> It's only that we didn't knew about it.

This has been well known since the 1990s (Eschelon, Carnivore, etc)

> All the big players can solve this issue for once and for all by providing end to end encryption.

You as the user really has to take responsibility for that, if the $BIGCO holds your keys, they can do whatever they want with them. I've put up GPG encrypted posts on Facebook, no big woo. Twitter is really only good w/ROT13 :P

The only thing thats sortof close to what you're looking for is OTR, but its only for instant messaging.

I have this thought, but not sure if it can be considered a decent solution to achieve machine blindness?

One of the reasons why PRISM/computers or search works is because they can parse the text we type off our keyboards. Those character codes is what helps 'em look for key words, and enables the opportunity to snoop.

So instead if we doodle our text [1] in free-form and in language/lingo of our choosing (augment steganography with raw data/images) then we are perhaps closer to making it difficult for the snooping algorithms to parse and detect the keywords these people bank on.

I have significant reasons to believe that OCR across all languages for 400BB mails a day is next to impossible with available technology today. Given the diversity of handwriting/languages worldwide and the the fact that people can invent characters to communicate with each other, this could become an impeccable method.

[1] Example: https://bubbleideas.com/letters/the-tiger-by-sir-william-bla... (Observe the second half of the page that is handwritten. Since private mails are meant to be read only by intended recipient, we don't need to fret much about search. Basic tagging would be enough.)

Handwriting won't work against backdoors.Backdoors can send screen captures. And since handwriting is slower than typing , most of the times, people won't use it - so it will be a good signal that one is trying to hide something.

> Handwriting won't work against backdoors.Backdoors can send screen captures.

It will be very hard to find/detect context from screen captures without human intervention. Which means only those who need to be snooped will need to be snooped by the Gov, not everybody.

> And since handwriting is slower than typing , most of the times, people won't use it - so it will be a good signal that one is trying to hide something.

This one is important. The example that I linked above has been written off the iPad. On the iPad it's relatively fast to write by hand (clunky keyboards?), and with a Wacom Pentab too, but then this could be the earliest of such experiences that we're talking about. If Tech focus/innovation can be centered around it, people would love doodling as much as typing.

I just did a non-scientific test and it took me a similar time to write on an android smartphone(with a good keyboard) and on paper.

This might work.

But there are some tools for handwritten OCR:

http://www.cvisiontech.com/ocr/file-ocr/ocr-tool-for-hand-wr...

And if we aim for people to write clearly ,unlike doctors, handwritten OCR doesn't seem like a hard problem relative to machine vision in general.

This is not a software engineering problem but social and political one. Thus, any attempt to solve it only by means of some smart engineering would be bound to be rendered uneffective for most or all of the masses.

It is a technological problem as well.

Let's suppose government allocated $200 billion to create this infra. But there are free tools which can render it useless. They are less likely to go ahead with such a plan then.

Whole premise of making inroads in service providers servers is that information stored there is human readable. Store it encrypted. Come up with stronger algorithms and security schemes and this problem is affectively solved.

If we can reach to a stage where only option remain to read a person's information is to arrest them or break into their devices, we have affectively pulled a project like PRISM down.

You are speaking these words as a matter of fact but you couldn't be more wrong. Government have access to this data by law when it's by a court order. Why is the first defence statement made that it's not done on US citizens? Because government will be pulled in court for this. And make no mistake, US companies can be pulled in court over this in foreign countries. So don't make flying statements like 'it was lawful' without understanding how court and executive branches balance each other.

I agree with your point about existence of such algorithms. But we can certainly improvise. What you have written here is like saying that Facebook is just a message board. Of course, technologies exists. It's about how to utilize them and make them easily accessible. And again, there's no framework to force any communication channel. There's no law that data stored on servers cannot be encrypted.

If it is lawful then there's no reason for Snowden to run and hide. In fact you are insinuating that what he has done is unlawful.

I agree it's a political problem, and can only really be solved through policy in the long term, but at the same time the big companies could do a lot more to bring end-to-end encryption to the masses, especially when it comes to chat apps.

I think a viral marketing effort by the people will be good enough to create a mass of people using something like textSecure:easy to use secure(OTR) IM, backwards compatible with the android messaging app.

Still that doesn't solve the the backdoor issue. This one is much harder to solve.