8 ms·
Ask HN: Is spam a real problem?
I'm considering developping an anti spam system. To check if there is a real need for it, I keep asking around me to conventional internet users if spam is a problem for them. I'm a bit surprized that nearly all people respond that it isn't because they don't see much of it.
Thus this idea seems to fail the "build something people want" test. What is your opinion about it ?
- jacquesm 16y agoWhether it is a problem or not depends on your usage of email. If false positives are unacceptable then spam is a real problemm. If you manage to come up with a 0 false positives spam detection algorithm though instead of doing you own (end user) marketing I'd go and talk to the 'big boys' instead.
- mike-cardwell 16y agoIs the user of your service the person with the mailbox, or the people running the mail system? Spam is a massive problem for email administrators.
- chmike 16y agoYes. I made the same conclusion. I had the opportunity to talk about it to the mail adlinistrator of a big internet access provider here in France. There is indeed more than 95% spam but 4 servers is all it takes to manage 4 million users ! The mail network traffic % is also peanuts compared to youtube and others. The only real nuisance for mail administrators is when they black list each others by automatic black listing systems. A human intervention is required to sort it out. The cost is thus still too low to justify adoption of my system. On the other side I had an interesting discussion with a big company manager. When I pitched him about my system, he immediatly calculated the cost of spam. With a few tens of seconds per day per employer to filter out the handful spam managing to get through the filters he came out with a cost around a million euros per year. A system sold for a hundred tousand euros to this company could make it. But this doesn't solve my network effect problem. Access providers don't have such incentive. This confirms that B2B would be the way to go.
- alnayyir 16y agoYou want to B2B this. Developing an anti-spam system isn't something you can just hack up though, being any better than the rest requires post-doc grade research backing your algorithms and bayesian filters. Unless you're some kind of latent super genius you're going to want to do a lot of reading of white papers on bayesian filters and how modern systems currently work before I think you really want to start hacking up code. I don't mean to discourage you, and it's good that you're doing some preliminary market validation, but the market is B2B, and the task is far more difficult than most.
- chmike 16y agoThe system doesn't use SMTP and is not another spam filter. The major difficulty is the network effect. It would help a lot if spam was perceived as a real problem. To my surprise it is apparently not. My analysis is that if there is no perceived cost to spam there won't be any obvious incentive to adopt a new system.