Angus @anguslm@qoto.org

@malin Nice. :D

@malin I can see that being useful though, especially if you can have versions like find-and-replace-all-x-in-brackets.

@malin Ah, you have to be inside the brackets already, I thought it was to take you to the next bracket and then alter it.

@malin I think i changes it to insert mode rather than completely quitting. What's the format for the ci[ command? ci[ then space then replacement word then enter? or without the space? or something else?

@malin It's a plugin, IdeaVim v0.51

@malin Oh cool, I can get fA df9 and dF. to work, but not the others. With the first one, hitting i exits VIM.

@malin Overleaf, what do you use?

@malin Yep, wrote my final year project in it. :D It’s wonderful.

@malin Java, and yeah, I just finished a small project and decided at the end that I wanted to make all the inputs case invariant by replacing each instance of equals with equalsIgnoreCase. Couldn’t help but smile at how annoying that would have been to do manually.

@malin Ooh, that's cool. Hadn't seen that before. BTW the code editor I'm using at work has VI on it and it's already been quite useful. Combined with a macro, the efficiency is wonderful.

@malin
Yeah, here's the handout:
https://docs.google.com/document/d/1CGHM-bpYBpfz730z7JeZtMdHa6YwTNuE6q1OS9w4kZ8/edit?usp=sharing

I've also got a new voting system idea, which I'm calling Burn From Both Ends. The idea is to combine STV and VTS by taking the percentage of people who put a candidate as their first choice, then subtracting the percentage who place them last to give them a total score. The candidate with the lowest score is then removed and the scores are recalculated and this is repeated until only one candidate remains.

So far it's returned the Condorcet winner for the two tests I used to break STV and VTS, and it's also shown a noticeable difference in that a polarising candidate (one who is always at the extreme end of each voters preference ranking) doesn't always come first or last. It seems part of Arrow's theorem is that this always follows from unanimity and independence of irrelevant alternatives, and leads to dictatorship. I'm pretty sure BFBE doesn't violate unanimity, so I guess it must violate independence of irrelevant alternatives, which is probably the condition I'm most prepared to sacrifice anyway.

@malin One of the six skills for tech consultancy. I chose intellectual horsepower, which is a mixture of intelligence and doing things to make the most use of it. I used the Mbuti Pygmies as an example for the expanding-your-horizons section.

@malin Nice one, thanks. 👍

@malin Hey, do you remember where the tribe with no words for good and bad were from? I’m working on a presentation for work tomorrow and I think they might be useful as an example.

@malin I found an article which seems to show this condition: https://www.cs.cmu.edu/~arielpro/mfai_papers/lecture6.pdf
Section 4.1 onwards

From what I understand, if the loser of the vote can alter the result to their favour by tactical voting, the system is manipulable. If they can alter the vote but it results in a worse outcome, it's not pareto optimal. If they can't alter the vote it's dictatorial. But the example they provide has only two voters, which seems to be the cause of the problem as it's trying to create a voting system which can resolve ties, which seems unnecessary. If voters 1 and 2 were replaced by 60% and 40% of the electorate respectively, the fact that group 1 will get their preference (a) no matter which order group 2 put their preferences in, is not a problem as they are a majority. It doesn't work for two people because you can't really have a democracy of two.

@malin I also did some more thought experiments on voting methods. It turns out my method of judging systems by seeing who would win in each possible pairing and if one would beat all other candidates, they should be the winner, is called the Condorcet method. It doesn't provide a winner in all circumstances however as cyclic collective preferences can be created (Condorcet paradox) though I suppose you could have coalitions for Condorcet ties. That's not much use for referendums however, so I wanted to try to make a system that returns a winner for Condorcet paradoxes, but returns the Condorcet winner whenever there is one.

My reverse STV method (VTS?) fails at that as if you create a 40, 30, 20, 10% grouping with the 40% group hating A, and all the rest putting A as their first choice but splitting their last vote between the other 3, A will get eliminated in the first round despite being the clear Condorcet winner.

My current favoured method is to basically just take a Condorcet method and subtract the majority percentage of their losses from the majority percentage of their wins to give a final score. Highest final score wins.

@malin Oh damn, that is bad. My computer’s motherboard died a few months ago. Pretty expensive to fix. Just got a job offer for a tech consultancy position I applied to.

@malin Yep, 😀 how's it going?