Comment on What is the point of the Nicole spam?
null_dot@lemmy.dbzer0.com 2 days agoSure but there’s no evidence that the typos effectively weed out the ones they don’t want.
Comment on What is the point of the Nicole spam?
null_dot@lemmy.dbzer0.com 2 days agoSure but there’s no evidence that the typos effectively weed out the ones they don’t want.
cynar@lemmy.world 2 days ago
No evidence that we have. The spammers obviously think it’s worth doing however, and they are the ones that would have the statistics.
null_dot@lemmy.dbzer0.com 2 days ago
All the evidence we do have demonstrates that the typos evade Bayesian filters and improve deliverability. This is demonstrably true.
When you hear hoof beats think horses not zebras.
Feyd@programming.dev 2 days ago
Provide the evidence?
cynar@lemmy.world 2 days ago
Does it however? I’m not up to speed on modern anti spam, but a huge number of spelling mistakes screams spam to me. I would be extremely surprised if it wasn’t the case. The best way to deliver spam is to make it indistinguishable from legit messages.
Also, the existence of spear fishing implies it’s a choice.
null_dot@lemmy.dbzer0.com 2 days ago
Do you mean to say, you’ve learned to associate spelling errors with spam because most of the spam you see… the spam that gets past your spam filters… has a lot of spelling errors?
That’s just not true. The best way to deliver spam is to send it from a reputable address, and to avoid looking like spam.
Bayesian filters need to be trained by a user identifying email as spam. From those emails it learns which words frequently appear in spam emails. Including spelling errors means more unique words rather than words that look like spam.