DIS: Coming clean
Alright, people seem to have started to get annoyed, so I believe continuing this experiment would violate my sacred and eternal duty to Treat Agora Right Good. So, yeah. I'm Greg. Or maybe I'm sending messages on Greg's behalf. See recent CFJs. As most of you accurately surmised, Greg's messages were generated by GPT-2, specifically a version of GPT-2 fine-tuned with Agoran mailing list logs since 2014. (The 2014 date is largely arbitrary—I might have been able to go a bit further back, but using much more data ran into resource limitations.) Greg was implemented using a combination of shell scripts, commands in my shell history, python scripts, a Google Colaboaratory Notebook, and me manually copy/pasting messages around. Notably, I manually pasted in and hit send on each message. I did this primarily because figuring out email APIs sounded like a PITA, but also because I wanted to be able to pull the plug in case it said anything horrific. I did, however, do as much as I could to avoid injecting my free will into the process. I operated off of two rules: for each message to the public forum, I would run a python script which had a 10% chance of invoking GPT-2 to generate a reply, which I would send verbatim. (GPT-2 barfs on overly-large input data, so I included a failsafe that automatically removed old messages in the thread until the input was small enough to work. Some messages (like the rulesets) were far too big on their own, resulting in the code generating "replies" without any context.) Additionally, each day after the first (which looks like is just going to mean "today"), I ran a script which had a 50% chance of generating a brand-new proposal. Had I been aware of CFJ 3790, I might have actually went to the trouble of having it send the messages automatically after generating them. I did "intervene" twice: for the registration message, I specifically asked GPT-2 to generate a message to BUS with a subject of "BUS: Registration". In my testing, this had about a 75% chance of generating a message that was a somewhat plausible attempt to register. Unfortunately, I got unlucky and my first "real" attempt to generate a registration message resulted in something completely random (a proposal, I think), so I generated a second one and sent that one. In another case, I discovered a bug with the large-input failsafe (turns out, GPT-2 can barf by silently returning the input with no additional output, or by throwing an exception; I was only handling the first case), so I fixed the bug and re-ran the generation. In every other case, I mechanically copies messages back and forth, following the plans I had made before sending the first message, without attempting to impose any editorial control. In my testing, Greg did occasionally borrow other people's signatures, but I didn't expect it to be this common. I considered preventing it from doing this by removing signatures from the training data (so it would never learn to include them), but I thought it was rare enough and amusing enough that it wasn't worth removing. In retrospect, I probably should have removed them. In my testing, I ran into several outputs that were interesting enough to save so I could show them later. Here are links to them: A proposal for something vaguely resembling a functional auction mechanic: https://gist.github.com/Gaelan/e7f7d3fc48c1abd08f0afb8049077acb A made-up FLR excerpt containing an interesting-sounding royalty mechanic: https://gist.github.com/Gaelan/8d092a17ed9c210685a4f4dd1e622ae2 Another ruleset excerpt, containing the core rules of an alternate universe Agora: https://gist.github.com/Gaelan/ee631f9f97b53df8483e342ef36b6618 A batch of attempts at starting new threads, with varying quality: https://gist.github.com/Gaelan/0c027e3f5b97dab700182aa663401f47 A fake rule called the "Register of Proposals", which looks like a semi-plausible implementation of proposals in an alternate-universe Agora: https://gist.github.com/Gaelan/0c6853f500799c5190a0a1ef474b098b I'd be happy to share the model at some point, but its a bit of a pain—I think it's 1.5 GB—so I'm not sure how best to do that. In the meantime, I'm happy to try out any inputs y'all are curious about. Also, From headers were included in the training data, so I should be able to ask it to generate message from specific Agorans. That might be fun. Happy to answer any questions, of course. Gaelan "Nothing in a democracy is sacred, and nothing in a democracy is sacrosanct. That's not to say it's never been broken, but it's been tracked for quite some time." [GPT-2 put that quote in someone's signature during one of my tests. As far as I can tell, it's not a real quote, but it sounds like a fairly interesting reflection on nomic.]
DIS: Coming clean?
My COMMUNITY SERVICE requirement in CFJ 3080 has already expired, but for the record, as requested by that judgement, I have also violated time limits to pay: 3054: 13 points* 3058: 25 points* 3066: 5 points 3067: 5 points 3081: 1 point *second criminal case initiated for me not paying the fine However, I did pay: 1906: 1 Note Therefore, I have paid 1 out of 6 fines during my time as a player, although 5 out of 6 were all imposed in the last two months. Other statistics: I have by far the record for criminal cases initiated against me: omd 50 Murphy 27 BobTHJ 19 also for criminal cases initiated against self: omd 9 ehird 3 Tanner L. Swett 3 I've been accused of breaking the highest number of different rules: omd 12 Murphy 9 scshunt 8 BobTHJ 7 G. 7 ehird 6 pikhq 6 root 6 Tanner L. Swett 6 Yally 6 Wooble 5 ais523 3 Sgeo 3 Zefram 3 Walker 2 Taral 2 OscarMeyr 2 Quazie 2 PerlNomic Partnership 2 Peekee 2 woggle 1 Hillary Rodham Clinton 1 Keba 1 Law-abiding Partnership 1 Tiger 1 j 1 I'm about middle-of-the-road for conviction percentage: Law-abiding Partnership 1. w1n5t0n 1. Tiger 1. Reformed Bank of Agora 1. Walker 1. Yally 0.7778 Murphy 0.7727 scshunt 0.6111 BobTHJ 0.5789 pikhq 0.5714 omd 0.5400 ehird 0.5385 root 0.5000 OscarMeyr 0. ais523 0. Sgeo 0. PerlNomic Partnership 0. Tanner L. Swett 0.2727 G. 0.1818 Wooble 0.1538 Peekee 0. Quazie 0. Keba 0. j 0. woggle 0. Taral 0. Pineapple Partnership 0. Zefram 0. Hillary Rodham Clinton 0. also for percentage of convictions sentenced to APOLOGY: Law-abiding Partnership 1. Reformed Bank of Agora 1. ehird 0.5714 Walker 0.5000 G. 0.5000 BobTHJ 0.4545 Tanner L. Swett 0. Yally 0.2857 pikhq 0.2500 omd 0. Murphy 0.1176 scshunt 0.0909 Wooble 0. Sgeo 0. PerlNomic Partnership 0. ais523 0. Tiger 0. w1n5t0n 0. OscarMeyr 0. root 0. and ditto DISCHARGE: PerlNomic Partnership 1. ais523 1. G. 0.5000 Wooble 0.5000 Murphy 0.3529 BobTHJ 0.1818 omd 0.1481 ehird 0.1429 scshunt 0.0909 Tiger 0. Reformed Bank of Agora 0. Tanner L. Swett 0. pikhq 0. root 0. Yally 0. Law-abiding Partnership 0. Sgeo 0. w1n5t0n 0. OscarMeyr 0. Walker 0.
Re: DIS: Coming clean
Given that gaelan exercised much control over greg by registering him amd by curating his training data, i think gregs actions should be considered just an extention of gaelan On Wed., 22 Jul. 2020, 10:15 am Gaelan Steele via agora-discussion, < agora-discussion@agoranomic.org> wrote: > Alright, people seem to have started to get annoyed, so I believe > continuing this experiment would violate my sacred and eternal duty to > Treat Agora Right Good. > > So, yeah. I'm Greg. Or maybe I'm sending messages on Greg's behalf. See > recent CFJs. > > As most of you accurately surmised, Greg's messages were generated by > GPT-2, specifically a version of GPT-2 fine-tuned with Agoran mailing list > logs since 2014. (The 2014 date is largely arbitrary—I might have been able > to go a bit further back, but using much more data ran into resource > limitations.) > > Greg was implemented using a combination of shell scripts, commands in my > shell history, python scripts, a Google Colaboaratory Notebook, and me > manually copy/pasting messages around. Notably, I manually pasted in and > hit send on each message. I did this primarily because figuring out email > APIs sounded like a PITA, but also because I wanted to be able to pull the > plug in case it said anything horrific. I did, however, do as much as I > could to avoid injecting my free will into the process. I operated off of > two rules: for each message to the public forum, I would run a python > script which had a 10% chance of invoking GPT-2 to generate a reply, which > I would send verbatim. (GPT-2 barfs on overly-large input data, so I > included a failsafe that automatically removed old messages in the thread > until the input was small enough to work. Some messages (like the rulesets) > were far too big on their own, resulting in the code generating "replies" > without any context.) Additionally, each day after the first (which looks > like is just going to mean "today"), I ran a script which had a 50% chance > of generating a brand-new proposal. Had I been aware of CFJ 3790, I might > have actually went to the trouble of having it send the messages > automatically after generating them. > > I did "intervene" twice: for the registration message, I specifically > asked GPT-2 to generate a message to BUS with a subject of "BUS: > Registration". In my testing, this had about a 75% chance of generating a > message that was a somewhat plausible attempt to register. Unfortunately, I > got unlucky and my first "real" attempt to generate a registration message > resulted in something completely random (a proposal, I think), so I > generated a second one and sent that one. In another case, I discovered a > bug with the large-input failsafe (turns out, GPT-2 can barf by silently > returning the input with no additional output, or by throwing an exception; > I was only handling the first case), so I fixed the bug and re-ran the > generation. In every other case, I mechanically copies messages back and > forth, following the plans I had made before sending the first message, > without attempting to impose any editorial control. > > In my testing, Greg did occasionally borrow other people's signatures, but > I didn't expect it to be this common. I considered preventing it from doing > this by removing signatures from the training data (so it would never learn > to include them), but I thought it was rare enough and amusing enough that > it wasn't worth removing. In retrospect, I probably should have removed > them. > > In my testing, I ran into several outputs that were interesting enough to > save so I could show them later. Here are links to them: > > A proposal for something vaguely resembling a functional auction mechanic: > https://gist.github.com/Gaelan/e7f7d3fc48c1abd08f0afb8049077acb > A made-up FLR excerpt containing an interesting-sounding royalty mechanic: > https://gist.github.com/Gaelan/8d092a17ed9c210685a4f4dd1e622ae2 > Another ruleset excerpt, containing the core rules of an alternate > universe Agora: > https://gist.github.com/Gaelan/ee631f9f97b53df8483e342ef36b6618 > A batch of attempts at starting new threads, with varying quality: > https://gist.github.com/Gaelan/0c027e3f5b97dab700182aa663401f47 > A fake rule called the "Register of Proposals", which looks like a > semi-plausible implementation of proposals in an alternate-universe Agora: > https://gist.github.com/Gaelan/0c6853f500799c5190a0a1ef474b098b > > I'd be happy to share the model at some point, but its a bit of a pain—I > think it's 1.5 GB—so I'm not sure how best to do that. In the meantime, I'm > happy to try out any inputs y'all are curious about. Also, From headers > were included in the training data, so I should be able to ask it to > generate message from specific Agorans. That might be fun. > > Happy to answer any questions, of course. > > Gaelan > "Nothing in a democracy is sacred, and nothing in a democracy is > sacrosanct. That's not to say it's never been broken, but it's been >
Re: DIS: Coming clean
> On Jul 21, 2020, at 5:22 PM, N. S. via agora-discussion > wrote: > > Given that gaelan exercised much control over greg by registering him amd > by curating his training data, There's a fair argument about forcing registration, but I'd argue that all of us, via genetics and education, had our training data chosen by other people. Gaelan
Re: DIS: Coming clean
On 7/21/2020 8:14 PM, Gaelan Steele via agora-discussion wrote: I'd be happy to share the model at some point, but its a bit of a pain—I think it's 1.5 GB—so I'm not sure how best to do that. In the meantime, I'm happy to try out any inputs y'all are curious about. Also, From headers were included in the training data, so I should be able to ask it to generate message from specific Agorans. That might be fun. I would definitely be interested in seeing what it comes up with for specific Agorans. -- ATMunn friendly neighborhood notary and Czar of Russia :)
Re: DIS: Coming clean
> On Jul 21, 2020, at 5:47 PM, ATMunn via agora-discussion > wrote: > > On 7/21/2020 8:14 PM, Gaelan Steele via agora-discussion wrote: >> I'd be happy to share the model at some point, but its a bit of a pain—I >> think it's 1.5 GB—so I'm not sure how best to do that. In the meantime, I'm >> happy to try out any inputs y'all are curious about. Also, From headers were >> included in the training data, so I should be able to ask it to generate >> message from specific Agorans. That might be fun. > > I would definitely be interested in seeing what it comes up with for specific > Agorans. > > -- > ATMunn > friendly neighborhood notary and Czar of Russia :) [posting links here for the benefit of non-discord folk] Here are some messages "from" to various Agorans: https://gist.github.com/Gaelan/e6f799142f98fa64f43344257b6b0ad5 I have no idea how much the From header actually affects the output, but it's amusing reading nonetheless. Also, here are some "proposals" from Greg: https://gist.github.com/Gaelan/26926a800092190db038140fa781a5b2 Gaelan
Re: DIS: Coming clean
On Tue, 2020-07-21 at 19:30 -0700, Gaelan Steele via agora-discussion wrote: > [posting links here for the benefit of non-discord folk] > > Here are some messages "from" to various Agorans: > https://gist.github.com/Gaelan/e6f799142f98fa64f43344257b6b0ad5 > I have no idea how much the From header actually affects the output, > but it's amusing reading nonetheless. The main correlation I notice (in this output and other output) is that the bot likes to have G. object to things. Bot!Aris and Bot!ATMunn also have a noticeably different writing style, although I'm not sure how well it matches the actual players. I think that Bot!Cuddlebeam might be an interesting test to run (Cuddlebeam has a fairly distinctive style in real life; will the bot pick it up?) -- ais523
Re: DIS: Coming clean
Just added CuddleBeam. There are definitely a few in there that seem somewhat cuddlebeam-y. Gaelan > On Jul 21, 2020, at 8:25 PM, ais523 via agora-discussion > wrote: > > On Tue, 2020-07-21 at 19:30 -0700, Gaelan Steele via agora-discussion > wrote: >> [posting links here for the benefit of non-discord folk] >> >> Here are some messages "from" to various Agorans: >> https://gist.github.com/Gaelan/e6f799142f98fa64f43344257b6b0ad5 >> I have no idea how much the From header actually affects the output, >> but it's amusing reading nonetheless. > > The main correlation I notice (in this output and other output) is that > the bot likes to have G. object to things. > > Bot!Aris and Bot!ATMunn also have a noticeably different writing style, > although I'm not sure how well it matches the actual players. > > I think that Bot!Cuddlebeam might be an interesting test to run > (Cuddlebeam has a fairly distinctive style in real life; will the bot > pick it up?) > > -- > ais523 >
Re: DIS: Coming clean
Does bot R. Lee (or V.J. Rada) have anything cool to say haha. On Wed, Jul 22, 2020 at 1:57 PM Gaelan Steele via agora-discussion < agora-discussion@agoranomic.org> wrote: > Just added CuddleBeam. There are definitely a few in there that seem > somewhat cuddlebeam-y. > > Gaelan > > > On Jul 21, 2020, at 8:25 PM, ais523 via agora-discussion < > agora-discussion@agoranomic.org> wrote: > > > > On Tue, 2020-07-21 at 19:30 -0700, Gaelan Steele via agora-discussion > > wrote: > >> [posting links here for the benefit of non-discord folk] > >> > >> Here are some messages "from" to various Agorans: > >> https://gist.github.com/Gaelan/e6f799142f98fa64f43344257b6b0ad5 > >> I have no idea how much the From header actually affects the output, > >> but it's amusing reading nonetheless. > > > > The main correlation I notice (in this output and other output) is that > > the bot likes to have G. object to things. > > > > Bot!Aris and Bot!ATMunn also have a noticeably different writing style, > > although I'm not sure how well it matches the actual players. > > > > I think that Bot!Cuddlebeam might be an interesting test to run > > (Cuddlebeam has a fairly distinctive style in real life; will the bot > > pick it up?) > > > > -- > > ais523 > > > > -- >From R. Lee
Re: DIS: Coming clean
(and can you add me if possible) On Wed, Jul 22, 2020 at 4:06 PM N. S. wrote: > Does bot R. Lee (or V.J. Rada) have anything cool to say haha. > > On Wed, Jul 22, 2020 at 1:57 PM Gaelan Steele via agora-discussion < > agora-discussion@agoranomic.org> wrote: > >> Just added CuddleBeam. There are definitely a few in there that seem >> somewhat cuddlebeam-y. >> >> Gaelan >> >> > On Jul 21, 2020, at 8:25 PM, ais523 via agora-discussion < >> agora-discussion@agoranomic.org> wrote: >> > >> > On Tue, 2020-07-21 at 19:30 -0700, Gaelan Steele via agora-discussion >> > wrote: >> >> [posting links here for the benefit of non-discord folk] >> >> >> >> Here are some messages "from" to various Agorans: >> >> https://gist.github.com/Gaelan/e6f799142f98fa64f43344257b6b0ad5 >> >> I have no idea how much the From header actually affects the output, >> >> but it's amusing reading nonetheless. >> > >> > The main correlation I notice (in this output and other output) is that >> > the bot likes to have G. object to things. >> > >> > Bot!Aris and Bot!ATMunn also have a noticeably different writing style, >> > although I'm not sure how well it matches the actual players. >> > >> > I think that Bot!Cuddlebeam might be an interesting test to run >> > (Cuddlebeam has a fairly distinctive style in real life; will the bot >> > pick it up?) >> > >> > -- >> > ais523 >> > >> >> > > -- > From R. Lee > -- >From R. Lee
Re: DIS: Coming clean
at 5:14 PM, Gaelan Steele via agora-discussion wrote: Alright, people seem to have started to get annoyed, so I believe continuing this experiment would violate my sacred and eternal duty to Treat Agora Right Good. I found it quite entertaining, fwiw!
Re: DIS: Coming clean?
omd, how did I get on your lists? I'm not even a current player. -- OscarMeyr
Re: DIS: Coming clean?
On Wed, Aug 24, 2011 at 4:54 PM, Benjamin Schultz wrote: > omd, how did I get on your lists? I'm not even a current player. The lists are based on Murphy's CotC database, and include all the old CFJs in that database. :)
