I have heard from several, nay MANY, of us here that postings were disappearing or "lost". I had a suspicion, pursued it this week, and found I was correct. Hosting a Forum on the web where anyone can create an account makes it a big target for spam, phishing, etc. If you have run or moderated an online forum I am sure you are aware of bad spamming taking over or ruining valid discussion.
SO you know where this is going - those "dropped" or "lost" or "removed" posts were accidentally identified as questionable. And "quarantined". An example of an actual post here that was affected was sent to me. the code sample had variable and type declaration accidentally created an English word that, while not overtly offensive could be considered sensitive. The code I can't reproduce here because, well, THIS article would be quarantined.
Fixing it - now that I know what's happening myself and the moderators are meeting with the Forum providers to get access to and do something about both quarantined articles and hopefully learn how to 'teach' the spam detection to pass more valid Fortran expressions. So we hope we can improve the situation. In the meantime, make a habit to scan code to see if, with removing spaces and capitalization, any code could be interpreted as having sensitive words. Or what I was told "slang or nonsense English language words" in your code samples. So pick non-word variable names, the more simplistic the better. we'll see what we can do to teach this Forum .
I have run many online forums (including this one). Technology exists to dramatically reduce the impact of spam - Intel has refused to use it. A hotword list that triggers moderation is not too bad, as long as the user is notified that's what's happening, but I'd think it would be useless in a technical forum such as this one.
Successful forums block spam before it is ever posted. There are IP blacklists (stopforumspam.com is one), algorithmic filters (Akismet for example), and throttling. Being smart about new accounts is also a potent weapon that Intel chooses to not use.
I remember being very frustrated at the limited tools available to me when spammers hit the IDZ forums. Weeks of this went by, and none of the more common approaches were implemented - it was all "cleanup after the fact".
I have been fighting spam on the Intel forums since 2009. There are many forms and variants of spam, and the spammers/scammers get better every day. If AI dominance could address the spam issue (not that it does not already), it could also be used to address bugs in syntax and code in a fortran compiler.
What you do not see is how low a spammer or scammer will go to post their garbage. Dolls used for only one purpose, morning after pills, and a host of other things (that MIGHT get caught by the filter, but more likely not).
Just as with anti-virus products, it is always a game of catch-up. What surprises me is the number of users who do not care about spam, and do not try to help the community by flagging it. I peruse every post on the forum every day looking for spam, so I know. Then, there are also the forum users who love to write in their own special, inappropriate, coded style to get whatever attention they can. Nonsense keywords? Yes, some are deliberately done by the forum users themselves.
Is a link hidden in a post about a wireless product on a technical support forum valid, or is it stealth SEO? Or, if not hidden, is it free marketing? There are so many examples, I could easily spend the day describing each one.
So, yes, some posts may disappear, and those posts can be brought back. Those few posts are nothing compared to the amount of spam. We hit a peak of 500 per hour several years ago, which made the forum unusable. Today, the forum is kept relatively clean, and it is being constantly monitored for spam and garbage.
Too much of the replies is focused on the impact of spam, and too little is focused on the impact of wrongfully eliding posts declared as spam.
It is (or was more predominantly) bad enough to compose a lengthy response to a technical issue, and then have it get eaten by the system (e.g. system says you have to log in again, or system crash, or system reboot). Most all of these replies are composed at the time of posting. They are thoughtfully prepared while focusing on the issue being addressed. After rejection, the poster has change focus on the "*$#!" system, and either cannot nor are willing to re-do the effort, at least to the extent of the first reply.
Nothing should be rejected without human review as well as informing and returning the offending post to the poster. Also, the human review must be technically competent about the subject at hand and of the nature of the populous using the forum.
We are told that those "mensajes desaparecidos" are quarantined, not deleted. However, a user cannot tell the difference and, as we learned in the case of Argentina's dirty war, "disappeared" mostly has the same effect as "dead".
If the Forum daemons will not inform the user about a decision to quarantine, in effect the message has been deleted. That has happened to me many times in this forum, and not once was I notified. If I had not happened to go back to a thread that was afflicted this way shortly later, I might not even have noticed that a deletion had occurred.
I had a couple of disappeared posts in the MKL forum. I informed the moderator of that forum, and he promptly fixed the matter. I have not seen the problem occur in that forum subsequently.
Please reevaluate your policies, Intel. The policy should be to
(1) reduce the fraction offensives/total_posts as close to 0 as possible
(2) keep the fraction proper_posts/total_posts as close to 1 as possible
It seems to me that you have forgotten or ignored Item (2). If this keeps going unmitigated, you will end up with a Fortran forum with zero posts, all of which are topical and non-spam.
It is depressing that repeated complaints by a number of users have been dismissed lightly. There are other Fortran forums that seem to handle the spam problem quite well. The old version of the Intel Fortran forum gave readers a box to check if they felt that a new post was inappropriate, and that was effective and required little effort from the users.
Sheesh, and Sheesh!
This forum has a "Report Inappropriate Content" option - click the three-dots icon to the right of a post. My experience, though, is that Intel takes 2-3 WEEKS to handle a spam complaint. I am privileged in that I know how to contact some Intel folk directly, but it isn't their primary job to police the forum. It feels to me that spam reports go to contractors (or offshore employees) who have not been trained in how to handle spam complaints, and while they hunt around to figure out what they should do, the spam multiplies. I get repeated emails saying they're working on it....
Most of the spam that has shown up here would not be caught by word blacklists. It tends to be offers for online viewing of sports games, or applications for fake colleges. What one doesn't see here is offensive language.
There are two fronts taking on spam. One, is the spam filter. This is the one that causes posts to disappear based on some rules that I do not know about. Those posts can be reinstated when brought to the attention of the community manager. The community manager can fill in the blanks here.
The other is the manual method. I, and other users, help the community manager with this. When we find a post that has been missed by the spam filter, we mark it as spam. Those post do not disappear. The community manager is notified, and the decision to delete the post, or unmark it is something the community manager (or designate) does.
The spam that is identified and handled manually is done so in roughly 24 hours. The action can be to delete the spam, and to disable the offender, or to unmark it.
So, the posts that are disappearing are being detected by the spam filter.
The post is still there at this writing. I will screenshot it and show you later, if it ever gets taken down. It is the sort of spam that common, rudimentary forum filters would stop and that keyword filters would not.
I could not resist - apology to the Titans
! SPAM.f90 ! ! FUNCTIONS: ! SPAM - Entry point of console application. ! !**************************************************************************** ! ! PROGRAM: SPAM ! ! PURPOSE: Entry point for the console application. ! !**************************************************************************** program SPAM implicit none ! Variables ! Body of SPAM print *, 'A search of GOOGLE shows there are 47251 different ways to cook SPAM as a hambuger, the best uses pineapple.' end program SPAM
Thank you for submitting abuse reports. It usually takes me about a day to get through everything. I don't usually work on weekends but did this for Steve
Our spam tool learns as it goes so it should get better over time, but the software posts are really hard on it because the code snippets sometimes seem non-sensical. That's why the posts disappear until I review and release them. We get about 2,000 posts/mth so sometimes it keeps me pretty busy and I appreciate your patience.
Thanks for pointing out spam and spammers. Just a simple 'ban this spammer' is all that's needed.
Thanks so much! Kind regards,
Support Community Manager
P.S. @JohnNichols , that spam picture was hilarious!
I thought the code had a small touch of humour, I mean it is legal Fortran. Completely useless Fortran. But legal.
I do like my hamburgers with pineapple - it is an Australian thing - it runs down your shirt and makes you look like a dero (derelict grandfather of dubious age and parentage)
Unfortunately, the damage is done. The spammer's goal is not to have the post remain in the forum, but to have the link in it indexed by Google. This takes a couple of hours at most. If you delete the post the next day, the spammer has won.
Filters based on word lists are useless here - spammers know how to avoid them, and they block legitimate posts. Here's what works:
I don't want any special favors - I just want Intel to be proactive at keeping these forums viable rather than reacting way too late to problems.
Steve's list of criteria and rules are quite comprehensive and make a lot of sense.
I add this suggestion to that list:
Until a new forum member attains trustworthy status (say, when the new member completes five posts none of which get blocked, deleted or recorded as spam/malicious, and a week has elapsed since the first post), any links are displayed as just the text of the URL, not as a hot link.
In economic terms, this is equivalent to the exams one takes to become a Dr. or Lawyer, etc.. This interestingly, whilst the absolute best solution, raises the issues of restraint of trade. Of course I do not want my grandmother performing brain surgery.
The real problem is the limitations that groups place on entry. The pass mark for the bar exam is one example.
So whilst I think it is a good idea, it raises other views.
But then again Fortran is not filled with radical people.
Some of these I have brought up before, some have not been discussed here. Doesn't matter, though, as they're unlikely to be implemented. Meanwhile we'll continue to have legitimate posts disappear under the banner of pretend spam prevention.