I’m a long-time lurker. I prefer to sit back and skim through comment sections, passively absorbing, and over the years I’ve seen a fair number. After a while, you start to get a feel for their dynamics. Typically, a blog post plays out something like this:
- Blog author posts something.
- Long-time commenters pop by with their two cents.
- Their chatter starts to wander off topic.
- Someone pops by with a strong opinion that’s vaguely off-topic.
- This kicks up an argument, which gets ugly and spirals away from what the original post discussed.
There are exceptions, of course; endless threads have no topic to wander off of, and if the thread is obscure and the topic well-defined the comments can stay topical indefinitely. The comment community plays a large role in this, too. A small band of thoughtful regulars are a blogger’s dream, while a large number of over-opinionated randos can (and often do) ruin any thread. If acrimony starts to trump argument, even a small community can turn dysfunctional.
It doesn’t help that our tools are few, blunt and prone to breaking. Voting systems can be gamed, while banning users or keywords is an all-or-nothing affair that barely works. Allowing comments for a limited window sounds great, but it doesn’t allow the regulars to build up much of a conversation. Banning all comments kills off the local community.
Aaaaand that’s about the extent of it. Maybe someday I’ll create a browser plugin that provides a personal ranking system, which automatically mutes or even hides users based on how you’ve rated their prior comments, but that’s low in my queue.
How am I going to encourage that small, thoughtful community to form? Here’s my current plan:
- Regular blog posts don’t allow comments, unless justified by the contents. This prevents comment threads from spiraling away.
- The “Community” post is an endless thread. Only one of them is active at a time.
- To provide a little structure, links to the regular blog posts will get dropped into the Community post as they go public. These can be ignored.
- The Community post will be linked somewhere along the side menu, but it won’t otherwise be advertised. This should keep the randos to a minimum, but without throwing out regulars too.
- The top of the Community post will outline the moderation rules in play. Those rules stay consistent over the lifetime of the Community post. If I want a significant change, the current Community post is locked and a new one is created. The new will link to the old, and vice-versa.
The first Community post is the one you’re reading right now.
The initial mod rules are fairly ill-defined and flexible, to keep the rules lawyers at bay. My guiding principle is to maximize information; it takes time and energy to read a comment, so you should try to convey as much as possible, as clearly as possible, in the least space. Critiques beat opinions, evidence wins over assertion. Strict enforcement of that doesn’t work with endless threads, but it’s still the ideal you should keep in the back of your mind.
The corollary is another matter, though: quit it with the oppressive language. If you lack the creativity to think up an alternative to “crazy,” you shouldn’t be posting here. Violence in any form is a no-no, and both stalking and harassment are low-grade forms of violence.
Speaking of which, I’d like to swipe an idea from football. They have a carding system to handle misconduct, which I think works in this context too. If you’re handed a yellow card, that’s a warning for unsportsmanlike conduct. A red card gets you banned from this thread, though not the entire blog. A black card is a permanent ban.
Got it? Then game on!
A belated thank you for the “Let’s Talk Websites” post – very informative.
256?
@ ^
Why are you prejudiced against 4^4?
@#^~
To navigate the story is to become one with it.
Divinity is the driver of chi. Peace requires exploration.
Well, the page turning after 256 comments is novel. I would have tried reading FTB during the hiatus but if you didn’t already have an IP address for the server it was a little hard to find it afterwards.
I remember the time Richard Dawkins managed to tweet a picture of one of his books where the QR code on the blurb had been altered to be the fourteen words. He (and the other surviving horsemen Daniel Dennett, and Sam Harris) all lurched to the right, and a good deal of the crapulent followers of movement atheism have followed suit, to the point where they’re echoing fash rhetoric either unwittingly, or hoping perhaps that no one is going to call them on.
To (non-existent) hell with all of them.
This time all I have is a serving spoon with drain holes. It works.
We are closer to the End than the Beginning.
https://en.wikipedia.org/wiki/Harold_and_the_Purple_Crayon_(film)
Weird, the name meant nothing to me until I saw what the character looked like. Some dim memory burbled up, I think of one of the animated adaptations. I still have zero memory of the children’s book.
As for the 256 comment limit, I thought that was obvious? Computers used to have all sorts of byte sizes, for instance this manual references byte lengths ranging from one to six bits. For the past three decades, everyone’s settled on eight bits to a byte. It lines up with ASCII (released in 1963), which heavily influences text file formats to this very day.
Math-wise, 2 is the only even prime in existence, the smallest Sophie Germaine prime, and it’s also the core of Mersenne primes. As it so happens, \(\sum_{n=0}^\infty 2^{-n} = 2\) . It’s the smallest number you can use to make a number system with, and the only two digits you need are zero and not-zero. If you’re doing anything related to numbers, base 2 is a natural choice. And there’s something elegant about \((2^2)^{(2^2)}\).
Where are you?
Busy and distracted, alas. It’s also kinda frustrating to ring the alarm bell about transphobia and yet see it not only march on but thrive; you can only push that rock up the hill for so long before boredom and frustration make it easy to move on to other things.
My writing was always a method I used to vent and cope, however. And I suddenly find myself with a lot to vent about. We’ll see how long that lasts.
OK.
I imagine you’ll have plenty to vent about, going forward. *weep*
UoC
???
Xanthë: Quite. My to-blog-about list currently sits at half a dozen entries, and that doesn’t include any of the bazillions of drafts I have kicking around.
chigau: …. AoC? I am confuzzled.
University
of
C……
No. It doesn’t.
Did you know that you say “sort of” alot?
Sort of? I know I have a few writing “ticks” that I’ve never been able to rid myself of, but “sort of” is one I haven’t noticed before.
As for your prior question, erm, University of Caldas? Am i doing this right?
I noticed it a lot on one of the sort of podcasts you did with PZ and GAS.
I just 🖤 elections.
Don’t you?
I voted.
Did you?
Of course! It’s literally the least you can do in a democracy. I’m a fan of compulsory voting, in fact.
Me, too.
If my, “Me, too.” is a duplicate comment, where is it?
Sitting in my spam folder, for unknown reasons. I’ve fished it out and approved it.
*cough*cough*cough*
OK.
Now it’s raining. That will help.
They are putting-up the noise tent. It seems a bit early.
And so it begins…
fine
Parades are weird.
Today is Bastille Day.
I wonder if someone will stage a jailbreak from one or more ICE detention centers.
That was a right pathetic excuse for an OHMYGODOHMYGODTHUNDERSTORM.
in my part of town
Gawd, it feels like we’ve been having a second spring flood in the middle of summer over here. The thunderstorm warnings have been near-constant.
Owlmirror:
I put the odds at 100%. ICE is not attracting the best and the brightest, plus no security system is completely secure. I’m more interested in whether or not the news of the jailbreak will be covered.
Oh, and I’m not just here to catch up on comments. That blog I just posted, on LLMs playing Battleship? It didn’t go as I planned.
Originally there was going to be a round three against ChatGPT, but by the time I had fired two thousand words into the draft I realized there was no need. The ending was originally going to be teasing that third round, but writing the section on ChatGPT’s text responses really soured me on my original ambiguous ending. Even if ChatGPT was following a strategy, its own words were contradictory and ever-shifting, you could read pretty much anything into them. Intuitively, that led me to write the ending I wrote, which wraps back around to Tracie Harris’ podcast and points out the same contradictions in Theo’s output. It’s a bit meaner on Harris than I intended, as a result; anyone can be fooled by bullshit, and LLMs are especially good at flinging bullshit around.
That change really undercut the original title. In hindsight, I really should named it “Mixed Signals.” Bah, better luck next time.
Incidentally, I am still interested in doing a third round. I know exactly what I would do, I even have a fair bit of programming code written to assist me. And with ChatGPT 5 just released, another over-hyped disappointment, I have a new and “improved” opponent. Dunno if it’ll be my next post, but I am sorely tempted.
How is “AI” and “LLM” and “ChatGPT” different from “a computer program”?
Let’s do a lightening round!
computer program: A structured recipe executed on a computer. Typically done via a von Neumann architecture, which has inputs, outputs, memory, and a state of execution. Other execution models exist, such as the lambda calculus, but in practice these are emulated on a von Neumann architecure.
Machine learning: Using a computer program to estimate the parameters of a function.
Model neuron: A simplification of a biological neuron. Typically consists of evaluating a linear equation, which multiplies the input numbers by weights to get the output (ie. A = ax + by + cz + …) combined with a non-linear transform (A -> 0 if A < 0 else A). Machine learning is often used to determine the weights.
Neural network: Two or more model neurons connected to one another. A “single-layer” network connects incoming numbers to the inputs of one or more neurons, with the output of those neurons being the output of the system. A “multi-layer” or “deep” system redirects the outputs of those input-facing neurons into another collection of neurons. A “fully connected” network connects every neuron in one layer to every neuron in the next layer.
AI: A marketing term created in the 1950’s by scientists to earn grant money for studying neural networks.
LLM: Large Language Model, a type of neural network aimed at taking in freeform text and outputting the same.
Transformer: A neural network that “transforms” a sequence of inputs into a vector (think of an arrow, where the length and direction of the arrow matters) in some abstract multi-dimensional space, or transforms a vector back into a sequence of outputs.
Generative Pre-trained Transformer: GPTs are a specific class of transformers that take in a sequence of symbols, usually broken up into “tokens” consisting of one or more symbols grouped together, and outputs the probability that a specific token will follow that sequence.
ChatGPT: The brand name for a specific GPT, which was trained to give outputs as if the input was one or more text messages from an informal social media “chat.”
The TL;DR is that a “computer program” is somewhat abstract and generic, while neural networks are just a bunch of matrix multiplications and functions chained together.
in a computer
“LLMs Can’t Code”
me neither
I think you meant “apophenia” rather than “pareidolia” — as the WikiP article itself notes, “apophenia” is the general term for seeing patterns and connections; “pareidolia” is the specific case of visual phenomena.