This will change the monetization of the web as we know it | Hacker News Reader

This will change the monetization of the web as we know it

sitefog.wordpress.com

Flagged for editorializing and unjustified hype in the headline. This has been tried N times before and failed every time; I don't see anything new that SiteFog is bringing to bear (other than using a slower programming language).

If you aren't using Javascript then it isn't the same idea. The whole point of this is that this requires absolutely no installation. It just works automagically when you visit a website.

The main technology hurdle that I see here is that it has to be imperceptible to the user when they "turn on" the botnet feature. This means the browser has to do some analysis of available CPU cycles and keep CPU utilization at a level where it does not affect performance. To do this well, the browser would need to be able to access detailed, realtime systems information, which is not part of any HTML standard as of yet.

(I agree with you on the editorializing/hype issue though)

Who's going to buy cycles from a service like this?

This is the real issue. There are two reasons why I don't think there are many target customers:

1. Most computationally intensive tasks require a bunch of supporting software/data. There are few tasks that can be easily reduced to a few KB of JS running on someone's computer.

2. Privacy. Most people that need lots of computation want to use it to work on their own data (drug companies, financial trading algorithms, NSA/CIA, etc), and they're not going to want some random person seeing even a limited set of their data.

This has been tried before using Javascript - there is no new thinking here. That aside, the problem as I see it, is that the following issues have yet to be solved:

1. You can't currently limit CPU cycles in Javascript reliably

2. You can't automatedly prevent malicious Javascript code from being uploaded and distributed while still allowing people to use the Javascript keywords and constructs that are required for most useful algorithms

3. Any data that is in a format to be processed is not secure because one can view the code and the data, thereby deducing its purpose with minimal difficulty

I find this sort of stuff very interesting 'techincally', but I think I need use-case of a customer who would pay for such a thing.

I can see a market for "anonymous data". Stuff you want to crunch, but not necessarily quickly, but cheaply. I could see uses for spam filters, research, search engines like Google. It might work and is at least likely to be profitable if bootstrapped tightly. Not sold on it being a game changer yet however.

How would it work for spam filters?

Regards

A lot of spam filtering is tokenizing and looking for patterns in TONs of email text. While it's best to spot these trends in real or near real time, there is still some value in crunching older data.

don't we need to be concerned about the quality of the javascript running in the users browser? It seems it doesn't take much for most browsers to get bogged down, so is selling cycles processed with javascript seems less than ideal.

Reply on news.ycombinator.com