid: is it possible for you to edit out the extra element icecream17 ?
Astrobytes: No it's fine MSmits
Astrobytes: I actually started vaping again recently
Astrobytes: I quit smoking 2+ years ago
MSmits: ohh ok
AntiSquid: isn't vaping still somewhat bad ? although clearly far less dangerous than smoking
MSmits: it's probably less regulated because it's not as old
JBM: i had a swedish doctor tell me chew-nicotine was actually rather good
Astrobytes: I don't give a fuck, I suffer with depression and anxiety sometimes and the vaping helps
wlesavo: AntiSquid it probably strongly depends on product quality
Astrobytes: JBM I used to snus all the time
MSmits: Astrobytes sure, just be careful about the company you buy from, they might not all be trustworthy
JBM: Astrobytes: so how was your health (and that of the other 99 people in the control group)?
Astrobytes: I buy nicotine and mix my own flavours with propylene glycol and vegetable glycerine
Astrobytes: And build my own coils. My vape is just a simple circuit
MSmits: I'm somehow addicted to cola with no caffeine and no sugar
JBM: we all do, though we wouldn't admit it
AntiSquid: the flavor i guess
MSmits: no, it has artificial sweetener
Astrobytes: JBM, my health while snus'ing? Great! I used it coming off cigarettes for a while actually
AntiSquid: aspartane addiction?
JBM: so you'd recommend it to non-smokers?
MSmits: if that's a thing, then sure
Astrobytes: No
JBM: that's kind of an orange flag
Astrobytes: Why start tobacco use as a non-user?
JBM: people start as non-smokers and do anyway
Astrobytes: There are tobacco free snus portions which I recommend if you like the flavours
MSmits: after you started, it might not be worth stopping, given the circumstances
MSmits: as Astrobytes said, if it helps with depression and whatnot...
MSmits: but if you never started, it's different
Astrobytes: I started smoking at 12. Fucking 12 years old.
JBM: there was that House episode where the solution was to fsckn' start smoking again
Astrobytes: Took me til I was about 17 to realise it was kinda hard to stop
MSmits: yeah that wouldnt happen these days i think Astrobytes, not as easily anyway
Astrobytes: Indeed
MSmits: btw, I failed at making Bandas early playout, while we were chatting about all this stuff
Astrobytes: Good thing with vaping is I know I can come off it in a week
JBM: and i'm failing at catching up on codejam
JBM: though i do feel like i'm making progres
MSmits: no i meant, i really did try, it just didnt work :P
AntiSquid: oh our distraction worked, hi five everyone, we beat MSmits
AntiSquid: nvm then
Astrobytes: lol
Astrobytes: The codejame problems looked interesting, I took a quick look earlier
Astrobytes: *codejam
MSmits: i made masks for counting the pink and blue dudes on various distances from the edge, to get a sensible eval, but couldnt make it work better than my other version
Astrobytes: Wear a different mask MSmits
MSmits: :mask:
AntiSquid: done a smoking cessation "course", to talk to people who wish to stop smoking, first guy i end up talking to says "hey i changed my mind,m i enjoy smoking, thanks for coming by though"
MSmits: lol wth
Astrobytes: haha
Astrobytes: That was me for years
JBM: yeah, round 1C is usually "doable, please be fast"
Astrobytes: The day I picked up a vape I gave away my tobacco (I smoke handrolled only) and that was it
JBM: (as opposed to 1B "pick the problems you'll finish carefully")
Astrobytes: I need to try and remember to enter it one of these years :P
MSmits: it's easier to quit smoking when your life gets easier
Astrobytes: Mine got harder MSmits
MSmits: my mother wasn't able to stop for like 40 years
MSmits: but when she retired she could do it
Astrobytes: So I contest your conjecture
MSmits: no i meant stop completely
MSmits: i remember her trying to quit before like 20 yrs ago, it goes well for a month, maybe two, then something bad happens and you start again
AntiSquid: some people can just stop anytime
AntiSquid: or maybe they are stingy ? idk
MSmits: Astrobytes i dont mean life gets easier when you stop
MSmits: i mean it's easier to stop when your life is easier
MSmits: AntiSquid it's genetic I think
MSmits: some people are more susceptible to Nicotine addiction
Astrobytes: No MSmits, my life was legit *a lot* harder when I decided to stop!
MSmits: ahh ok, that might just be you though
MSmits: doesn't have to be a general thing :)
Astrobytes: The vape really worked for me.
MSmits: generally, I think bad habits are harder to break when you're not feeling too great
Astrobytes: I tried everything else previously
Astrobytes: Yes, that's a fair generalisation
MSmits: yeah, thats maybe a bit like what i did with my cola addiction
MSmits: I replaced it with non-caffeinated version...
MSmits: caffeine was giving me loads of problems
MSmits: i got hyperventilation and stuff
Astrobytes: My mum did that. Was addicted to diet coke for years afterwards. With the caffeine tho :P
Astrobytes: Really? How much were you drinking?
MSmits: ah, i never did the sugar version
MSmits: 1 bottle a day
MSmits: 1,5 L
Astrobytes: My mum used to get through 6L Diet per day when I was I kid/teen
MSmits: thats crazy
Astrobytes: Plus coffee
MSmits: 6L? How do you even do that
Astrobytes: Hardly ate
AntiSquid: and how is she now?
BenjaminUrquhart: ah yes the coke diet
Astrobytes: She was always on the move, driving, on flights etc
Astrobytes: She's alright, high blood pressure but nothing too serious
MSmits: yeah but still 6L, thats 8 of the biggest cups isnt it?
Astrobytes: Yeah but it was in small amounts from like 3-4am til midnight
MSmits: hm ok
MSmits: well if it's diet, it's mostly caffeine + intestinal problems
Astrobytes: She doesn't drink soft drinks any more
MSmits: aspartane will become dangerous after over 30: I think
Astrobytes: Nor do I
MSmits: 30L i mean
MSmits: ah ok
Astrobytes: I drink water or fresh orange juice
MSmits: water is best
MSmits: orange juice is a bit sugary I think
Astrobytes: Been a fan of it all my life
Astrobytes: Yeah don't overdo it on fruit juice
Astrobytes: I sometimes juice my own fruit and veg
Astrobytes: We do need sugar as humans you know
MSmits: ahh, I eat carrots as candy. Other than that no veg and fruit besides what's in the powder I make shakes out of :P
Astrobytes: Just not too much ;)
Astrobytes: lol, heathen
Astrobytes: Dip those carrots in some nice homemade hummus
MSmits: the carrots is just to have something to chew occasionally
MSmits: apparently chewing is a need
Astrobytes: I have that issue as a vegetarian sometimes
Astrobytes: "Wow, I used a fork today!"
MSmits: I would be a vegetarian if I didnt go to a snack bar every 1-2 weeks
AntiSquid: as opposed to chop sticks?
MSmits: thats the only time i eat meat
Astrobytes: Spoons AntiSquid, I eat a lot of rice and dal, and other Indian stuff
Astrobytes: BUT chopsticks tonight
Astrobytes: noodles on the menu
AntiSquid: curry and nut roasts dinners eh?
Astrobytes: Fuck nut roasts
AntiSquid: :FD
Astrobytes: You can have a different veggie meal every day for over a year just by choosing Indian veg dishes from different areas
Astrobytes: IF you can get the veg
MSmits: I'm sure that's great, but it sounds horrible to me
Astrobytes: Foodist!
Astrobytes: :P
MSmits: nah, I'm just an extreme fussy eater
Astrobytes: Hadn't noticed that at all...
MSmits: that in combination with trying to eat healthy, is why i started doing the shakes
AntiSquid: thai curry is only curry i really like :p
MSmits: at least then i wont slowly kill myself :P
Astrobytes: I love a good Thai curry
Astrobytes: I have issues cause of fish sauce and shrimp paste but I break my rules sometimes
Astrobytes: *dried shrimp
Astrobytes: I still miss some nice meat things, but I don't mind so much because I can cook tasty shit
MSmits: thats good
Astrobytes: Not literally shit ofc
MSmits: I dont judge
Astrobytes: :P
MSmits: you know what bothers me sometimes
MSmits: seeing parents pass their bad habits like smoking/drinking/obesity on to their children
MSmits: with the excuse: "well i do it, cant forbid them then can i? "
Astrobytes: Happens alllll the time
MSmits: yeah
MSmits: it's child abuse
AntiSquid: https://www.youtube.com/watch?v=SGwE4hnk9rI Astrobytes
Astrobytes: MSmits this song is literally about that https://www.youtube.com/watch?v=Xsp8DmgtjHY
MSmits: ah right
Astrobytes: AntiSquid I've tasted some great stuff, aaaaannnd some really fucking vile shit
MSmits: so you did taste shit
AntiSquid: maybe they wanted to mimic the taste of innards ?
MSmits: again, I dont judge
Astrobytes: LOL MSmits :D
Astrobytes: You're supposed to be listening to the Offspring
MSmits: I think we've all been curious from time to time
MSmits: there's only so many times you can hear someone say it tastes like shit
MSmits: and think, am i the only one who doesnt know?
Astrobytes: Well... I guess there's only one thign for it
Astrobytes: *thing
MSmits: right
Astrobytes: Personally, I'm content with imagining
MSmits: so you're imagining it now
Astrobytes: (╯°□°)╯︵ ┻━┻
Astrobytes: stupid game
MSmits: what game
Astrobytes: fucking daisy tulip bastard thing
MSmits: oh, i havent started that
MSmits: seems hard
Astrobytes: It's 'fun'
Astrobytes: Trying to come up with a decent eval that's consistent
Astrobytes: And prune some stuff
Astrobytes: The codeine tablets aren't helping matters tbh
MSmits: prune a lot it would seem?
MSmits: coughing?
MSmits: or back?
Astrobytes: On the larger maps yeah
Astrobytes: Back
MSmits: ah ok, not getting any better?
Astrobytes: Docs again on Monday
Astrobytes: Fucking agony man
Astrobytes: If I sit up straight and don't move I'm cool, or stand up straight and don't move I'm cool
MSmits: sucks :(
Astrobytes: You see the issue
MSmits: yes
reCurse: So don't move
Astrobytes: Tried
Astrobytes: Unfortunately I cannot just sit in the chair all day and swallow painkillers
JBM: you could torture patiente too
MSmits: it helps if you know exactly what it is and that it's going away on its own
Astrobytes: lol JBM
JBM: you missed your path there pal
Astrobytes: Yes MSmits, it's a bit up in the air atm unfortunately
Astrobytes: Anyway, I'm not here to moan
Astrobytes: Honest
MSmits: thats ok by me, it's not like you do it all day
MSmits: well not on chat anyway
JBM: was about to bounce on that ^^
Astrobytes: :D
MSmits: :P
Astrobytes: Arrr I'm done.
Astrobytes: Stupid eval/game/universe :P
reCurse: +1
Astrobytes: no JBM, it's the game you see, it's bugged and/or trolling me you see...
AntiSquid: the galactic spirit be damned
Astrobytes: hehe
Astrobytes: Anyway, think I'll go and supplement my painkillers with a shot or two of palinka and some nooodles
Astrobytes: Or noodles
Astrobytes: See you all tomorrow
AntiSquid: bye
AntiSquid: gn8
MSmits: gn all, going to sleep also
LelouchVC2: :o
LelouchVC2: Night
cegprakash: does anyone know a good C++ stl library for matrix multiplications, vector multiplications etc?
cegprakash: I'm planning to start with NN soon
ZarthaxX: make it yourself
ZarthaxX: if you are talking outside CG, use npython with numpy
cegprakash: if I do it myself it'll be full of bugs
cegprakash: xD
LelouchVC2: Worst advice ever, make ur own library!
cegprakash: but I think for c++ it may be worth it
cegprakash: while a python guy trains for 1 hour I would train under 10 minutes
inoryy: that's not how it works
AntiSquid: lol ZarthaxX response :D
AntiSquid: why don't you make it for him ZarthaxX ? :p
LelouchVC2: He would, but he's busy with other libraries!
PortalBot: he's busy saving the world!
KiwiTae: its a good exercise though , ZarthaxX good advice
Amrhossam: join clash
Amrhossam: now
Zenoscave: Amrhossam no
Amrhossam:
- /
Amrhossam: what are you doing guys
Zenoscave: anything but clash
Amrhossam: follow me on twitter then
Zenoscave: nope
Amrhossam: @amrhossam96
Amrhossam: you said anything
Amrhossam: stick to your word
Zenoscave: Anything != everything ;)
Amrhossam: clever
LelouchVC2: Something with clash of code and leap years
LelouchVC2: I'll figure it out one day
romrd468: well for some reason velocity*1000 for csb works ok
LelouchVC2: I'm starting to hate this site
LelouchVC2: Every clash these days seems to have the weirdest prompts that barely describe the problem
LelouchVC2: And passing all the tests rarely means 100% score anymore
LelouchVC2: Then Python noobs come in and solve it in half a second with one line.
fvla: i do resent some fastest mode prompts
fvla: that are just 1 line math print statements
LelouchVC2: And now more predominately ruby coders
fvla: which language do you use?
LelouchVC2: :((
LelouchVC2: c++
fvla: you need to use a scripting language
fvla: you really do on clash of code
fvla: because c++ doesn't have good text processing out of the box
fvla: if you don't like that, then stay away from clash
fvla: and do bot programming and puzzles
LelouchVC2: C++ has some pretty good string processing which comes in handy
LelouchVC2: Just usually takes more code and a little more logic
fvla: you can't do it fast like in python
LelouchVC2: I'll use a script language when I'm sucking obama's d
LelouchVC2: Which might be soon
fvla: format(int(input()), 'b')
LelouchVC2: ;-;
fvla: take 1 line of single integer input and convert it to 0s and 1s
fvla: it's not just "a little more logic" in c++
fvla: and don't suck obama's d
fvla: but don't be too stubborn either
fvla: python is great for prototyping
fvla: i don't know about ruby, but that's just because i never needed to use it
fvla: perl is also better at some tasks, but if you want maintainable code, stick to python
fvla: javascript is more like c++ and java, but a scripting language...and i have a personal vendetta against it, but it's used too many places to not use it...
fvla: but learn python/ruby/javascript, then use it for quick prototyping
LelouchVC2: *shudders*
LelouchVC2: I went through crash courses on python/javascript - didn't like either :/ Can use them but I'd be looking up documentation all the time
fvla: what, you don't look up documentation for C++ all the time?
LelouchVC2: Not really
fvla: i mean, once you get used to python
fvla: you won't be looking up too much
LelouchVC2: I suppose, I'll get to it during the summer I guess :((
LelouchVC2: "i think" ~ Proofs Math Professor
fvla: don't learn it for the sake of clash, though
fvla: learn it for the sake of writing useful programs fast
LelouchVC2: Lol, no worries, just good to have a quick and dirty language handy
LelouchVC2: yeee <3
fvla: pytorch, for instance, which is better than tensorflow
fvla: and i use python for file processing often
LelouchVC2: Yea, well I'll see how Python goes
LelouchVC2: I wanna do C# when I get the chance *slurp*
LelouchVC2: I messed around with it for a small program, was nice to use. C++ but easier
LelouchVC2: well im outta here
LelouchVC2: thanks for the talk <3
KiwiTae: LelouchVC2 c# is like java
BenjaminUrquhart: I mean, yes but actually no
BenjaminUrquhart: that took wayyy too long to send