SpiceWare's Blog

It is obvious Facebook is pushing an agenda and only allows posts that go along with the narrative without allowing others with different viewpoints a chance utilize free speech to speak their opinions and have an open and honest debate.  It seems the word compromise has disappeared from the public.  Not everyone is going to agree and squelching anyone with a challenging a point of view is dangerous and outright censorship.

 

I think I'll likely delete my account as well once I notify my immediate friends and family.  Thanks for the nudge to get me to do so.

Edited August 20, 2021 by tremoloman2006

22 minutes ago, tremoloman2006 said:

I think I'll likely delete my account as well once I notify my immediate friends and family. 

 

I did the same - told everybody they had about a week to download any photos I'd taken of them, and explained why I was doing so with the above screenshots.

Life is great without (un)social media. Trust me!  

The problem is AI moderators running on flawed overly generic algorithms have been put in the role of judge, jury, and executioner.

 

It was fine when AI was in charge of your spam filter.  If something accidentally got flagged as spam that wasn't spam you pulled out.  It's a different matter when 3rd parties are deciding how to censor or ban you based on these same flawed/incomplete routines.  Nobody wants to have to pay real people to do this sort of work, even to review cases.

 

 

The initial blocking is IMO not the main problem (there always will be overblocking). Not being able to resolve overblocking fast and easy is what kills the whole thing. But for the latter you need real people who understand your writing.

An AI moderator is fine, the volume is too much for a human to oversee in a timely manner.

 

No way to dispute the AI decision is the problem. Not a single "disagree" I submitted resulted in a reversal, nor even a response.

Yeah, that's pretty ridiculous, no doubt about it.  Especially since they really give you no explanation at all as to what caused the post to be removed, and good luck trying to protest it, that almost never results in a reversal.  Some of those may have been removed because of the external links. If there's one thing Facebook hates, is when you link to an external site that takes people away from Facebook. 

 

The only one that makes even a small amount of sense is the post that mentions COVID. But even then, I've seen so much misinformation about COVID that gets past Facebook's filters, I don't even know why they bother trying.

 

I don't blame you at all for abandoning Facebook.

 

 ..Al

13 minutes ago, Albert said:

Some of those may have been removed because of the external links.

 

I considered that, though when I noticed the Liftoff comment missing I did two more comments: one with just the text, and one with just the link.  Both posted without issue, and the people I was talking with immediately liked both comments so I know they weren't blocked.

 

 

7 minutes ago, SpiceWare said:

I considered that, though when I noticed the Liftoff comment missing I did two more comments: one with just the text, and one with just the link.  Both posted without issue, and the people I was talking with immediately liked both comments so I know they weren't blocked.

There's really no rhyme or reason to what they do, and since they don't share the actual reason posts like this are removed, all we can do is guess. It is super frustrating. I've seen all manner of posts removed where the rationale is not clear at all.

 

 ..Al

24 minutes ago, Albert said:

There's really no rhyme or reason to what they do, and since they don't share the actual reason posts like this are removed, all we can do is guess. It is super frustrating. I've seen all manner of posts removed where the rationale is not clear at all.

 

 ..Al

The reason they won't tell you why is they don't know why themselves.  They have a neural network they've "trained" by feeding it a bunch of posts marked "bad" (and hopefully some marked "good") until it correctly marks another bunch of posts.  The more similar your post is to the "good" training & testing posts, the more likely the AI will mark it correctly.  But if your post is more like the "bad" training & testing posts, then the AI is going to mark it incorrectly.  And there's no way to ask the AI "why is this post "bad"". 

 

Modern AI is a big black box - garbage in - gospel out.

6 minutes ago, EricBall said:

The reason they won't tell you why is they don't know why themselves.  They have a neural network they've "trained" by feeding it a bunch of posts marked "bad" (and hopefully some marked "good") until it correctly marks another bunch of posts.  The more similar your post is to the "good" training & testing posts, the more likely the AI will mark it correctly.  But if your post is more like the "bad" training & testing posts, then the AI is going to mark it incorrectly.  And there's no way to ask the AI "why is this post "bad"". 

 

Modern AI is a big black box - garbage in - gospel out.

I'd be interested in reading about this, do you have a source?

 

Thanks,

 

 ..Al

Took ya' long enough.  

 

I tried Facebunk years ago, because a friend of mine said that was the best way to stay in touch with him. The whole experience was so phenomenally annoying (and I wasn't even using it regularly) that I killed my account after just a few weeks.

 

Never missed it.

 

It's a bit annoying that some people/businesses/groups only post updates there, and if you don't have an account you can only read a handful of posts before content gets blocked. But it's not annoying enough to do anything about it.

 

On 8/20/2021 at 1:28 PM, EricBall said:

The reason they won't tell you why is they don't know why themselves.  They have a neural network they've "trained" by feeding it a bunch of posts marked "bad" (and hopefully some marked "good") until it correctly marks another bunch of posts.  The more similar your post is to the "good" training & testing posts, the more likely the AI will mark it correctly.  But if your post is more like the "bad" training & testing posts, then the AI is going to mark it incorrectly.  And there's no way to ask the AI "why is this post "bad"". 

 

Modern AI is a big black box - garbage in - gospel out.

On 8/20/2021 at 1:36 PM, Albert said:

I'd be interested in reading about this, do you have a source?

https://en.wikipedia.org/wiki/Artificial_neural_network#Training

IMHO it's a huge problem with neural networks.  One other item which isn't mentioned is the raw input data is often "pre-processed" to make it more compatible with the neural network.  Of course, this pre-processing might introduce or amplify biases.  

Yeah. Don't use public forums to keep in contact with people. I just use discord to keep in contact with friends. Especially after my old scratch account got Thanos snapped over something similar in concept to this. (Killing all 5 years of hard work and personal progress). Small communities usually have better moderation so it's better to stick around those.

Phone calls or letter writing still works well. Even a bit of risqué texting doesn't hurt.

Interesting - received this yesterday from Facebook, just as the 30 days came to an end.  Since I didn't request a password change, and have no desire to reactivate my account, I clicked the let us know.