Hacker News Reader

Comment by eof | Hacker News Reader

Interestingly, someone has proposed basically this idea in 1997 http://en.wikipedia.org/wiki/Hashcash

The idea is that sender has to do a non-trivial amount of computation to find a hash below a certain threshold for receivers to accept it; but the amount of computation would be low enough to not adversely effect day-to-day email transacting; something like 1/20th of a second worth of computation of a cpu would probably be about right.

As a side note, spam hasn't been an issue for me for a long time.

View on HN

I like the idea but wouldn't that also make a lot of legitimate use cases of email too costly? Think of a service like MailChimp.com: They allow you to easily manage and send out newsletters. Computing the hashes for 5000 newsletter receivers would take approximately 5 minutes of raw CPU power.

Also: The CPU power needed has to increase every year because CPUs get better and better (and thus cheaper).

Then again: Let's assume Hashcash would eliminate spam or reduce it: This would probably mean a production boost (no humans have to detect spam anymore) and spam filters cost CPU as well...

In the end it may be a 0-sum game but reverse the burden. The more I think about it the smaller the potential problems seem to be...

Hashcash is actually used by some spamfilters, but not as a sole determining factor. For legit services like MaichChimp, they could be whitelisted (so always clear spam filters) or they can get a bonus to their rating (negative? whichever way would make it less likely to get marked as spam) and then determined to be spam on its contents rather than the absence of a Hashcash value.

I completely agree! I implemented Hashcash a couple years ago into my email wrapper and the algorithm is (in it's essence) suprisingly similar to Bitcoin: given this data, find a hash that has a consecutive number of x zeros in it.

I don't know about the 1/20th of a second though. Back in the day, the slowest Hashcash stamp with 5 ending zeros I could produce (with optimized code of course) was marginally above 1 second. But the good thing about Hashcash is that, much like Bitcoin - the difficulty scales. Just ask for more zeroes and it'll take longer. Email clients could also use the number of zeroes to help email classification: 0 zeros = SPAM, 5-8 zeroes = HAM.

HashCash is not surprisingly similar to Bitcoin; HashCash was the inspiration for Bitcoin. HashCash is one of the few examples of previous work that the Bitcoin paper bothers to cite.

I was going to post the same thing, have always like the idea of hashcash and surprised it hasnt recieved greater usage specifically for things like captchas although in the case it would be a reverse captcha :)

edit

Could potentially be used a a form of ddos mitigation if the browser was required to design a tiny bit of hash cash for each request, although I haven't really thought through all the details.

Reply on news.ycombinator.com