4 ms·

This. Google has the best spam filters so far. I wonder though, with the advent of new LLM models, it should now be trivially possible to build a zero-shot spa

This. Google has the best spam filters so far. I wonder though, with the advent of new LLM models, it should now be trivially possible to build a zero-shot spam-filtering bot that is self hosted.

cycomanic 4y ago
It used to have, but I would say it's been getting worse. The number of false positives is definetly going up and I would argue that that already gives us an indication of how important Google sees email. If they would consider email a channel that carries imporrtant information they would optimise to reduce false positives not minimise false negatives.
Semaphor 4y ago
Having an old Gmail account and a current fastmail one, i disagree.
- pixelesque 4y ago
  Are they the same target email address though? If not, surely that's not a fair test if the gmail account has been around for 10+ years or so? (I have the same situation, an 18-year old gmail account and a 6 year old fastmail account, but the reason I don't get ANY spam at all in the fastmail account is I only use it for certain things and it's much newer, so I'd argue at least in my case, that's not a fair comparison).
- massaman_yams 4y ago
  Gmail is subject to specific targeting by spammers in a way that fastmail is not. The returns for spending weeks or months finding a niche way through Gmail's filters are justified by the number of gmail addresses that can be targeted, which is probably 3 orders of magnitude larger than the total number of fastmail subscribers.
  - Semaphor 4y ago
    Good point, though it's still in favor of fastmail (which, as a paid service, will always be several magnitudes lower in users)
- alex_lav 4y ago
  I was on fastmail for about a year. It didn't filter nearly as much, and Fastmail as a service was constantly experiencing outages. Pretty much weekly. I would say it's an almost unanimously inferior service.
  - Semaphor 4y ago
    I had 1 outage in 7 years, 2 more i read about while I was sleeping …
    - alex_lav 4y ago
      Did you use it from dec 2021 to dec 2022?
      - nsteel 4y ago
        I did use it during that period (and still do). I've never seen an outage. Are they regional outages? Or do I just go to bed at the right time to consistently avoid them? But the spam filter isn't as good as Google's, 100% agree with that.
        alex_lav 4y ago
        > Are they regional outages? I would guess so? It was days. Many days of "No mail was lost but access is down" messages.
        Semaphor 4y ago
        Sounds like some super weird, extremely local issue. And yes, I’ve been using it as main mail account for (after checking) almost 9 years now, so that includes that period.
        alex_lav 4y ago
        > extremely local issue. I don't think so. They posted on twitter about it as a genera purpose outage. Anyway. I'm glad they exist and I'm glad you've found value in them. I just couldn't continue paying for an email provider I could never access.
meltedcapacitor 4y ago
LLMs make it much easier to make harder to detect spam...
- dmw_ng 4y ago
  The same tech works in both directions. Spammer creates 1000 email variants using a LLM, spam filter collapses those 1000 variants back into easily classifiable embeddings
  - Gigachad 4y ago
    Isn't this exactly the same problem as detecting if text was AI generated which has largely agreed to be impossible?
    - kolinko 4y ago
      Kind of, but not exactly. In the end it's content that matters, not the form that it's send to - I don't care if it's my grandma sending me offers for viagra pills, or a spammer, and I don't care about the language either - I just don't want to get such offers. On the other hand, if there is an e-mail that may be interesting to me, I don't care if it was sent by a human, or by a machine - I want to see it in my inbox. In other words - it's not about distinguishing human/machine written text. It's about distinguishing content that is worthwhile for me from the one that isn't.

This. Google has the best spam filters so far. I wonder though, with the advent of new LLM models, it should now be trivially possible to build a zero-shot spam-filtering bot that is self hosted.

cycomanic 4y ago

It used to have, but I would say it's been getting worse. The number of false positives is definetly going up and I would argue that that already gives us an indication of how important Google sees email. If they would consider email a channel that carries imporrtant information they would optimise to reduce false positives not minimise false negatives.

Semaphor 4y ago

Having an old Gmail account and a current fastmail one, i disagree.

pixelesque 4y ago
Are they the same target email address though? If not, surely that's not a fair test if the gmail account has been around for 10+ years or so? (I have the same situation, an 18-year old gmail account and a 6 year old fastmail account, but the reason I don't get ANY spam at all in the fastmail account is I only use it for certain things and it's much newer, so I'd argue at least in my case, that's not a fair comparison).
massaman_yams 4y ago
Gmail is subject to specific targeting by spammers in a way that fastmail is not. The returns for spending weeks or months finding a niche way through Gmail's filters are justified by the number of gmail addresses that can be targeted, which is probably 3 orders of magnitude larger than the total number of fastmail subscribers.
- Semaphor 4y ago
  Good point, though it's still in favor of fastmail (which, as a paid service, will always be several magnitudes lower in users)
alex_lav 4y ago
I was on fastmail for about a year. It didn't filter nearly as much, and Fastmail as a service was constantly experiencing outages. Pretty much weekly. I would say it's an almost unanimously inferior service.
- Semaphor 4y ago
  I had 1 outage in 7 years, 2 more i read about while I was sleeping …
  - alex_lav 4y ago
    Did you use it from dec 2021 to dec 2022?
    - nsteel 4y ago
      I did use it during that period (and still do). I've never seen an outage. Are they regional outages? Or do I just go to bed at the right time to consistently avoid them? But the spam filter isn't as good as Google's, 100% agree with that.
      - alex_lav 4y ago
        > Are they regional outages? I would guess so? It was days. Many days of "No mail was lost but access is down" messages.
        Semaphor 4y ago
        Sounds like some super weird, extremely local issue. And yes, I’ve been using it as main mail account for (after checking) almost 9 years now, so that includes that period.
        alex_lav 4y ago
        > extremely local issue. I don't think so. They posted on twitter about it as a genera purpose outage. Anyway. I'm glad they exist and I'm glad you've found value in them. I just couldn't continue paying for an email provider I could never access.

meltedcapacitor 4y ago

LLMs make it much easier to make harder to detect spam...

dmw_ng 4y ago
The same tech works in both directions. Spammer creates 1000 email variants using a LLM, spam filter collapses those 1000 variants back into easily classifiable embeddings
- Gigachad 4y ago
  Isn't this exactly the same problem as detecting if text was AI generated which has largely agreed to be impossible?
  - kolinko 4y ago
    Kind of, but not exactly. In the end it's content that matters, not the form that it's send to - I don't care if it's my grandma sending me offers for viagra pills, or a spammer, and I don't care about the language either - I just don't want to get such offers. On the other hand, if there is an e-mail that may be interesting to me, I don't care if it was sent by a human, or by a machine - I want to see it in my inbox. In other words - it's not about distinguishing human/machine written text. It's about distinguishing content that is worthwhile for me from the one that isn't.