[ring... ring...]

“Help! Help! This is an emergency! Our civil rights are going to shit!”

]]>is that a statistical fact or conjecture? Do odds and averages really work that way? If Yahoo (for example) keeps account records for one year, five years or forty years it would be interesting (statistically) to see if year one is any different than year forty. Court orders, accidents and crimes can’t really be put on a statistical graph, can they?

Right, what could possibly go wrong?

http://attrition.org/dataloss/

http://www.theregister.co.uk/2004/06/07/hdd_wipe_shortcomings/

http://news.cnet.com/AT38T-leaks-sensitive-info-in-NSA-suit/2100-1028_3-6077353.html

http://egocentral.invisionzone.com/index.php?showtopic=2778

http://www.techcrunch.com/2006/08/06/aol-proudly-releases-massive-amounts-of-user-search-data/

]]>But for a large population – say, all Yahoo users – it’s easily possible to study the odds for the population as a whole with great accuracy. (I doubt this study has been done for Yahoo, because Yahoo probably haven’t and nobody else has the data, but it’s possible. It’s been done before for other institutions.)

Odds of disclosure are strictly nondecreasing with time, so yes, in this case, odds and averages really do work that way.

]]>The Child Exploitation & Online Protection Centre is lobbying hard for them to be exempt from this provision of RIPA as they think it should be part of the cost of doing business for the ISPs, but it can be argued that attaching a price tag to each request helps moderate law enforcement’s insatiable desire for data on each and every one of us.

Is HMG’s desire to have a central database of all internet communications managed by them just a desire to get the cost of unlimited access down to a more manageable level?

]]>Sweden, France and the UK are going to choke on their data if their idiot governments persue their various 3 strikes policies. And their customers will choke on their freshly-inflated bills. ]]>

The important thing is not what they charge but when and how they grant access.

eg.

a) Anyone from “the law” at any time.

or

b) With a court order, under some circumstances.

I can understand Yahoo wanting to be reemberised for their time and these fee’s arnt much.

Your paying for an engineer to spend half hour to an hour getting the relivent information from a database, formating it nicely, and sending it off.

If the database is well secure, and the ISP’s are doing their job. This shouldnt be too easy (ie, not everyone on the staff should have access), and they should double check everything.

Id be more worried if they didnt charge.

]]>This is irrelevant. There is some chance of “harmful disclosure” in year one, some chance in year two, and so on. Call these event one, even two, event three, etc. The probability of disclosure in either year one or year two (P(e1 or e2)) is necessarily as large as or larger than disclosure in year one alone. The key insight is that you can’t get to year two without passing through year one first.

So, odds of harmful disclosure are zero at time zero, and increase monotonically and asymptotically to one.

]]>false.

Information NOT kept cannot be released. A null option’s chance is always zero. A zero sided die cannot roll a 1. There is a constant zero percent chance.

If there is a constant low chance of kept information being released, then the passage of time, in fact, increases the odds of said release. Roll a 10,000 sided die repeatedly and eventually you wil roll a 1.

]]>Id be more worried if they didnt charge.

Giving them a profit motive to put you in jail… what could possibly go wrong?

]]>You are no more likely to roll a 1 on *any individual roll* but we’re talking about the cumulative probability of lots of rolls.

You can think about it like this: For a 6-sided dice, the probability of rolling a 1 is 1/6. So, the probability of NOT rolling a 1 is 5/6 (0.83..).

For two dice rolls, the probability of *not* rolling a 1 is (5/6 * 5/6) = 25/36 (0.694..), so the probability of rolling a 1 is 1 – (5/6 * 5/6) = (0.305..), already higher than a single roll.

For 100 dice rolls, the probability of rolling a 1 is 1 – (5/6)^100 = 0.999999. In other words, extremely likely.

]]>It’s one thing to provide info under a court order for no profit, but if a company is profiting on your writing’s being sold to a client no matter what the reason you can demand a cut of the profit’s as copyright holder in your own writings.

I’d love to see that lawsuit.

Then either ISP’s would need to verify they don’t make a profit from selling your writings or you would have a right to royalty statements every time your stuff was sold to law enforcement for a profit.