i just read makerworld’s new guidelines and i have to say, it barely will make a change. part of the guidelines says that inappropriate models MAY get removed. whats this MAY? relying on users to report them is a terrible idea for the following reasons:
1: the whole pint of removing those types of models is so users dont see them and in order for that to happen, users need to see the model which defeats the purpose.
2:not everyone has the same defintion for a model needing removal.
3:people will still put out these models since the risk of getting reported arent so high.
what i suggest is the following:
1: makerworld should have a few humans take 2 seconds per model to see if its fit for publication
2: fine people who make obviously unfit models to discourage people from doing so.
3: take out the word “may”. it makes makerworld look weak, and unable to enforce its own rules.
I don’t think you quite understand the scale of MakerWorld; the number of new models uploaded daily and the ACTUAL time it takes to properly review a model makes this a far more challenging task than you might realize.
MakerWorld does not have the legal authority to impose fines, and attempting to change the Terms of Service to make such a thing possible would result in a full-on revolt among all users.
The thing that makes an authority look weak is tough talk and failing to follow through. “May” creates uncertainty, and uncertainty leads rule-abiding people to typically engage in self-censoring behavior, which reduces the moderating workload even further. May creates deterrence.
you are correct that there are a lot of models uploaded daily but it takes 2 seconds a model to see if there is anything wrong with it and they have enough employees to make it work.
what i meant by fines is to fine them in makerworld points like they do sometimes.
i’m not looking for tough talk. a little more confidence would still be better than uncertainty. and besides, they should be following through regardless.
Wish in one hand. Many of us have been saying this and more but Bambu doesn’t seem terribly interested in our opinions.
The “may” part comes from letting users police which saves an employee or several since it would need to be manned 24/7 unless you want delays added to posting.
It’s absolutely not ideal. The rules don’t get enforced evenly. Some users even get bans implemented where others keep on trucking. I doubt you’ll find many who like the way the present system is implemented.
But it’s their platform and their rules. They want to avoid legal issues - understandably. Every one of us can vote with our feet whenever we want to but doubt we’d be missed. Refresh the new models page a couple of times and see how many models get posted every minute. It’s surprising.
A couple of people can man a support system that is chat based, having enough people to go and check every single upload is going to cost them a lot and have them not doing actually useful stuff for the company that helps generate revenue
There really isn’t a benefit to them wasting that manpower
I guarantee you that it takes longer than 2 seconds on average over a day.
A human might be able to do 5 approvals in 10 seconds, but any refusals take extra time to provide a reason, errors take time to fix, and humans tend to go crazy if you ask them to do 12,600 models a day. Ask the AI workers (where AI stands for “Actually India”).
Another example of how the turing test is discredited. Some support chatbot just got ID’ed as a human. (Not saying there are no humans, but I am saying the humans have a lot of AI assistance)
you’re right that it will cost them a few extra workers but if that will build their trust its worth it for them. they are making billions anyway. they may as well build up trust and get the extra 10 workers
Except there is no lack of trust in that department, so its cost for no real gain, especially as once they start actively checking every upload they themselves become liable for anything that slips though, unlike now where they are not held responsible for what users posts because they operate on a reactive basis
there is definitely trust to be gained especially because this is what makerworld is striving for: a safe site full of models appropriate for everyone.
Just to give you an idea: Hundreds of models are uploaded every hour. Source: MakerWorld. You can check it out for yourself. As I write this, the rate was 250 per hour - and we’re talking about 24/7.
That amounts to 6,000 models per day. Let it be a little less one day and a little more the next, just to get a sense of the scale.
4,2 models per minute. Just think about how many employees would have to be hired to work in shifts. Check the cover and photos, read the text, and view the actual 3D model. In addition, various languages. Sure, that would be desirable, but it’s impossible to pull off without spending a fortune on labor costs.
Sure, if Bambu doubled its printer prices and MakerWorld were a pay subscription service, then they could definitely do that. But who’s going to pay for that? Exactly… certainly not the customer.
Anyone who asks for something like that should also be willing to sign up for a MakerWorld pay subscription - if that were possible.
What makes you think that’s even possible in such a short time?
Why don’t you try it for yourself? Try doing that for an hour and see how it goes for you, then we’ll talk some more.
Decisions should be well-founded and in compliance with the rules, not random. To ensure this, the 3D model must be viewed from all angles, along with the photos and text. Especially since there are some very complex models - and, unfortunately, some that are cluttered with unnecessary elements.
Imagine the outcry if hundreds of models were blocked because the wrong decisions were made. We’re still talking about people here, not robots.
I don’t mean to pick on you, but either you have no idea what you’re talking about, or you just want them to take a quick look without really checking whether the models comply with the rules.
…and you ignored the question of who’s going to pay for it.
Be honest: Would you pay around $20 a month for a MakerWorld subscription?
this isnt a full scale examination. this is a quick scan to make sure the model isnt inappropriate or racist or doesnt fit with guidelines. no one needs to spend their time zooming in on every last triangle. thats absurd!
if i was being paid it would be a lot easier.
i completely agree that no system is perfect but i would tell workers that they should spend an extra minute or two if they arent sure and if they still cant make up their mind to put it up anyway and hope for the best. this isnt supposed to have a 100% success rate. 99.9% is fine too.
i dont think you understand the numbers. are you trying to tell me they would need $20 million a month for this? makerworld has over 1 million users! 5 workers would be enough and in china work is probably cheaper than the US’s $18 an hour
The numbers that you are using are way underexaggerated.
For some models, it might take 10 seconds to open it up and look for obviously fake images. Others, when people are using better AI models or putting in more effort, can easily take 5 to 10 minutes. Then there is the question about interpretation, when you rush someone there is going to be a lot of false positives / negatives. Someone could post a “88” model, but if the Chinese worker does not get that reference immediately off the top of their head, good, now there is ■■■■■■■ crap on makerworld.
I also find that you exaggerate the efficiency of the average worker. In a 10 hour work day, I probably work 6 hours or so… and that is well above the average of 3 hours for a 8 hour work day.
Just go do it.
Sit down for an hour, look through the new models, report the ones that you think break the rules.
Maybe you will understand the process more. I know me and Sharky have done this out of being annoyed.
Then keep track of how many models you can go through.
Actually closer to 15 million users from what I read.
Still think how many people would go elsewhere if they had to pay a subscription when there are still plenty other platforms that dont have one.