alias spoken_squiggle -> spoken_scribble

CyclicallyCynical said in forum #418207:

Scribbles are completely random, uncontrolled, and very chaotic. This word also refers very specifically to children’s doodles, like with crayons. You’d never call a children’s doodles “squiggles”, virtually everyone calls them scribbles.

See, this is why it was deprecated in the first place. Having a tag refer to multiple things at once is not desirable. While I did say you should submit a BUR to undeprecate the tag if that's what you want, I think a better solution would be to make a new scribble tag with a qualifier in parentheses. Feel free to come up with one; I have no clue what qualifier would work for this.

3: Using statistics, this calculates to the average % of mistagged posts being 86.67% with a 95% confidence interval of roughly 75% to 99%. In other words, only 30 images are needed to determine that there’s only a 5% chance the actual percentage of mistagged posts is under 75%. If you want to try being smart, consider using math instead of your intuition. If you disagree with my proof, you can go and look at more images and honestly count them yourself.

You want some math? Here you go:

5: I think this is what needs to happen. This whole fuckfest is due to some smoothbrained activities that had to have occured for the word scribble to be replaced with squiggle - like wtaf happened there?!

6: I was asking people to explain why they didn’t read my whole post before downvoting, or why they didn’t leave a comment explaining why they downvoted. What does this whole forum even exist for, if not to discuss topics? Why are approvers downvoting my post, and then don’t explain shit to me, when I’d just posted my FIRST post? If they don’t help new contributers understand how tf to use this thing, who will? Buddha? Jesus of Nazareth? Diddy?

CyclicallyCynical said in forum #418210:

Eh, I don’t think these folks seem like anything I post. I feel like I’m getting trolled or something, I don’t know.
I’ll try and post it, but I can already smell my charred corpse burning at the stake from over that hill.

You need to stop taking these BURs so personally. Downvoting just means they disagree. They're not trolling you. I've had near-unanimous downvotes on some of mine with just one or two users explaining why. I've even downvoted some of my own BURs. Their purpose is to prompt discussion, not to make you feel validated.

The ones not answering may believe their point was already made by others and see no need to repeat what was already said. We like to keep things concise and efficient.

Also, you're not in any position to criticize what other users do when you act overly aggressive like this. These rants of yours do nothing to strengthen your arguments and will reduce the likelihood others will listen to you. Next time you post here, take a moment to compose yourself first.

Blank_User said in forum #418245:

See, this is why it was deprecated in the first place. Having a tag refer to multiple things at once is not desirable. While I did say you should submit a BUR to undeprecate the tag if that's what you want, I think a better solution would be to make a new scribble tag with a qualifier in parentheses. Feel free to come up with one; I have no clue what qualifier would work for this.

You want some math? Here you go:

The Math

Let's take your calculation again (we'll assume 95% confidence for everything), but this time, we'll say your sample of 30 contained only 4 mistagged posts instead, so about 13.33%. If we plug the numbers into this calculator, we get a confidence interval of 1.169% to 25.5%, meaning according to your logic, there's only a 5% chance of the percentage of mistags being above 25.5%.

Now let's say you picked another sample of 30 and exactly half of them were mistags. That gives a confidence interval of 32.11% to 67.89%, so there should only be a 5% chance of the mistag percentage being outside of that range.

Are you beginning to see the problem?

The confidence interval is different for each sample. It does not tell you what the probability of the mean being a certain value is. What 95% confidence actually means is that when you take several random samples of the same size using the same method, 95% of those confidence intervals are expected to contain the mean while 5% are not. But you only took one sample, one that wasn't even random, and tried to draw conclusions about the entire dataset.

I'll admit I had to look some of this up (this explains things pretty well) since it's been at least a decade since my last Statistics class, but I've had enough exposure to math and science to sense when something isn't adding up. That's how intuition enters the equation.

My brother in Christ.
The 25.5% is literally just the inverse of the 75% I said.
Also, a 5% chance of the mistag being outside of the 32.11% to 67.89% range sounds right.
So no, I'm not seeing anything wrong with the math, at all.
The thing with math is, your intuition is not always right. It's okay to admit you were wrong.

Blank_User said in forum #418245:

You need to stop taking these BURs so personally. Downvoting just means they disagree. They're not trolling you. I've had near-unanimous downvotes on some of mine with just one or two users explaining why. I've even downvoted some of my own BURs. Their purpose is to prompt discussion, not to make you feel validated.

The ones not answering may believe their point was already made by others and see no need to repeat what was already said. We like to keep things concise and efficient.

Also, you're not in any position to criticize what other users do when you act overly aggressive like this. These rants of yours do nothing to strengthen your arguments and will reduce the likelihood others will listen to you. Next time you post here, take a moment to compose yourself first.

I believe I've been extremely composed this entire time. Could you point to a message I made that was unhinged?

CyclicallyCynical said in forum #418247:

I believe I've been extremely composed this entire time. Could you point to a message I made that was unhinged?

forum #417422

CyclicallyCynical said in forum #418246:

My brother in Christ.
The 25.5% is literally just the inverse of the 75% I said.
Also, a 5% chance of the mistag being outside of the 32.11% to 67.89% range sounds right.
So no, I'm not seeing anything wrong with the math, at all.
The thing with math is, your intuition is not always right. It's okay to admit you were wrong.

You're still not getting it. It's not your math that's wrong. It's how you're interpreting the results that's wrong. Your conclusion depends on what posts happened to be in your sample. Again, if we assume your understanding of confidence intervals is correct (it isn't) and we combine the results from both of our samples, we get these two truth statements:

• There is a 5% chance that the percentage is under 75% (your sample).
• There is a 95% chance that the percentage is under 25% (my sample).

Both of these statements can't be true at the same time. Confidence accounts for all the samples, not just one or two. We don't know which of those two samples is among the 95% of samples that contains the mean. It possible that neither of them are in the 95%.

By the way, if you were lucky enough to see mistags in all 30 of your sample posts, your confidence interval would be practically 0 to 0, meaning you could've concluded with your flawed reasoning that every single use of the tag was wrong. Conversely, if all 30 were tagged correctly, then the interval would be 1 to 1, which according to you would mean the tags are correct in all 7,103 posts. It should be pretty clear how absurd it would be for both of those to be true at the same time.

Please read the article I included before. It has penguins in case you need more incentive.

HyphenSam said in forum #418248:

forum #417422

Yikes. Well, at least there's some improvement, even though that wasn't exactly a high bar to clear.

Still, I think they've been given enough warnings. If they keep acting like this, I'll give them a Negative feedback if someone else doesn't beat me to it.

Blank_User said in forum #418249:

You're still not getting it. It's not your math that's wrong. It's how you're interpreting the results that's wrong. Your conclusion depends on what posts happened to be in your sample. Again, if we assume your understanding of confidence intervals is correct (it isn't) and we combine the results from both of our samples, we get these two truth statements:

• There is a 5% chance that the percentage is under 75% (your sample).
• There is a 95% chance that the percentage is under 25% (my sample).

If you want to be exact, using a 95% confidence interval based on a sample of 30, the estimated mistag rate is 86.67% with an interval of approximately 74.67-98.67%. This means that if we repeatedly sampled 30 images and constructed intervals this way, 95% of those intervals would contain the true proportion of the binomial distribution.

Your point of the interval being wide due to the small sample size is correct, so you could say my evidence weak. But to meaningfully push back against my claim, you'd need to review more images (~100-200) so any conflicting data would be statistically significant.

Both of these statements can't be true at the same time. Confidence accounts for all the samples, not just one or two. We don't know which of those two samples is among the 95% of samples that contains the mean. It possible that neither of them are in the 95%.

Of course not, as they come from different hypothetical examples. Your scenario was the exact inverse of mine. I wasn’t claiming that both intervals were simultaneously true; rather, I was illustrating that the same logic applied to inveresed data leads to reversed conclusions. You’re right that a sample of 30 produces a wide interval, but your critique only becomes materially relevant if you produce conflicting data. Otherwise, we can acknowledge the limitation of the 30 sample size and move forward, review more samples, or just agree to disagree.

By the way, if you were lucky enough to see mistags in all 30 of your sample posts, your confidence interval would be practically 0 to 0, meaning you could've concluded with your flawed reasoning that every single use of the tag was wrong.

Conversely, if all 30 were tagged correctly, then the interval would be 1 to 1, which according to you would mean the tags are correct in all 7,103 posts. It should be pretty clear how absurd it would be for both of those to be true at the same time.

This is a textbook straw man, and you know it. You've entirely misrepresented my original claim.
You've reframed my claim to "I can prove things about the dataset with 100% certainty using a small sample size." That's not what I said.
Also, your application of Clopper-Pearson intervals for binomial data is incorrect.

What I said: "this calculates to the average % of mistagged posts being 86.67% with a 95% confidence interval of roughly 75% to 99%"
In other words, there's a 95% chance of repeated experiments (sampling 30 random images) yielding ~86.67(+-12)% of images with mistags.
My original statement is a bit vague, but still correct. A better explanation would've been that if we repeated the experiment, 95% of the resulting CIs would contain the true proportion of the binomial distribution.

Your application of Clopper-Pearson intervals for binomial data is incorrect.
Even if all 30 sampled posts were mistagged as you say, a proper 95% confidence interval would not be 0-0 (100%?), it would be ~88-100%. A Similarly, 0/30 would not give 1-1 (0%?), but ~0-12%. The numerical premise of your argument here is incorrect.
The confidence interval (+-%) comes from the sample size. With 30 samples, this is ~12%.
So, the range would not be 1-1 (100%) or 0-0 (0%) in either case, it would be 88-100% and 0-12%, respectively.

To your statement: “It’s absurd that both conclusions could be true at the same time.”
As I stated above, this was never my argument. Your intervals come from different (inversed) hypothetical samples.

Please read the article I included before. It has penguins in case you need more incentive.

Yikes. Well, at least there's some improvement, even though that wasn't exactly a high bar to clear.

So, calling someone a fucking asshole on a porn site of all places is me going "unhinged"?
Aren't there 10,000s of anal images on the site of assholes getting fucked already? :)
Jokes aside, of course I'm calling someone an asshole if they leak a 'private' message I sent them.
This was the message, by the way:
"Hey there, did you actually read what my post suggested or just downvote it for no reason?
I'd like to understand what your issue with it is so I can fix it :)"
Please tell me how this comes off as "unhinged." At worst, I'm being a smartass.
This site hosts content I'm not going to even indirectly reference here.
I seriously doubt anyone here would consider this "unhinged".

If he was truly hurt by that one message, he could've just ignored it, or told a mod.
But instead, he chose to be confrontational and stir the shit-pot by posting private messages in public.
From my perspective, I was being more respectful to him by taking things to DMs instead of confronting him publicly.
Additionally, he went out of his way to insult me for spending 6 hours trying to make my first BAR presentable later for no reason. That is the context to the message you've linked.

Still, I think they've been given enough warnings.
If they keep acting like this, I'll give them a Negative feedback if someone else doesn't beat me to it.

Who are you talking to when you start rudely referring to me in the third person here? I've messaged a few different supporters/approvers with questions related to these BURs and threads, including those who participated in their discussion. No mods has given me a "warning" for anything other than the message I referenced above - which I immediately acknowledged and have since respected. If I'm breaking some other rules I'm not aware of, you can point me to them and I'll read them. On the contrary, I've encountered a ton of otehr very kind and helpful user who've taught me about tagging/BURS/etc., which I greatly appreciate! Like I've stated, I'm actively learning more about the site to help contribute more and meaningfully.

I've been firm, sarcastic, and maybe smart with some of my language, but nothing like what you're insinuating. Your response to being debated should be to bring better information and a better arguments to the discussion, not threaten to give me "Negative" feedback without clear justification. If you disagree with what I'm saying, then elucidate why. If this isn't the place to do exactly that, then where is? Where else on the site is it appropriate to have debates on BURs? If you're not allowed to reply to messages in your own threads, where else would you be? If you think I'm wrong but can't be bothered to explain why, that's fine too. Nobody is forcing you to return to this thread.

thought i was done with statistics after college. im not reading all that

also excuse you this is NOT a porn site. it's actually a site where random people can sign up to upload terrible scat images and get upset when they're not approved for some weird reason.

im not going to respond to or read this thread anymore probably. i keep zoning out whenever i try to read too many paragraphs in one post with insufficient paragraph breaks. my adhd probably

Ylimegirl said in forum #418254:

thought i was done with statistics after college. im not reading all that

also excuse you this is NOT a porn site. it's actually a site where random people can sign up to upload terrible scat images and get upset when they're not approved for some weird reason.

im not going to respond to or read this thread anymore probably. i keep zoning out whenever i try to read too many paragraphs in one post with insufficient paragraph breaks. my adhd probably

Yea it's pretty funny to me that people are complaining about me using the words "fuck (sex)" and "asshole (anus)" when there is literaly scat, guro, toddlercon on here....... like give me a fucking break.

This is getting wildly out of hand, off-topic and personal for something that the average user doesn't give two shits about.

WRS said in forum #418277:

This is getting wildly out of hand, off-topic and personal for something that the average user doesn't give two shits about.

Yeah. It doesn't matter whether they're misusing confidence intervals in their methodology when the main argument itself is flawed.

Suppose we assume that 85% to 90% mistag percentage for spoken squiggle is correct. If the goal is to tag scribbles and squiggles separately, why would we want to alias spoken squiggle to spoken scribble? That would result in 10% to 15% of posts being mistagged, and with the alias in place, those mistags can't be fixed.

Blank_User said in forum #418303:

Yeah. It doesn't matter whether they're misusing confidence intervals in their methodology when the main argument itself is flawed.

Suppose we assume that 85% to 90% mistag percentage for spoken squiggle is correct. If the goal is to tag scribbles and squiggles separately, why would we want to alias spoken squiggle to spoken scribble? That would result in 10% to 15% of posts being mistagged, and with the alias in place, those mistags can't be fixed.

I just read the BUR command page again and you’re 100% right about future mistags.
I didn’t realize you could remove tags with update. That makes alias’ing make no sense.

This would be a better BUR:
update spoken_squiggle -> spoken_scribble -spoken_squiggle

I’ll put the request through and see what people think

BUR #54775 has been rejected.

mass update spoken_squiggle -> spoken_scribble -spoken_squiggle

This should fix the mistagged posts without causing future posts to be mistagged
Nevermind then, fuck it. Keep all the posts mistagged then.

The bulk update request #54723 (forum #418103) has been rejected by @Sistercynical.

The bulk update request #54775 (forum #418334) has been rejected by @Sistercynical.

how many of these are you going to make

FubukiKai said in forum #418338:

how many of these are you going to make

The most recent one actually makes sense if you read it, but sure.
Seems like everyone wants to keep the posts mistagged, atp I don’t give a shit anymore.
All of the squiggle and scribble tags are beyond salvaging, so fuck it at this point.

Sistercynical said in forum #418339:

The most recent one actually makes sense if you read it, but sure.
Seems like everyone wants to keep the posts mistagged, at this point I don’t give a shit anymore.

You still haven't justified why there should be two tags when they are often used interchangeably a spoken scribble and a spoken squiggle indicate the same thing.

You also haven't solved the fact that scribble was deprecated for being ambiguous forum #278659, it stands to reason if scribble is ambiguous then spoken_scribble could equally be mistagged on images like post #2112942

zetsubousensei said in forum #418341:

You still haven't justified why there should be two tags when they are often used interchangeably a spoken scribble and a spoken squiggle indicate the same thing.

You also haven't solved the fact that scribble was deprecated for being ambiguous forum #278659, it stands to reason if scribble is ambiguous then spoken_scribble could equally be mistagged on images like post #2112942

Did I not say I don’t give a shit anymore? And btw scribbles and squiggles are NOT the same.
Instead of contributing to help solve the problem, you’re just pointing out why you didn’t like mine.
All of the scribble/squiggle tags are fucked beyond repair, I give up trying to help fix them.
If you come up with any possible solutions yourself, then give them. But I’m done.

Edit: Wrote this before seeing the latest crashout. Keeping it for informational purposes.

Sistercynical said in forum #418328:

I don’t see how my methodogy is incorrect, but I digress.
Would 10-15% of posts being mistagged not be better than ~85% being mistagged as they are now?
This BUR can be rejected if it doesn’t seem like it’s worth it, but it feels like a win/win either way

If I understand you correctly, your position is "I'd rather have scribbles and squiggles tagged separately, but if we can't do that, then we should use scribble tags for all of them since they're the most common." That's not a win-win. That's a compromise. But if that's what you're settling for, then we need to make sure it actually would lower mistags. If you get the percentage wrong, then switching them could increase mistags instead.

The flaw in your methodology I was referring to was assuming the true percentage had a 95% chance of falling within the range of a specific confidence interval. This explains what I'm talking about. For everyone else sick of reading about math, there's an easier way to understand this.

The smaller the sample, the less certain you can be that it's representative of the entire group. It's possible for the sample to have a much higher or lower mistag percentage than the entire group. Two people can take different random samples and get completely different results, but the true percentage would stay the same. That's why the admins wouldn't be convinced of your estimate. If the mistags were higher than usual in your sample and it ended up being about 1 in 5 being mistagged after those first 30, but the admins believed your conclusion was accurate, then the mistag percentage would increase from about 20% to about 80%. The result would be the opposite of what you want.

By the way, if you attempt another statistical analysis of a tag, consider using order:random when selecting your samples. You don't know what trends could be present when you sample the most recent posts.

But before any of that is possible, a replacement for scribble needs to be determined. The new tag would need a qualifier to distinguish it from child's drawings. In other words, scribble would remain undeprecated and we would have a new scribble_(???) tag with some qualifier in the parentheses. Squiggle would be aliased to this new tag, assuming we'd only be keeping one of them.

Just to be clear, this is not an endorsement. I'm only listing conditions your BURs would have to meet to have any chance of succeeding.

Blank_User said in forum #418348:

Edit: Wrote this before seeing the latest crashout. Keeping it for informational purposes.

If I understand you correctly, your position is "I'd rather have scribbles and squiggles tagged separately, but if we can't do that, then we should use scribble tags for all of them since they're the most common." That's not a win-win. That's a compromise.

Disagree, this is a win as it reduces mistags from ~85% to ~15%. Not sure what your point is exactly.
If you can agree there's a problem, but disagree with the move, then come up with a solution yourself instead of ridiculing the solution.

Also, this is what you're calling a crashout...?
"I don’t see how my methodogy is incorrect, but I digress.
Would 10-15% of posts being mistagged not be better than ~85% being mistagged as they are now?
This BUR can be rejected if it doesn’t seem like it’s worth it, but it feels like a win/win either way"

Which of those sentences qualifies as a "crashout"?

But if that's what you're settling for, then we need to make sure it actually would lower mistags. If you get the percentage wrong, then switching them could increase mistags instead.

The flaw in your methodology I was referring to was assuming the true percentage had a 95% chance of falling within the range of a specific confidence interval. This explains what I'm talking about. For everyone else sick of reading about math, there's an easier way to understand this.

The smaller the sample, the less certain you can be that it's representative of the entire group. It's possible for the sample to have a much higher or lower mistag percentage than the entire group. Two people can take different random samples and get completely different results, but the true percentage would stay the same. That's why the admins wouldn't be convinced of your estimate. If the mistags were higher than usual in your sample and it ended up being about 1 in 5 being mistagged after those first 30, but the admins believed your conclusion was accurate, then the mistag percentage would increase from about 20% to about 80%. The result would be the opposite of what you want.

I already said this in my above message:
"You’re right that a sample of 30 produces a wide interval, but your critique only becomes materially relevant if you produce conflicting data.
Otherwise, we can acknowledge the limitation of the 30 sample size and move forward, review more samples, or just agree to disagree."

Instead of talking about the BUR, you've started going down this rabbit hole trying to critique some math I used.
You continue to slander me unjustly, which I've very much not appreciated, and contributed nothing to this discussion.
This has become a waste of time. Like I said above twice already, I don't give a shit anymore. I'm never coming back to this thread.