Bitcoin Forum

While we are waiting for a reaction from theymos / see his comment below/ I suggest to add this to the rules.

theymos' comment on the subject, I have provided already such table but still no reaction.

The OP was edited together with the subject. The original OP is below this line.

---

Here is what I found.
What they do is replacing some letters with the corresponding Cyrillic as they are lookalike, know also as homographs.
This kind of activity is a bit weird for me, I couldn't find any similar comment with the same text, tho.

I checked the letter "a" so all marked with yellow are in Cyrillic.
https://i.imgur.com/eFs6pT4.png


Here are more examples with the same approach. So shall I report such activity?

https://i.imgur.com/iKPypX8.png




I thought that they cover the copy/paste postings but you have to research every single case.

There are many more like those, this is just a brief example.

I think this kind of unnecessary activity should be considered as a rule-braking and be punishable.
What you think?

There was a discussion about this exact issue a couple of months ago, see here: https://bitcointalk.org/index.php?topic=3094459.0 (https://bitcointalk.org/index.php?topic=3094459.0)

Bottom line is it is still plagiarism. Report them all, and include a link to the original post in your report.

I've seen this method before, they copied from a post in the same thread :P not so smart
I'll give you a hand 8) here's all related post sources that they copied from
btw, I didn't file report on them yet because I don't want to take the credits
These are your findings, you found them first ;) so you have the privilege to report them

---

Original
Copy:

---

Original
Copy:

---

Original
Copy:

---

Original
Copy:

Thanks for the help, both of you, it's OK to report them, no worries. I have already reported many (over 900 now) so I don't really care if someone will take the credit for it. It's just to have some action against such activity.
I'll just continue the same way :)

https://media.giphy.com/media/Nm8ZPAGOwZUQM/giphy.gif
^{Click here if unable to see image} (https://media.giphy.com/media/Nm8ZPAGOwZUQM/giphy.gif)


It's just unbelievable for me. They took it into the next level. Really cleaver thing...  :o
I was not familiar about this kind of tricks, it took me a while to figure out what was going on.  

I wish they use these skills for something good.  ;) Really waste of energy when using the skill for spamming.

Here is the result with one simple word "the", "e" is Cyrillic

All "the" are marked with yellow. The once that have the with cyrillic are with bold text  

https://imgur.com/ItwkJ8R

I have 5 pages to report now :D

What if someone copied your post ( especially when he/she is a newbie ) ?
And maybe his only goal is to destroy you  ( through " a permanent ban" ).

I am not sure I am reading you correctly, but there is no penalty for having your posts plagiarized by someone else.

In your example, the person who plagiarized (ie. the copier) would get banned. There would be no action taken against the original poster.

You can check the dates on the posts, so don't worry, no one can cheat here. Well, if you are sharing same thoughts and expressing them in the same way just as a coincidence, it will be easily marked, checking the post history of both persons.

This one you reported is copy and pasting as well:



The accounts don't seem to be related from what I can see but they're all trying to mask their plagiarism in the same way. I wonder if there's a guide out there that tells them to do this to avoid detection? I usually try find a unique word or typo as they're easier to spot like 'altcoyins' was here.

I accidentally found the first one by searching for Cyrillic letter "a" in the search engine and I was really impressed that we can search by only one character, awesome feature. Normally I check for non English comments outside the Local boards by searching specific characters. Then I saw many results in English with the Cyrillic letter "a" highlighted and just started reporting. Now every few days I'm reporting some posts like this.

Also I have a question to you as a moderator:

I'm not sure what you mean. If you report the same post twice there will be two reports in the cue, but not sure what you mean by 'two people report'. If you mean you report one post but with additional info on another user that needs to be looked into then it will still be one report.

Thanks for the answer. I guess it was a bad formulated question.
The whole idea was if a bad post is reported 5 times from 5 different users does it go higher in the queue (because of the more reports) or there are just 5 different report in the queue.

Oh. No. There will just be five different reports in the cue in the order that they were made. The report cue isn't very sophisticated but I think I read that on the new forum software it will have them all grouped together and will be more efficient. 

I still think that using of homographs have to be forbidden and added to the rules.
By doing so, we will not waste time in searching for plagiarism. On the other hand using homographs has no positive advantages in the forum, the only reason someone will use it is to cover something, it cannot be done by mistake, it takes quite an effort and it is done on purpose.
So the best and easy thing is to be forbidden.

• If you have more than 2 changed characters with homographs in your post, it should be deleted.
• If you have more than 2 posts breaking this rule you should be banned for a week.
• For more than 2 post the user should be permaban/nuke.

While we wait for a solution from the admins, in the mean time we can just clear up some of the examples below from the last few days.































...
and many more ...

I support this request. I can see no valid use for their concealed use in a standard English text stream. I appreciate all the hard work you guys put into cleaning up the forum, and improving the quality of the threads.

Damn these shitposters would never give up. How in the world do they get such ideas of spamming? This is the reason why spam is not being controlled and the forum is still a trashcan full of blabbered out English from a shitposter.

Mods may ban them for plagiarizing but something needs to be done - like IP ban and Hardware MAC address tag so that they can be forever banned from the forum. :P

But I guess we can all sleep.

Well of course they're never going to give up as long as they can make more money spamming bitcointalk than they could doing manual labor in [insert Asian country here].  This is pretty clever of them, but my question is:  Can google not catch these substitutions?  Seems like even if you include a phrase in quotes, they return a lot more results than are warranted anyway, and I figured their engine would be a lot more powerful than some shitposter's homograph-substitution scheme.

I concur with OP & Jet Cash.  If this is what spammers are doing these days to fly under the radar, it should probably be banned.

I have around 50 hompgraph reports waiting to be investigated + the ones I've mentioned above, which I found only on the first page of the search result for the last 10 days.
I'm on vacation at the moment and have no pc around, quoting on the phone is pain in the arse.
So I guess there are much more of those shitposters.
Let's add the to the rules so they can get what they deserve.

I just added poll, so hopefully we'll see the big picture in a few days, of course if most of the people are active and willing to vote :)

Useful link to look at when checking for these: http://unicode.org/cldr/utility/confusables.jsp

I'll keep bumping this thread until there is some reaction on the case.

What can be done >

• Theymos adds feature that automatically converts the homographs to Latin outside the Local section
• We all together stop this madness as we list using homographs on the rules

The spammers already made an improvement. Instead of changing all the letters with homographs, now they change only one letter, which is more difficult to detect / at least they think so/.

Here is an example form the last few days. The marked with yellow letter is Cyrillic "a".
... image loading (https://i.imgur.com/2V7326p.png)
https://i.imgur.com/2V7326p.png

---

Thanks man for helping, this is a good tool to see what type of characters to look for.

Here's an even better tool for checking for homographs: https://www.textmagic.com/free-tools/unicode-detector (https://www.textmagic.com/free-tools/unicode-detector)

Just copy and paste the text in, and any Unicode character will be highlighted in red. If you suspect someone of using homograph plagiarism, you can go their profile and copy in an entire page of recent posts to check them all in about 10 seconds.

This is a great tool, thanks.

There is no problem in detecting them but reporting them. Let me explain.
Using only homographs is no rule-breaking, the problem is that those who use this technique are trying to hide copy-pasting.
But to accuse someone in copy-pasting first you have to correct the post back to normal Latin characters, and then search for the original posts. Which takes time, even if you are using Word with "replace all" option.

I just want to add the homographs to the rules, because using them is no beneficial for the forum at all.
Doing so, you can directly report the homographs and skip the plagiarism part.

---

Bump, last 3 days history with more than 70 cases of using homographs:
https://i.imgur.com/F9np4wB.jpg

Are there any legitimate reasons for needing to use monographs in a forum post? I can think of none. Can anyone correct me?

If there are none, then the only reason to use them is to hide plagiarism, in which case they should be banned.

No reasons whatsoever..

OK then, a question:
What it takes to add something to the rules, when it comes to something abuseful like homogprahs?

---

Another bump. I'll do this until there is a reaction on the case.
Already 22 cases from today only (https://imgur.com/a/yi1fGZd) /one is a copy/paste report actually/

---

And a new bump and anoter 24 cases only for today...
https://i.imgur.com/aV8afqd.png

I am not sure if there might ever be a legitimate need for the use of these symbols. If not, we may want to change SMF settings so that users cannot use these symbols, or that these symbols are automatically changed to the letter they are designed to look like.

We are waiting for reaction from theymos and I know it can take a few more months, in the mean time I just want to report all those using homographs but I have to find another reason to report them, coz using homographs is not against the rules... yet.

You could argue it's not English, which isn't allowed on the English boards, but it's a bit far fetched.

Since there is no legitimate use for it, they should just be banned. It's clearly abuse.

I gave up looking for them though, because I can't see which accounts are banned already.

I have around 50 hompgraph reports hanging as unhandled for more than a month now and I stopped reporting them. Not enough time to waste on listing them on my rule-breakers list, so I'll push this thread a bit until we have a clear solution to the hompgraph problem.
Actually I don't see so many now, just 30-40 per day max, before they were like a few pages in the search results.

---

Bump, another 20-ish for the past day. I'll start reporting them as soon as they are forbidden.

---

Bump.I have reported already a few from today, lets see what the mods will do.

This topic requires a higher priority. While it used to be a relatively small problem, I'm afraid I've pushed massive spammers to using homograph attacks by getting many of their spambots banned.
See here (https://bitcointalk.org/index.php?topic=4642588.0) for many examples of spambots who started using homograph attacks today, which they didn't do yesterday.

I do check for homographs almost ever day, or in the worst cases evety other day. I can tell you that almost all of the "single character" hompgraphs popping up in the search results ( I just check the past day/two days posts ) are made from newbies for the first time, refering to the "a very" case I have asked you to check a few days ago.

Let's be honest, all those are bots, it is known.
From time to time I spot a regular hompgraphs with many vocals replaced, but those are very rare now, and mostly posted by the "usual suspects".
I stopped reporting them as I got a bad report on one case - this one below, and I still have 49 hanging.


So, I know this is getting bigger, it was big before too, but as I started reporting the thing started to look better.
Seems like the time spend reporting is a bit waisted, as the banned accounts are easily replaced by new ones, and everything done with a script.

---

bump

Just a few from the latest ones,
Please, tell me again that this homographs are not used to hide plagiarism/copy-pasting?

Big drawback is that LoyceV's script cannot search for homographs. I have to hunt them manually.

C'mon let's add them to the rules people!!

I can search for them, but only for Newbies and it's more work.
If you want me to search for them, can you make a list of common words and show the both the  homograph version and the HTML equivalent?
Example:

wallet

If one word has different variations, post them all. I'll continue downloading new patrol pages next week, I'm not online much at the moment. But feel free to use this time to collect them for me.

Finally the homographs problem is solved and I'm locking this thread.
I can reopen it if the situation get worse.
Thank everyone for the support. :)

---

 150 new hompgraphs are posted every day, im monitoring this for the past 6 days.
I think it's time to reopen this thread again.

I want to add them to the rules so we can report them directly without looking for plagiarism.

I guess if homograph character set(cyrillic characters) support is disabled in this forum where primary language of posting is English then it will be very easy to detect the homogarphs, They  will convert to gibberish and everybody will understood that cyrillic  character are used to hide copy-pasting. I guess it will reduce the work of everybody.

I believe that the use of homographs definitely suggests the intention to hide plagiarism and so a form of ASCII binding could be set up to deny the use of homographs. This would prevent plagiarized character strings appearing identical to the original if a web scraper or bot was to view the strings, but the scraper should still be able to tell that 99% of the text is copied.

I'm not sure if Bitcointalk has a bot that checks for plagiarism but I assume it does, (or else the use of homographs here wouldn't make a difference) and so another alternative would be to check any text that uses 'mimic characters' with more scrutiny. For eg, a string without the use of homographs could clear a plagiarism check if 70% of the text is original, but a text with homographs might only pass if 90% is original, or not pass at all because the only logical reason for the use f homographs is to evade recognition.

A third option would be to just report simular looking posts to moderators, but even here, the use of homographs to conceal plagiarism is negligible.

So the use of homographs logically means that it is the posters intention of hiding plagiarism from bots so the best course of action would probably be to implement some function into the web crawler that check Bitcointalk and give it permission to delete and ban all posts and accounts that use homographs. The Armenian characters set has characters identical to the Latin character set, eg. o, n, u, S and Լ. There isn't a Latin board here, so removing Latin characters from this forum could work.

Instead of trying to make new rules to stop everything which will likely not be enforced any more heavily than it already is, use this discovery to build tools to find these people faster and report them.

No, the characters are converted automatically to Latin, but searching for them get them listed, see here for example :
https://i.imgur.com/j0CPpxA.mp4

I want to add them to the rules so they can be easily reported.


If I get them banned for using hompgraphs then I can easily reported them, even list them automatically.