We have been getting some SPAM that is hard to tell if it is or not. Some of you have been really helpful in discerning this. Thank you for your efforts. I'm pretty good at growing fruit but I don't really know a lot about computers. Some of you are adding specific details that makes it clear that they are SPAM. I don't know how to do that, but if you explain it to me, I can get rid of it. Awesome work!
John S
PDX OR
John,
The two posts from “mariahcarey” are more than likely spam. I know on the surface they sound almost legit, but I doubt it. In the July 1 “Thinning 2024” post at 11:40 pm, the poster stated they were in SE Georgia. In the July 1 “When to open your grafts?” post at 11:49 pm, they stated they were in northern Idaho. Kind of hard to be in two places at once, right? Also, if you highlight the “Thinning” post, there is another word at the end-- “wordle”. If you highlight the “When to open” post, the word “mapquest” appears at the end. Suspicious. The sentence construction of the two posts from “amorapotter” that just popped up while I was writing this is odd. Again, there are hidden words at the ends of the posts that appear if you highlight the posts.
JeanW
John, thanks for addressing these directly.
If I'm only 80% sure a new poster is fake, I'll delete to post but not bad the user. Perhaps we should edit the post with a blanket message like, "post deleted" author please follow up with clearly human generated relevant posts.
Sometimes I wonder if we're just helping train the spam AI to get smarter.
Good eye, JeanW.
This "hidden word" technique is created via hacking of the HTML code, that is beyond just AI text generation.
First, the background color of the text must be mined, then the text color is made equal to the background color.
In this case: "<span style="color: #f5f5f5">wordle</span>"
#f5... being light gray.
All dubious posts should be text-highlighted via mouseover-select to reveal a random and otherwise invisible keyword.
This is an old trick to associate the keyword with a website and improve its ranking by search engines.
So it doesn't even matter if no one sees and clicks on the keyword, there is still a benefit to the spammer.
These hidden link posts should be deleted as soon as practical as the Google (and other search engines) algorithm can easily detect them and penalize the HOS website.
https://loganix.com/hidden-links/
^ all the hidden link basics, a wordy article several minutes to browse, but it is relatively recent.
Apparently hidden links are still a thing.
Thanks for the help, people. I deleted the two from Mariah Carey. Thanks for the tip, @JeanW. She didn't respond to my question. You'd think with a voice like that, she'd be able to speak up. If you don't respond to a simple question, you're probably going to get deleted if it's suspicious, or suss as the kids would say these days.
John S
PDX OR
Both recent untidybalk posts are spam (February blossoms, Olive tree...).
Note that the hidden link text is now being placed near the right margin or some blank lines down, requiring a mouseover
of the entire webpage area to reveal the spam link text.
Such crude efforts will remain easy to detect and remove regardless of improvements in AI text portion.
I have not seen these latest. Actors posts usually contain outbound links to gaming sites and if using a PC the easiest way to see them (for me at least) is using the (a) mouse and invoking the posters image/icon of a slanted 8 (the servers permalink). Then all links in the users post hidden or not get found by invoking the (b) keyboard tab key.
I found out you would not have never understood my way that I use to see the links in actors posts that I described to you previously unless you're logged out. Sorry about that John. If your logged out many of the markups for jumping around that are there when logged in go away when logged out. So log out and you can go to Larry's post #9 for a practice, and mouse click the figure 8 icon next to his numeral 9 and start pressing the keyboards tab key. If it takes you twice to get to an object in the next post (ie. #10) then it's because an object is in there via Larry's post #9 that contains a hyperlink from the object. Then practice comparing any of the other trusted sources such as yourself as long as you know you never published a link the way Larry did and count again. So compared to his you while playing on yours should have one tab key less, and that's how you might best know how to quickly spot hidden objects (that are advertised hyperlinks) from new registered members.
The Rooney method described above does use the least number of keystrokes and mouse actions to locate links, hidden or not.
Plus not having to be logged in. But perhaps a moderator then has to log in to take action.
The tab key in this context will also pause on any posted images, and it also pauses on the "Visit my website" link below a member's profile area to the left of the main post.
So far I don't think any spams have included on-topic images, so this should not confuse the tab process.
The tab pause sequence on a forum page can be reversed using Shift+Tab, so you can work starting at the newest (last) post.
You all have been doing a fantastic job of letting us know about SPAM. Some of the SPAM is obvious, like buy these things! I have just been banning them by myself without help. Others have stuff hidden. You have trained me how to see some of it, which I appreciate. For example, quote or reply and it becomes visible. On others, I still can't see it, but you explain why it's SPAM. Great work! There are some uninteresting posts that seem banal and inane, but I don't necessarily feel empowered to ban them until you explain why. Keep up the great work.
John S
PDX OR
Recent first-timer posts of mid-Nov2024 do not have any hidden keyword links;
they are simply non-sequiturs or offer outdated advice that could waste the time of other readers.
If they do not clarify their postings within a couple of days after established member's rebuttals,
then deletion would not be a big loss to the Forum.
That's a fine line. I want people who are interested in growing fruit to have a place where they can learn more. I don't want the HOS forum to be an expert's only forum. "Go away kid, you don't know enough!" is not a motto I want to use. At the same time, if someone is suspected of SPAMMING the group, I wouldn't mind banning them. I'm trying to be careful about it, and I appreciate you all letting me know what they're doing. I 'm learning about it, but I'm sure I don't know as much as a lot of you.
John S
PDX OR
John, I am assuming your recent comment on your thinning pawpaws topic was the result of
a first-timer posting who commented on the time-consumption of hand-pollinating.
There were none of the typical spam markers in the post, but the first-timer, the brevity of the comment, the odd user name,
and the lack of any positive forum contribution of the text led me to do a Web search of the posters' forum name.
(Perhaps Jafar saw something I didn't.)
They appeared on quite a few websites, usually just one posting and all quite recent, and in an odd variety of topics unlikely for one person.
On one site their account had already been banned. I would be curious to know if the quoting function
of this forum can be automated, because that is what that poster did. So it could be someone just playing around.
Still, it adds nothing to the forum. I would not expect any moderator to take the time and search around as I occasionally do
just as a matter of curiosity. It might be quicker to just post a follow-up question to those posters, and if no response for days, delete them.
Topic: Fruitscape Designers, recent posting dated 20dec2024.
The keyword in this posting is the non-sequitur "Tucson".
username legacylandscaper can be traced to the website of the same name.
They specialize in landscaping and only mention use of "toolbox plants"
No mention of fruit or fruiting plants.
This is a Tucson landscaper/hardscaper.
First posting for the username and 9 years after the fact.
You all have done a great job of showing me how to detect some of the SPAM. I deleted more today. The quoting strategy is excellent. However, there are a few more where I can't tell if they are SPAM or not, so keep letting me know, with your more sophisticated eyes, if you suspect that something is SPAM. It's helping.
On some of them, they are the last post, and I can question them. No response, deletion. But on others, I can't tell.
Thanks,
John S
PDX OR
They try to find hiding places. There are three more from this month you had missed. You can get them shown by placing and activating the following line in the forums search:
color: #ffffff
That line is part of the code that makes links not show but I think I can get you to see them through viewing the post through the print activation screen. Each of those three will display viewable texts in the quotes section for each. The added text is there to see if or when your printer options are configured to include toggling 'background colors' one way or the other.
I will email you a screenshot to clarify what I expect your screen to see if it makes it any easier?
I get 783 results when entering color: #ffffff in the search box.
color: #ffffff equal to color: "white")
And that still shows against the pale green background.
So I am missing something about this new reveal process.
Those posts with spam links can be a penalty to the Forum.
I think those posts with a generic and perhaps AI-generated content, that are echoic or agreeable to previous posts, may be harmless unless they induce boredom by sheer numbers.
They often do not offer or solicit anecdotal or academic input to further the topic, but may ironically be beneficial to the Forum metrics, showing more traffic to the Forum website.
jafar said
What do you think of this post?
https://forums.homeorchardsoci.....-1/#p40089
I get the red squiggles on anything the computer doesn't understand, including many varieties in my posts, and your name, Jafar.
John S
PDX OR
jafar said
Neat trick Rooney, thanks. I'd already pulled up several of those to review.
You're welcome.
@ Larry, stay on your way of blocking in text to reveal the tricks misquoted text if that's working for you.
My explanation is, as explained before, opening the post in a print mode and then in the options you can toggle background colors on and off and the translucency blinks the word on and off. Which is hard to miss and that's my point. That somebody missed it before.
Jafar:
You get two in this thread now because I literally typed #ffffff twice, above.
I now get 785 results for color: (783+2 because of my recent entries).
** The trick is just to search for #ffffff, that gets 7 results. Do not use Color:
Then, in the search results, when topic link is opened, on my FF browser, after a brief pause, the page will auto-scroll the offending posting to the top of the page. There, with a mild squint, the white-text link can be seen, or moused over for better viewing.
A couple of these are currently legit, but a handful of spam links are still present.
So Rooney: yes, good trick; simple #ffffff search will include manageable results that will auto-scroll to page top, saving moderator's time.
Can use mouse-over or squint in lieu of print mode.
...but many AI postings do not employ hidden links.
A month ago I had started tracking forum statistics at the bottom for the number of new registries and there has been a five fold increase. The last 24 hours alone exceeded 10 for the first time.
As moderators who delete registries (AI) over the last month, are your deletions about in line with this five fold increase?
Assuming the answer is yes then about 90% of spam content could be getting screened out by the math, so maybe that's a pretty good screen to be left in place against AI. This five fold increase is impossible to explain by a sudden rush of human interest in gardening alone unless young school age individuals are currently on spring break.
I surveyed a few comments via topics of spam/forums etc; including trends, and various administrators are using various tactics that they claim can work for a while. The best bet is to be 'unique', or in other words be clever enough to invent a new way to counter circumvent spam that's unique enough to last as a wall long time.
So here's one thought that I think could last and it comes from myself as a regular user of this form (not moderator) who's seen in the last changes in PHP syntax (scripting) that somebody created an UNANSWERED TOPICS function at the top.
Okay then, whoever scripted that in PHP which amounted to knowing a first post should also be the coder to monitor first time posters which as we know AI/bots usually are. These first time posters when they make these posts should temporarily be brought through to a separate forum topic page. Like I said just temporarily. Maybe 2 weeks.
In that category regulars such as myself can still short notice flag them based on our general understanding of gardening etc; including coding.
ET said
I noticed that by toggling reader mode (that is, clicking the tiny document icon right next to the space for URL) in Firefox browser it's easy to reveal the white hidden link of (some) spam posts.
If my suggestion of committing to put newest posters comments in a viewable public access then that folder can have different styled rules applied to it so that the background changes the kind of contrast your talking about automatically. When one of us sees nothing fishy about the post then the idea is to click a link that lets it pass into the intended forum as desired by the new member. When-or-If we get ranking permissions to act as 'approving people' as such then we must keep out virus definitions current so as to never unintentionally release our forum passwords or the added expense of hiring out these coding changes would be a lost expense.
ET said
I noticed that by toggling reader mode (that is, clicking the tiny document icon right next to the space for URL) in Firefox browser it's easy to reveal the white hidden link of (some) spam posts.
Fantastic idea, ET! I used that and was able to decipher a couple of the SPAMs that I couldn't tell before. Thanks,
John S
PDX OR
Idyllwild
simplepress
jafar
Marsha H
Viron
1 Guest(s)