do it, ofcourse.
the chance of crashing a computer is much lower than the change of the botnet crashing a computer. i don't imagine they were really this reserved with the small pox vaccine. "should we innoculate?" ofcourse.
ssl provides an encrypted layer which is secure enough to transmit credit card information over the internet on a regular basis. it should be plenty sufficient.
what i don't think is sufficient is how the info is distributed. the thing can be shut down / censored too easily. to make the information distribution resiluant, it needs to be decentralized. that's why i think the website should provide an rss feed that can serve new leaks as torrents. torrent clients equiped with rss scanners can automatically download and seed the leak - this would essentially create thousands of backups of the data as quickly as possible, while also creating thousands of backup connections ("mirrors").
with a per-bandwidth (Mbps) pricing scheme, somebody noted that having a virus or spyware on your computer may become uber-$$$.
also, there are times in the day where there is more traffic on the internet, and times where there are less (higher and lower demand).
There are two solutions that could be done independantly or together:
traffic-adjusted pricing model:
Instead of paying for a fixed amount of bandwidth, you could pay for a bandwidth "factor" or ratio. If you pay twice as much as person x, you get twice their bandidth, thus if there are 2 person x's on the connection, vs. 1 person x, you get less bandwidth.
bandwidth you recieve = your "bandwidth factor" * total available bandwidth / sum [ consumer's "bandwidth factor" ] over all consumers
The bandwidth you get is the total available bandwidth, divided by the sum of the bandwidth factorsnumber of customers transfers data, times your This would be a traffic-adjusted model - you pay more for each amount of bandwidth when the bandwidth is in higher demand. (It would be great if you could dynamically change this ratio.)
Bandwidth self-throttling:
In addition to traffic-adjusted pricing, or apart from it, it would be nice if a customer could set a "maximum bandwidth usage" so that they could regulate their own bandwidth usage and thus have more control over their internet bill. When combined with a traffic-pricing model (such as described above), they could also use this to save money by not paying for more bandwidth than they need.
advantages of pay-per-use
As a side note, I'd like to point out that w/a pay-per-use model, on average a person will pay the same for their bandwidth as they had before. However, since bandwidth will be more efficienctly distributed, they will actually be getting more bandwidth for their buck, on average.
It would also, by turning bandwidth into a commodity, make ISP more competition-based, as they would compete over price-per-bandwidth, which would amount to how good their infrastructure is per the size of their customer base. When an ISPs loses customers, their price-per-bandwidth will go down proportionally, and that'll attract more customers, and vice-versa. This negative feedback loop creates a stable equilibrium at the optimal price point (assuming there are multiple ISPs).
where x and y are prices for downloading and uploading data, respectively. These values could be tied to certain physical connection technologies. For instance, a cable modem connection might have different x and y's than, say, a T1.
This model is just for reference, though, there is a much better model, that happens to be "optimal" in an important sense:
Bandwidth adjusted model (i.e. second-order model):
sum[ x * Mbps * Mb downloaded at that rate + y * Mbps * Mb uploaded at that rate ] over all Mb
since Mb transfered = Mbps * seconds, this is the same as:
sum[ x * Mbps * Mbps download rate used + y * Mbps * Mbps upload rate used ] over all seconds
This second model means that downloading, say, 100 Mbs on a 100Mbps connection would cost 10 times as much as downloading 100 Mbs on a 10Mbps connection.
It means "faster costs more".
I think this second pricing model would be ideal. I originally thought it up when thinking about how one should price processing power in a cloud computing model, but it works for data transfer just as well. This way, everyone is always paying the same price as everyone else for every unit of bandwith that they use. This is the only pricing model that garauntees this.
Assuming each $ represents the same "demand", supply and demand will drive bandwidth distribution in the direction that maximizes the total value gotten from it (summed over all of the customers). I.e. this will maximize the efficient utilization of the total available bandwidth by minimizing the Kullback-Leibler divergence (maximizing the self-entropy).
Slashdot Mirror
do it, ofcourse. the chance of crashing a computer is much lower than the change of the botnet crashing a computer. i don't imagine they were really this reserved with the small pox vaccine. "should we innoculate?" ofcourse.
ssl provides an encrypted layer which is secure enough to transmit credit card information over the internet on a regular basis. it should be plenty sufficient.
what i don't think is sufficient is how the info is distributed. the thing can be shut down / censored too easily. to make the information distribution resiluant, it needs to be decentralized. that's why i think the website should provide an rss feed that can serve new leaks as torrents. torrent clients equiped with rss scanners can automatically download and seed the leak - this would essentially create thousands of backups of the data as quickly as possible, while also creating thousands of backup connections ("mirrors").
with a per-bandwidth (Mbps) pricing scheme, somebody noted that having a virus or spyware on your computer may become uber-$$$. also, there are times in the day where there is more traffic on the internet, and times where there are less (higher and lower demand). There are two solutions that could be done independantly or together:
traffic-adjusted pricing model:
Instead of paying for a fixed amount of bandwidth, you could pay for a bandwidth "factor" or ratio. If you pay twice as much as person x, you get twice their bandidth, thus if there are 2 person x's on the connection, vs. 1 person x, you get less bandwidth.
bandwidth you recieve = your "bandwidth factor" * total available bandwidth / sum [ consumer's "bandwidth factor" ] over all consumers
The bandwidth you get is the total available bandwidth, divided by the sum of the bandwidth factorsnumber of customers transfers data, times your This would be a traffic-adjusted model - you pay more for each amount of bandwidth when the bandwidth is in higher demand. (It would be great if you could dynamically change this ratio.)
Bandwidth self-throttling:
In addition to traffic-adjusted pricing, or apart from it, it would be nice if a customer could set a "maximum bandwidth usage" so that they could regulate their own bandwidth usage and thus have more control over their internet bill. When combined with a traffic-pricing model (such as described above), they could also use this to save money by not paying for more bandwidth than they need.
advantages of pay-per-use
As a side note, I'd like to point out that w/a pay-per-use model, on average a person will pay the same for their bandwidth as they had before. However, since bandwidth will be more efficienctly distributed, they will actually be getting more bandwidth for their buck, on average.
It would also, by turning bandwidth into a commodity, make ISP more competition-based, as they would compete over price-per-bandwidth, which would amount to how good their infrastructure is per the size of their customer base. When an ISPs loses customers, their price-per-bandwidth will go down proportionally, and that'll attract more customers, and vice-versa. This negative feedback loop creates a stable equilibrium at the optimal price point (assuming there are multiple ISPs).
Simplest model (first-order):
x * Mb downloaded + y * Mb uploaded
where x and y are prices for downloading and uploading data, respectively. These values could be tied to certain physical connection technologies. For instance, a cable modem connection might have different x and y's than, say, a T1.
This model is just for reference, though, there is a much better model, that happens to be "optimal" in an important sense:
Bandwidth adjusted model (i.e. second-order model):
sum[ x * Mbps * Mb downloaded at that rate + y * Mbps * Mb uploaded at that rate ] over all Mb
since Mb transfered = Mbps * seconds, this is the same as:
sum[ x * Mbps * Mbps download rate used + y * Mbps * Mbps upload rate used ] over all seconds
This second model means that downloading, say, 100 Mbs on a 100Mbps connection would cost 10 times as much as downloading 100 Mbs on a 10Mbps connection. It means "faster costs more".
I think this second pricing model would be ideal. I originally thought it up when thinking about how one should price processing power in a cloud computing model, but it works for data transfer just as well. This way, everyone is always paying the same price as everyone else for every unit of bandwith that they use. This is the only pricing model that garauntees this.
Assuming each $ represents the same "demand", supply and demand will drive bandwidth distribution in the direction that maximizes the total value gotten from it (summed over all of the customers). I.e. this will maximize the efficient utilization of the total available bandwidth by minimizing the Kullback-Leibler divergence (maximizing the self-entropy).
four words: grass-powered lawn mower.