Can we talk about the language filters on GT7?

[benjoi84]

Something's been on my mind for a while now, a minor annoyance but an annoyance nonetheless.

(Preface: I am aware this issue stems from Sony's Terms of Service and not GT7, but my complaint comes from within the game so here we are).

When sharing a replay, I'll sometimes get a "bad language warning" for no apparent reason. Well, there is a reason, it's just that they don't tell you what the offending content is. It wouldn't really bother me too much, but I've now got two warnings from Sony for my 'behaviour' and I feel like I'm totally helpless to prevent it, apart from abstaining from posting content. As you can see my latest infraction, here in this example, there's really very little that could be interpreted as offensive;

So, the upshot of this situation is I had to try deleting each word one by one to work out what was causing the issue (I'm doing this on the Daily Race post race lobby, so time is limited). After many iterations of the original comments, here's how my shared content looks now it's 'legal';

It turns out, the word 'follows' triggered the language filter. Why didn't the system just tell me that was the problem? Moreover, how can that be a problem? And what of the difficulty in identifying it... It can't be that offensive if it took process of elimination to work it out!

Has anyone else had problems with this?

 

[Real_StiFex]

Yep it happenned to me a lot in the context of leaving a comment on a photo scape or a livery style. Either to express admiration or to praise the content creator.
Many times I had the trouble of having to guess the "offensive" word and try to circumvent around meaning having often the obligation of redacting the whole comment another way...
Anyway regarding the matter the thing is, I think, by design, they cannot tell you the offensive culprit word, this in order to prevent an "ill" atempr from people to collect intel on the whole list of banned word and expression.
In order to keep and protect the guidelines efficient the full list cannot be known.


And I agree that the bad effect of this pratice is that the base user without any ill manners is impacted, collateral damage of sort...

As an exemple I recently did myself a serie of 3 black & white scapes and shared them only to discover that my description was flagged because "black & white" was inapropriate. As well as the term "2/3" as in 2 of 3
To designate the pictures order

Well it is a larger reflexion over the subject of "grey zone" censorship.
Trying to elducorate content leads to dumb things in the regular environement when a color is deemed inapropriate.
Maybe IA censorship would help with language ( another subject in regard to that sure)
Anyway it is a cat and mouse game with bad offenders trying to go around and find other fabracted terms to make offense without being noticed....

 

[Famine]

The primary thing to understand is that it's not anything to do with PD. This is all SIE's work; SIE handles all content moderation for its first party titles from studios like PD which are and always have been wholly under their auspices.

Anyway regarding the matter the thing is, I think, by design, they cannot tell you the offensive culprit word, this in order to prevent an "ill" atempr from people to collect intel on the whole list of banned word and expression.

It's actually marginally simpler than that. SIE staff are bound by the same terms of service in communications. They can't say "you used the offensive term 'spoonhead' in your message" because they can't use it either.

Similarly, I can't tell you that 🤬, 🤬, 🤬, and 🤬ing 🤬 🤬er are censored on GTP because... well, they're censored. That's as precise as you 🤬ing get.

They absolutely could do it a lot better than they do though, as it's pretty convoluted. For a while we couldn't use the term "engine", and "LSD" is probably still off the table (the non-drug meaning being too niche for SIE to grasp). Some terms are censored because they're truncated, reversed, anagrams, or non-English versions of offensive words, but that can account for a lot of perfectly valid communications.

There's zero reason why a human moderator - and they are all human moderators at SIE - should look at that post and decide that it's offensive, but I guess they have a script to follow and that says that "t f" in "Swift follows", or "pens", or some other thing, is a strike.

 

[benjoi84]

@Famine prompt reply and good explanation appreciated...

Seems just another way in which is us honest folk are penalised for the antics of others... To think I could potentially be prohibited from using a service I pay for through quite literally no fault of my own.

You can't even learn anything from the experience!

 

[Real_StiFex]

I will not quote you both for the post lenght sake but I concur to all of that too.
I had knowledge about invovlment of SIE - TermsOf Service
I wanted to enlarge the discussion to a broader sense in order to point the limits between aimed effects and consequence of censorship.
Which is in my opinion a much due work and more necessary in order to fight hate speech and bad behaviour all the wise.
We still trying to figure out the balance.

What will happen when a filtering list will include every terms expressions aliasses and convulated figure speech in all language ( World wide dialects) will result in this kind of situation.
Add on top of that exploits and stratagems from bad actors to avoid detection.
It is known ( I wish I had sources at hand) to end up in a worse situation in the long term.
That is why I think we have to find new method to fight harassement and fight hate speech globally.
But that is another kind of discussion way off my reach...
Have a good week end folks 👍

 

[kjeldsen]

delete

 

[Real_StiFex]

That is an opinion you have.
Mine is that tackling those question globally Would have much more efficience and possibly better results in order to fight for and regain serenity on the online space.
It is a vast and spicy subject and I am not saying that we need to do something right here right know on a gaming platform be it X or Y
In this form I agree with you that it is de facto the best solution currently available.
But that in the long run only counting on those kind of tools will deserve the "cause" or the aimed intent by impacting the " goods" more than the "bad" to over simplify what I mean.
Sorry to digress but I want to be clearly undestood in what I express.
Please feel free to reach me in PM or otherwise I will stop here because I don't want to slip in a really unwanted trap... 😅

 

[kjeldsen]

delete

 

[Draggon]

Today I discovered that SIE thinks Italian is a bad word. Good grief...

 

[hawkeyez]

One time it was just me and my friends in my lobby, I typed “man that car is a turd” and it was deemed offensive. So I tried “this car sucks” nope didn’t go. I think I tried typing something three times and it didn’t let it through. Yet I’ve had randoms join the lobby with offensive usernames (I don’t care at all) but I can’t write turd? Lol

Sticks and stones may break my bones but words can never hurt me. I know words CAN be hurtful to some but I feel like that’s up to the person receiving those hurtful words to just it brush it off and move on.

Let people report bullying and such if they feel the need. The automatic censorship is frustrating sometimes, as you say it’s a word that is used on everyday sentences with no real power but suddenly it’s somehow offensive?

 

[kjeldsen]

delete

 

[Korvregn]

I actually thought that it was flagged because they thought you were roasting people from Finland.
Like: ”Finally I found ONE decent finnish person, after only meeting fins wanting to cut me with their knife in the sauna”

Sorry Finland, you are all decent and you know I love you guys!

 

[shadow0460]

Too many people get offended by certain terms, moreso random 💩 than use of the 🤬 bomb.
🤬 that 💩.
Offended? Well I happen to like the poo emoji. 💩 What better thing to stare at you from word salad than a smiling pile of piddle? 💩

 

[benjoi84]

Too many people get offended by certain terms, moreso random 💩 than use of the 🤬 bomb.
🤬 that 💩.
Offended? Well I happen to like the poo emoji. 💩 What better thing to stare at you from word salad than a smiling pile of piddle? 💩

Indeed. I have a saying that goes "If someone is determined enough to be offended, they will find a way"

 

[Cold Warrior]

Okay I get they also can't use the term themselves. So why can't they then just highlight for you what they don't like? Instead of it being a guessing game. After some trial and error I just learned the word "participated" is considered offensive in GT7. Trying to post a replay for my league. Don't understand why. I thought we are supposed to be inclusive.

 

[Mafio]

I think it’s pretty evident its not a PD thing by them adding a car whose name is on the banned word list.

Ford ******.

 

[Cold Warrior]

I think it’s pretty evident its not a PD thing by them adding a car whose name is on the banned word list.

Ford ******.

Interesting. Never heard of a Ford ******.

 

[B80]

Interesting. Never heard of a Ford ******.

Ford Orion (with an extra letter...)

 

[Meythia]

So why can't they then just highlight for you what they don't like?

So anyone could simply use a different swear term?

The filter should only ever be a problem in cases like the current one, when a name equals a banned word.