Even though we have a powerful bad-word detector on Scratch, there are still some loop holes and mishaps that occur. Sometimes a user may post “this sucks” or “I hate this” on a good project. Some may write bad words in dfoasgnadrilwnav9dsrandomlakd9ownf text. Some may write bad words backwards, which isn't spotted at first.

The problem is that the B-W detector can't detect every harmful word in every way, and this is bad. We're an open target for bullying. Scratch Team, for the safety of our young and the good riddance of threats, we need you to fix the B-W Detector ASAP! Here's what you can do to improve the safety of Scratch:

1. Have the B-W detector read text backwards, forwards, and find bad words in garbled text, if it doesn't already.
2. Increase the pool of bad words to block. See the examples above.
3. Keep a close tab on suspicious usernames. Some may be the new accounts of banned users seeking vengeance.
4. Also detect bad words in usernames as well. Some names can be just as mean as comments.
5. Any other suggestions users may make.

Please do respond, ST.

#1 oh it does do that,that is why “grapes” are censored (from what i heard)

#2 maybe. but if pakistan is on the blacklist, they may need to remove some.

#3 realize that people that have nothing to do with each other might make similar names.

#4 True. I +1 that.

#5 N/A

the “it sucks” reminds me of a company(I think it was swedish vacuum cleaner company) who's slogan was “nothing sucks like *company name*”

Thanks for your suggestions! But please realize that no bad word filter is going to be able to outwit the ingenuity of dedicated people 100% of the time. There will be things that are missed - and that's why there is a Report button on every comment. We have been working quite hard on the filter - but remember that each time we add a new rule to it - it adds to the number of “false positives” that need to be looked at - and adds to the work that our moderators have to do. So we have to balance how much it tries to find - you would be amazed at the number of “bad” words you can find in ordinary comments - that really aren't bad at all.

ProdigyZeta7 wrote:

1. Have the B-W detector read text backwards, forwards, and find bad words in garbled text, if it doesn't already.

Yes - it does some of this now. But we have to be careful not to have it try too many patterns or the false-positive rate skyrockets (it finds things it thinks are bad that really aren't).

ProdigyZeta7 wrote:

2. Increase the pool of bad words to block. See the examples above.

That is easily done - and when we find words that we are missing on a regular basis, we will add them. Just keep reporting them when you see them - our moderators see your reports and will tell us when a word is not being caught and we can add it.

ProdigyZeta7 wrote:

3. Keep a close tab on newly created users. If there are similarities in usernames (i.e.links removed by moderator) that have a background of swearing, bullying, etc. they shouldn't be creating new accounts, unless they have parental permission to do so.

You are exactly right! We have a number of safeguards in place now - and we will be adding more soon.

ProdigyZeta7 wrote:

4. Also detect bad words in usernames as well. Some names can be just as mean as comments.

Indeed they can. As of last week, we have a simple filter on usernames now - we may increase it's scope after we see how it works out.

ProdigyZeta7 wrote:

5. Any other suggestions users may make.

Absolutely! If people have ideas to help with this problem, please let us know. It's a tough problem to manage - we're doing far more than we were on the old site now - but there will always be room for improvement.

Paddle2See wrote:

Thanks for your suggestions! But please realize that no bad word filter is going to be able to outwit the ingenuity of dedicated people 100% of the time. There will be things that are missed - and that's why there is a Report button on every comment. We have been working quite hard on the filter - but remember that each time we add a new rule to it - it adds to the number of “false positives” that need to be looked at - and adds to the work that our moderators have to do. So we have to balance how much it tries to find - you would be amazed at the number of “bad” words you can find in ordinary comments - that really aren't bad at all.

ProdigyZeta7 wrote:

1. Have the B-W detector read text backwards, forwards, and find bad words in garbled text, if it doesn't already.

Yes - it does some of this now. But we have to be careful not to have it try too many patterns or the false-positive rate skyrockets (it finds things it thinks are bad that really aren't).

ProdigyZeta7 wrote:

2. Increase the pool of bad words to block. See the examples above.

That is easily done - and when we find words that we are missing on a regular basis, we will add them. Just keep reporting them when you see them - our moderators see your reports and will tell us when a word is not being caught and we can add it.

ProdigyZeta7 wrote:

3. Keep a close tab on newly created users. If there are similarities in usernames (i.e.links removed by moderator) that have a background of swearing, bullying, etc. they shouldn't be creating new accounts, unless they have parental permission to do so.

You are exactly right! We have a number of safeguards in place now - and we will be adding more soon.

ProdigyZeta7 wrote:

4. Also detect bad words in usernames as well. Some names can be just as mean as comments.

Indeed they can. As of last week, we have a simple filter on usernames now - we may increase it's scope after we see how it works out.

ProdigyZeta7 wrote:

5. Any other suggestions users may make.

Absolutely! If people have ideas to help with this problem, please let us know. It's a tough problem to manage - we're doing far more than we were on the old site now - but there will always be room for improvement.

What you said got me thinking….what if you not make a blacklist, but a whitelist of words people can say, just like what some people make for safe chats? I know that sounds Horrifically tedious, but It could work….

stickdude123 wrote:

What you said got me thinking….what if you not make a blacklist, but a whitelist of words people can say, just like what some people make for safe chats? I know that sounds Horrifically tedious, but It could work….

I -1, but it's up to the ST if they want to add this “feature”.

Paddle2See wrote:

Thanks for your suggestions! But please realize that no bad word filter is going to be able to outwit the ingenuity of dedicated people 100% of the time. There will be things that are missed - and that's why there is a Report button on every comment. We have been working quite hard on the filter - but remember that each time we add a new rule to it - it adds to the number of “false positives” that need to be looked at - and adds to the work that our moderators have to do. So we have to balance how much it tries to find - you would be amazed at the number of “bad” words you can find in ordinary comments - that really aren't bad at all.

ProdigyZeta7 wrote:

1. Have the B-W detector read text backwards, forwards, and find bad words in garbled text, if it doesn't already.

Yes - it does some of this now. But we have to be careful not to have it try too many patterns or the false-positive rate skyrockets (it finds things it thinks are bad that really aren't).

ProdigyZeta7 wrote:

2. Increase the pool of bad words to block. See the examples above.

That is easily done - and when we find words that we are missing on a regular basis, we will add them. Just keep reporting them when you see them - our moderators see your reports and will tell us when a word is not being caught and we can add it.

ProdigyZeta7 wrote:

3. Keep a close tab on newly created users. If there are similarities in usernames (i.e.links removed by moderator) that have a background of swearing, bullying, etc. they shouldn't be creating new accounts, unless they have parental permission to do so.

You are exactly right! We have a number of safeguards in place now - and we will be adding more soon.

ProdigyZeta7 wrote:

4. Also detect bad words in usernames as well. Some names can be just as mean as comments.

Indeed they can. As of last week, we have a simple filter on usernames now - we may increase it's scope after we see how it works out.

ProdigyZeta7 wrote:

5. Any other suggestions users may make.

Absolutely! If people have ideas to help with this problem, please let us know. It's a tough problem to manage - we're doing far more than we were on the old site now - but there will always be room for improvement.

There may be no perfect B-W filter, but we can improve one until we think it's perfect. Just recently (like in the last hour), another user posted “hate it” on a valuable project, and it seems the B-W filter is missing more than we think. I'm also trying to get another user to stop using the “r” word in his projects, and I do believe he hasn't read the Community Guidelines yet. The problem is that new Scratchers are more likely to skip reading the rules of Scratch because they assume that our policy is the same as any other website. Again, we may not be able to stop 100% of bullying, but we can try for 99.9%.

Bump. Someone commented with bad words spelled backwards…

Censoring bad words if they're backwards is kinda ridiculous and unnecessary.

I heard that skyscraper used to set off the filter.

coolitsusa wrote:

Censoring bad words if they're backwards is kinda ridiculous and unnecessary.

I heard that skyscraper used to set off the filter.

Well, about 10 minutes ago, someone wrote “(substitute for lacking intelligence) (synonym for donkey)” backwards on the newest featured project, showing the weakness in the bad-word filter. I'm not sure if the ST will catch it, but I had to report it.

stickdude123 wrote:

Paddle2See wrote:

[snip]

What you said got me thinking….what if you not make a blacklist, but a whitelist of words people can say, just like what some people make for safe chats? I know that sounds Horrifically tedious, but It could work….

No, this won't work. There are about 2,000,000 words in the English language (not counting words made up on the internet like facepalm, etc.). This also doesn't count misspellings or the >30 other languages out there.

coolitsusa wrote:

Censoring bad words if they're backwards is kinda ridiculous and unnecessary.

I heard that skyscraper used to set off the filter.

I used to play on a Minecraft server with an oversensitive filter. It kept censoring random phrases like “skyscraper”, “grapes”, “if you can”, etc…

Making a filter too sensitive is not a good thing. Also, some people found a way around the filter in question.

stickdude123 wrote:

What you said got me thinking….what if you not make a blacklist, but a whitelist of words people can say, just like what some people make for safe chats? I know that sounds Horrifically tedious, but It could work….

I wouldn't really work.

Think of all of the words in every language in the world.

Also, how about misspellings like “hey wuts upp how r u doin? Hav u beeen wurking on ur awsom praject?”

"3. Keep a close tab on newly created users. If there are similarities in usernames (i.e. links removed by moderator j/]this) that have a background of swearing, bullying, etc. they shouldn't be creating new accounts, unless they have parental permission to do so."

However, there might be different people having similar usernames. For example, some of mikeono's fans make accounts like [insert word here]Ono, and my username is similiar to JH1010's.

jh1234l wrote:

"3. Keep a close tab on newly created users. If there are similarities in usernames (i.e. links removed by moderator j/]this) that have a background of swearing, bullying, etc. they shouldn't be creating new accounts, unless they have parental permission to do so."

However, there might be different people having similar usernames. For example, some of mikeono's fans make accounts like [insert word here]Ono, and my username is similiar to JH1010's.

Okay, I'll eliminate that ambiguity. It's just sometimes users who get banned try to go around their block by creating new accounts with similar names.

Following.

Also it detects censoring.

when @greenFlag clicked // Don't smash it, though.
possibly forever {
repeat until <(eureka :: sensing) = [true]>
think up [an idea v] :: sensing
end
make [a project v] with (thoughts :: sensing) :: events
} @loopArrow :: control cap

I agree. Although some people may find loopholes, the majority of Scratch community would follow the rules.

upsilon wrote:

I agree. Although some people may find loopholes, the majority of Scratch community would follow the rules.

Yep, those ugly trolls.

Tell me, does Scratch have a filter for B-words in other languages? Especially Chinese.

And how about one for About me and What I'm working on?