/tech/ - Technology

Technology & Computing


New Reply
Name
×
Sage
Subject
Message
Files Max 5 files32MB total
Tegaki
Password
[New Reply]


ma1rtfjr.jpg
[Hide] (38.3KB, 441x88)
I just had an interaction with the website of some average business, and it made me wonder if email and regular communication is dead. That business has a contact form on their site, they receive 5+ spam emails every single day even though their contact form has a google captcha on it. Meanwhile, people struggle to receive legit emails because every email provider blocks literally everything.

If you put more heavy-handed captcha solutions on the site, you start filtering normalfags and hindering your business. While being able to filter bots would be a great victory, but it's only a half-victory because spammers can use some third world captcha farms to use human labor to solve captchas.

How would you logically fix this? How do you allow people to contact each other without having to deal with this AIDS? The dude who created Freenet has talked about a few techiques that sound interesting. The first is a token cost-based system, there's some thing that generates tokens over time or whatever, and then you must pay those tokens to send messages, and the received can customize how many tokens are required to send a message to them.

The second is a human network -based system where you choose to trust certain people, and whoever other people create a linked list from you to other people, the closer they are to the people you trust the less suspicious they are. The network of trusted people could then be used to judge people and websites/services as trustworthy or untrustworthy.

Also as a side-note: there's several contexts where spam exists. Filtering posts that go into a social media website is different than filtering messages that a business contact email receives. In the former case it doesn't really matter if some people have a hard time or even if some real people get filtered, but in the latter case it could be the difference between staying in business and not. You can use a more heavy handed barrier of entry like requiring a phone number or a really difficult/long captcha when creating an account or mark new/inactive accounts as untrustworthy until approved by a moderator or some algorithm, but that kind of tricks don't really work for someone's personal/business contact or whatever.
Expanding on the different contexts:

For small websites, you could probably filter 100% of spam just by making an unconventional captcha or a weird javascript form that bots don't recognize as a form. Spammers probably won't bother to create a custom solution for some tiny website or even look directly at the website.

Similarly you could hide most email addresses by adding it into an image. Image-to-text conversion is pretty good these days so it's not a perfect solution though. I wonder if you could somehow mask it, like by making a gif animation that starts out empty but then shows the email, or rendering it as an SVG path.
Replies: >>13000
>>12999
>Similarly you could hide most email addresses by adding it into an image.
I've always wondered how effective those hiding in plain sight techniques are, like when people say something along the lines of "you can contact me by sending an electronic message to ass on the domain known as balls.com". I suppose AI might make extracting those in an automated fashion more feasible?
Replies: >>13003
KISS, just make spamming expensive. Ban any subnet that spams, and then reject all emails from them with a note that says they have to pay to get unbanned.
>>13000
I doubt spammers are using some kind of content analysis tools to find email addresses because nobody really hides them like that, at most people add spaces or parentheses around @ or replace it with "(at)" or something. I've seen a few websites that put their email onto an image.

I think sharing the contact address indirectly is very powerful at stopping the vast majority if not all bot spam, but it'll only work as long as it's unpopular so it's not any kind of final solution to spam. Replacing the "@" and "." with images would probably be enough too, it's unlikely that any bot is smart enough to figure out there's an email there unless they render the whole page and use an image-to-text converter on that (which they might start doing if everyone starts obscuring their addresses).
There are some very advanced techniques to del with spam. Take a look at rspamd, you can use bayes classifier https://rspamd.com/doc/configuration/statistic.html and even neural network https://rspamd.com/doc/modules/neural.html to deal with spam.
With how llms are popular today, I am surprised that it is not commonly done.
Replies: >>13005
>>13004
There's way too many people, even in business settings, who speak like troglodytes, and chat bots are indistinguishable from humans at this point. I don't see how analyzing sentence structure can detect a bot for that reason. AI may be able to do something but I'm not very confident about it because a lot of spam mail sound very similar some kind of business offers, who knows though.
1709159499183-tegaki.png
[Hide] (11.7KB, 500x500)
That's what happens when you put your email address somewhere where it can be scrapped or give it to a company that sells it.
There is no solution to that if you want a permanent address inbox open to anyone.

You could regularly change the public address while staying connected with important people with SimpleX which I made a thread about but nobody is gonna use that just to write some stranger one message. I could be wrong though.

So there you go. It's gonna stay fucked.
I receive no sms spam though. Probably because it costs money to send them or because I haven't given it to any shitty websites nor posted on the web.

As others said, putting it in an image is probably a good solution.
Maybe put it vertically in the image and make @ larger, so it can't be ocr'd as easily.
[New Reply]
7 replies | 2 files
Connecting...
Show Post Actions

Actions:

Captcha:

Select the solid/filled icons
- news - rules - faq -
jschan 1.4.1