[Newbie scrutiny instead of jail] Every new user's first post: loyce.club/patrol

[LoyceV]

I'll keep it short
The recent spam wave made me realize I can quite easily track the first post of every user. See loyce.club/patrol/

Please Report (or Merit) the posts when needed

It's updated every 5 20 minutes. It currently treats every user as a new user (but only once), so the next post from any user will show up. After a few days it should show only first posts from (mostly) new users.

Sample:

[RapTarX]

Just checked out and it's good one since it includes report option & merit. It can be useful for merit source as well for finding latest post and merit them.
But it would be pretty good if it created new tab when we click on report or merit option. Otherwise, whenever I'm clicking on report, I'm going to the report page and then to the board the post was. But if your site would create a new tab when click on a link, anyone can easily report.

[LoyceMobile]

But it would be pretty good if it created new tab when we click on report or merit option.

I don't like that. Now you can choose by using right or middle mouse button.

[Timelord2067]

I saw you posted a link to another patrol page:

Code:

https://bitcointalk.org/index.php?action=recent;patrol

Can you just outline how your patrol page differs from the other?  I'm still looking over that other one.

could the patrol page be narrowed down to a specific section (e.g. meta or reputation (not very good examples, but still...) ? ) ??

[Upgrade00]

I saw you posted a link to another patrol page:

That's a link to the forum's default patrol page, it tracks posts of newbie ranked users, this would include all their posts and replies till they get the required activity and merit for the next rank. Go to forum profile information, you'll find the option to activate patrol link.

The loyce patrol page tracks first post of users only (subsequent posts are not tracked). It primarily targets new members.
From the topic title it follows the suggestions on some other thread about newbie scrutiny as an alternative to newbie jail.
The tool would be used to analyze first posts and to determine which new accounts are possible bots and spammers.

[tranthidung]

could the patrol page be narrowed down to a specific section (e.g. meta or reputation (not very good examples, but still...) ? ) ??

You have customization for patrol page. Please see Hidden pages/ features on the forum

|

Patrol

|

|

|

Recent posts

|

Last 100 recent posts (from all accounts)

|

|

Patrol

|

Last 200 recent posts (from Newbie accounts)

|

|

No-bounty patrol

|

Last 200 recent posts (from Newbie accounts (outside bounty threads)

|

|

Customized recent posts

|

Last 100 recent posts (from all accounts) in a specific board (ex: 1 ~ Bitcoin discussion)

|

|

No-bounty customized patrol

|

Last 200 recent posts (from Newbie accounts (outside bounty threads) (ex: 1 ~ Bitcoin discussion)

|

[NotATether]

I think that for the link spam, you can filter all new posts that have at least 3 URLs in them and put them on a different page. It will help moderators find the spam sooner than just a bunch of reports.

[PrimeNumber7]

There are many new users whose posts are removed due to spam (among other reasons). It would probably not be uncommon for one spambot to create thousands of accounts that eventually get banned in a short period of time.

Unless your page will update when a user is nuked or a post is removed, it will result in a lot of duplication of work for anyone using your site to find spammers. Someone could find hundreds of spammers on your page that have already been nuked.

[nullius]

Objections:

• It won’t work.  This is trivial for spambot authors to work around.  All that a bot needs to do is to make one post which will have a high probability of passing human review from someone who is not following the thread in which the post is made.  The old spammer trick of copying and pasting an earlier post in a long thread can be fully automated—as can a text spinner, to prevent catching that with a bot, too.
• It will result in overreporting of legitimate humans.  Most people are injudicious.  Most people will not use this only to report obvious spambot activity.  Many people will apply their own biases when reporting; and moderators are not always perfect in filtering that out.
  
  This always happens; but I think that it will be worse when those biases are focused on every new user’s first post, especially since people using this will also suffer confirmation bias when they are looking for something bad.  Use of this tool will bias the whole process to err toward the deletion of new users’ first posts.

Unfun fact that I have not hereto mentioned:  My own first post was trashed.  I am still not sure why.  There is a reason why I have oftentimes referred to “the first post in my post history”.  My actual first post on this forum was deleted almost immediately.

Given that experience, I know how discouraging it can be to a human with a Brand New account.  I don’t expect for it to discourage the spambot authors.

[LoyceV]

The loyce patrol page tracks first post of users only (subsequent posts are not tracked). It primarily targets new members.

Correct, although it includes all old members too. The active users will all disappear from the list soon enough, but old users who wake up will still show up.

I think that for the link spam, you can filter all new posts that have at least 3 URLs in them and put them on a different page.

I could, but for now I want to see how this turns out after most existing users disappear from the list. It was a quick fix the way it is, and I don't want this to turn into another one of my time-consuming projects.

Unless your page will update when a user is nuked or a post is removed, it will result in a lot of duplication of work for anyone using your site to find spammers.

At the moment, I don't check for banned users. I update my list of banned users once an hour, but it has an exception for users who recently posted. Modlog doesn't show time stamps, and I made this exception to remove unbanned users from my list again.
I don't check for deleted posts. For now I just want to see if this turns out useful.

Someone could find hundreds of spammers on your page that have already been nuked.

In that case stick to the top of the list only.

[Pmalek]

I would have to agree with nullius regarding point #1.
Eventually, these spammers will find out about this feature and understand that all they need to do to avoid it is to create one post that won't be considered spam, and after that they can continue with their routine. Instead of one, would you consider tracking the first 3 or 5 posts (and listing them one below the other) of each new user? Chances that they will go through the trouble of creating 3-4 legit posts before they start spamming are significantly lower.

But it would be pretty good if it created new tab when we click on report or merit option. Otherwise, whenever I'm clicking on report, I'm going to the report page and then to the board the post was. But if your site would create a new tab when click on a link, anyone can easily report.

Don't click on the merit/report button directly on the page, instead right click on it and open it in a new tab. That will take you straight to the merit or report window.

[LoyceV]

Eventually, these spammers will find out about this feature and understand that all they need to do to avoid it is to create one post that won't be considered spam, and after that they can continue with their routine. Instead of one, would you consider tracking the first 3 or 5 posts (and listing them one below the other) of each new user?

I can add that when needed, but for now I prefer to keep the list short. It's always a cat and mouse game against spam, and I have no illusions this will stop them all. I did however already use it to report several posts that I wouldn't have reported without this.

[mole0815]

Wow this is very helpful.
I guess you never run out of good ideas either
Will visit the site more often and thanks to the shortcuts directly even more practical.
And if then really only the first posts are taken, the tool is a real asset ... thanks!

[PrimeNumber7]

IMO this is something that needs to be done via automation. Most spammers are using automation to make spam posts and the mods nuke *many* newbies. If you use automation, you can report posts automatically, and in the report note specifically state that the account should be reviewed to get nuked and that you are using a script to make the report.

Wow this is very helpful.

As a FYI, there is always the patrol.

[mole0815]

Wow this is very helpful.

As a FYI, there is always the patrol.

Yes, of course I know that. Thanks!
But the new/additional page offers a few nice features that makes it quite easier to work through

[LoyceV]

And if then really only the first posts are taken, the tool is a real asset ... thanks!

It's not really their first post, but the first post they make since (a few hours before) I created this topic.
It's currently down to showing about 50 posts per hour, I expect it to get a lot lower the coming days/weeks.

IMO this is something that needs to be done via automation. Most spammers are using automation to make spam posts and the mods nuke *many* newbies. If you use automation, you can report posts automatically, and in the report note specifically state that the account should be reviewed to get nuked and that you are using a script to make the report.

I can only dream of that I still don't have a reliable posting bot, so I can't automate reporting either. But if it's a bot anyway, MindlessElectron can take care of that without reporting.
I agree it would be better to remove posts that have been deleted, but that requires a lot of scraping to check, while I didn't need any additional scraping to create the current version.

there is always the patrol.

I barely use that because it's filled with bounty spam reports. That's allowed, but makes it harder to find spam that can be reported.

I count a total of 3699 users posting until now

[wooI_Ioow]

 Post 55819586   caught by tool is de facto  from Legendary. A bug?

[LoyceV]

Post 55819586   caught by tool is de facto  from Legendary. A bug?

No, it's the first post he made since I created this project. I show those.

[FatFork]

LoyceV, I like this idea very much. Can we have a timestamp (posting time) for each post as well? It would be helpful to see the rate of new spam being generated.

[Upgrade00]

Suggestion: Make links within posts un-clickable.

Since this project was inspired by the wave of spam links being posted on the forum, it is likely that the first post of some of the new members would contain phishing or malware links, and scrolling through, especially on mobile a user can mistakenly click on one of such links.
I do not know if it's possible but making links within tracked posts not clickable could help.
If someone is interested in the links posted they could follow the post number hyperlink back to the forum and check it here. If it is a spam post which would likely already be deleted, they can always check it on archive (although, there's little reason to want to do so).

[LoyceV]

LoyceV, I like this idea very much. Can we have a timestamp (posting time) for each post as well?

I don't keep the exact time of each post, only the time I scrape it. This is usually a few seconds later. I've added it (Amsterdam time).

Suggestion: Make links within posts un-clickable.

Done!
After some searching, I added the following code to loyce.club/patrol/theymos.css:

Code:

/* The following code came from https://css-tricks.com/how-to-disable-links/ (so not from theymos) to disable accidentally clicking on potential phisting links */
    .ul {
	  pointer-events: none;
    }

I like this solution, the forum consistently uses <a class="ul" href=", so I can quite easily disable all links without editing the original HTML-code for each post.

If someone is interested in the links posted they could follow the post number hyperlink back to the forum and check it here. If it is a spam post which would likely already be deleted, they can always check it on archive (although, there's little reason to want to do so).

Or just manually copy/paste the link into a browser

Update: quotes and image links are still clickable.

[Upgrade00]

Or just manually copy/paste the link into a browser

Thanks for the quick response. Of course manually copying and pasting would be the easiest response, but I was considering a situation where hyperlink was used and it appeared as the personalized text not the original link, wasn't sure how exactly it would appear.

Update: quotes and image links are still clickable.

Well newbies can't post images, so na harm there I guess.
By quotes, do you man quoted replies of others?

[LoyceV]

I was considering a situation where hyperlink was used and it appeared as the personalized text not the original link, wasn't sure how exactly it would appear.

Good point, those require looking at the page source, or my archive.

Well newbies can't post images, so na harm there I guess.

It could still be a link to malware (say a directory called image.jpg so the browser opens image.jpg/index.html) but I haven't seen them yet. If someone clicks an image link and gets malware, I expect them to be smart enough not to download it.

By quotes, do you man quoted replies of others?

Yes. Apparently those use a different tag. That's fine, it links to the forum anyway.

[hosseinimr93]

Suggestion: Make links within posts un-clickable.

Done!

Now, it's impossible to find posts like the one mentioned by TheBeardedBaby in this topic.
For finding such hidden links, I usually hover the mouse cursor over the links and check them at bottom of my broswer without opening them.
Despite of this, I agree with disabling links.

My suggestion is to replace the links with tooltips, so that the real link is shown in a text box when you hover the mouse cursor over the disabled link. Off course, I don't know whether it's possible or not.

[LoyceV]

For finding such hidden links, I usually hover the mouse cursor over the links and check them at bottom of my broswer without opening them.

I thought of that after my change too.

My suggestion is to replace the links with tooltips, so that the real link is shown in a text box when you hover the mouse cursor over the disabled link.

If anyone has the CSS for this, that would be really cool

[FatFork]

For finding such hidden links, I usually hover the mouse cursor over the links and check them at bottom of my broswer without opening them.

I thought of that after my change too.

My suggestion is to replace the links with tooltips, so that the real link is shown in a text box when you hover the mouse cursor over the disabled link.

If anyone has the CSS for this, that would be really cool

I don't know about tooltips (at least not without javascript) but you can add this to CSS to display the contents of the href tag. You'll probably want to add some CSS selector in front as this applies to the entire document. Replace the square brackets with something else if you don't like them.

Code:

a::after {
     content: " [" attr(href) "]";
}

[LoyceV]

I don't know about tooltips (at least not without javascript) but you can add this to CSS to display the contents of the href tag. You'll probably want to add some CSS selector in front as this applies to the entire document.

Sample: https://loyce.club/patrol/test.html
This looks terrible, and I don't know enough about CSS to fix this.

[FatFork]

I don't know about tooltips (at least not without javascript) but you can add this to CSS to display the contents of the href tag. You'll probably want to add some CSS selector in front as this applies to the entire document.

Sample: https://loyce.club/patrol/test.html
This looks terrible, and I don't know enough about CSS to fix this.

Just give me a second...

[TryNinja]

Sample: https://loyce.club/patrol/test.html
This looks terrible, and I don't know enough about CSS to fix this.

It does indeed.

At least do something like this:

Code:

.post a::after {
    content: " [" attr(href) "]";
    color: #666;
}

Makes it readable and doesn't break the rest of the page.

[FatFork]

Dammit! I can display href content in the tooltip but I can't find a way to display the tooltip with pointer-events disabled. Anyway, here's CSS so maybe someone has an idea (TryNinja?). I've added CSS selectors to apply only to posts.

Code:

.post a.ul {
  position: relative; 
  z-index: 2;
/*  pointer-events: none; */
}
/* hide the tooltip by default */
.post a.ul:before,
.post a.ul:after {
  visibility: hidden;
  opacity: 0;
  pointer-events: none;
}

/* place tooltip above the element */
.post a.ul:before {
  position: absolute;
  bottom: 150%;
  left: 50%;
  margin-bottom: 5px;
  margin-left: -80px;
  padding: 7px;
  width: 200px;
  -webkit-border-radius: 3px;
  -moz-border-radius: 3px;
  border-radius: 3px;
  background-color: #000;
  background-color: hsla(0, 0%, 20%, 0.9);
  color: #fff;
  content: attr(href);
  text-align: center;
  font-size: 12px;
  line-height: 1.2;
}

/* make tooltip look like a speech bubble */
.post a.ul:after {
  position: absolute;
  bottom: 150%;
  left: 50%;
  margin-left: -5px;
  width: 0;
  border-top: 5px solid #000;
  border-top: 5px solid hsla(0, 0%, 20%, 0.9);
  border-right: 5px solid transparent;
  border-left: 5px solid transparent;
  content: " ";
  font-size: 0;
  line-height: 0;
}

/* show tooltip on hover */
.post a.ul:hover:before,
.post a.ul:hover:after {
  visibility: visible;
  opacity: 1;
}

@TryNinja you are missing a class selector in your code. It should be

Code:

.post a::after {

[LoyceV]

At least do something like this:

It's an improvement, but now the gray links can't be selected. And it still makes the posts messy.

Dammit! I can display href content in the tooltip but I can't find a way to display the tooltip with pointer-events disabled.

This is currently running on loyce.club/patrol/test.html. The tooltip kinda defeats the purpose if the URL is still visible on the bottom of my browser.

I won't change this (yet). A tooltip would be perfect, but I don't think anything less than that is worth messing up the posts this much when this spam is still very rare.

[nullius]

Objections:

• It won’t work.  This is trivial for spambot authors to work around.  All that a bot needs to do is to make one post which will have a high probability of passing human review from someone who is not following the thread in which the post is made.  The old spammer trick of copying and pasting an earlier post in a long thread can be fully automated—as can a text spinner, to prevent catching that with a bot, too.

What I predicted, I have now observed in the wild:  A spambot account starts with a few posts that look plausible when viewed out of context, with text that looks copied and pasted from somewhere.

• Nuke user: N/A in topic #0 by member #2900957

Meta Log

DateTime	Action	Detail
12/18/2020 12:28:01 AM	Changed to Archived status	Nuke user

Spamfighter advice:  Never unfollow old threads.  Context is your friend.  You best know the threads in which you yourself have actively participated.

I caught this one because its second post was on an old thread that is still on my watchlist.  The post was fully grammatical, and it did not contain any links (not even hidden links).  Nobody would report that as spam, let alone as a spambot automated post.

Because I remembered the context of the thread, I immediately detected that the post made no sense:  Its content was totally irrelevant to the subject matter of the thread.  It was not related to any of the prior posts in the 13-page thread—not even as a topic-drifting side discussion!  But if you were to view the post without reading the whole thread, then you would not see it as spam.  It doesn’t look like spam.

Whereupon I examined the account’s post history, and I found that its first post was of the same nature:  More or less grammatical, no links, sensible within its four corners, and totally irrelevant to the context of the thread.  Read that post, and ask yourself if you would report it.  Then, view the context and ask again.

A list for scrutinizing each account’s first post (or even its first few posts) would not catch this.

I can think of some easy ways for the spambot authors to attempt blending into the context.  For obvious reasons, I don’t want to explain.  Suffice it to say, an intelligent reader who has actively participated in a discussion has the best vantage point for catching any automated tricks.  —Unless the spambot authors were somehow to make their bots write good, original posts that contribute to the substance of discussions!  That would not be a bad outcome.

[LoyceV]

This project got stuck on a post without userID. I don't know what happened there.
I only saw it after 3 days. It works again.

[FatFork]

This project got stuck on a post without userID. I don't know what happened there.
I only saw it after 3 days. It works again.

Is it possible that the scraper was trying to get a post that had been deleted in the meantime?
I see a deleted post in the modlog.

Delete reply: Re: Build a bitcoin app for the tonido app store and win an ipod touch ? in topic #3110 by member #2911613

[LoyceV]

Is it possible that the scraper was trying to get a post that had been deleted in the meantime?

Only if Recent Posts showed the post without the username.

[LoyceV]

I've reduced the update frequency from once a minute to once every 5 minutes.

[LoyceV]

The 1000 first posts cover only one week. I would have expected most of those users to be new users, but it turns out many of them use an older account (which means they just made the first post since I started this topic 5 months ago).

[The Sceptical Chymist]

The 1000 first posts cover only one week. I would have expected most of those users to be new users, but it turns out many of them use an older account (which means they just made the first post since I started this topic 5 months ago).

I know you posted this a month ago, but I just noticed it and find it an interesting fact.  Any idea what's going on with that?  I'm thinking many of them are account farmers, but who knows.  I can't think of another reason why someone would register an account and wait that long to make their first post--and when that's happening with a number of accounts, it makes me a bit suspicious.

The best way I suggested is to provides a bot that can put human to test to be sure that an account user is not a bot but a human.

Oh, you mean like the captcha thing that the forum already has?

[LoyceV]

Any idea what's going on with that?  I'm thinking many of them are account farmers, but who knows.  I can't think of another reason why someone would register an account and wait that long to make their first post--and when that's happening with a number of accounts, it makes me a bit suspicious.

I can think of legit reasons, such as a user who's trying to get a technical problem solved. But if I look at the 10 latest entries, I get this:
Freshly joined altcoin spammer
Freshly joined Twitter spammer
2 years old account creating a 2 word topic
7 years old account posting PhoenixMiner which may or may not be malware
Fresly joined generic altcoin shitposter on a tech board
Freshly joined Telegram spammer
Freshly joined Telegram spammer, in the same topic as the previous user
Fresly joined plagiarism, creating a few topics, no doubt hoping for a quick Merit
3.5 years old account posting in Italian << This one looks like a real user, I've Merited his post.
Freshly joined account posting his username and one Portuguese word

This is more or less as I expected. Real users are the rare exception.

[LoyceV]

Bump into 2022! I stumbled upon this cronjob and had totally forgotten about it. It still works, but I set updates to be less frequent (every 20 minutes).

Since I started this, 185,433 users have made at least one post.

Bump

Bump!

Bump! Go report some spam

[BenCodie]

Hey, nice tool! I've had a scroll through and I noticed that some members who aren't actually posting for the first time are showing up...so it's a little hard to distinguish brand new members from members who aren't brand new. Adding the member rank, activity and/or post count at the time of the post could help distinguish this easily without visiting each post and checking.

For example, user BitDials has over 108 activity and this post showed up in entry 11 (at time of writing):
https://loyce.club/archive/posts/6172/61722610.html

It's not the first post from this user (far from it) but in the list you would not be able to tell in any way until you actually go and view the post.

[LoyceV]

I noticed that some members who aren't actually posting for the first time are showing up...

That's intentional: every user who posts shows up once, because every user can be a spammer.

For example, user BitDials has over 108 activity and this post showed up in entry 11 (at time of writing):
https://loyce.club/archive/posts/6172/61722610.html

That means he hasn't posted since I created this topic, and indeed, he just woke up from October 2020.

It's not the first post from this user (far from it) but in the list you would not be able to tell in any way until you actually go and view the post.

Correct. I'm just surprised how many old users are still waking up to make their first post in years.

[LoyceV]

I was reminded of this thread, so it deserves a bump. Unfortunately, it still nows most first posts aren't worth reading.