Hacker News new | ask | show | jobs
by mabynogy 2889 days ago
I'm not into crypto but I found that method: https://en.wikipedia.org/wiki/One-time_pad
2 comments

One-time pad requires a pre-shared key of the enomorous length to be effective in the World Wide Web. An impossible plan can be vacuously better and simpler at the same time, guaranteed.
Only the length of the message.
The issue with one time pads isn't their security. That's fine.

The issue is key management. Both parties need the same key and it has to be at least as large as the data you want to send. Each set of parties needs a different key.

If you had a method to securely transmit such keys then you could just transmit your data over it instead.

This is why one time pads are only used by countries to communicate with staff overseas. You can send the pads by diplomatic courier for use in communication later. There is no equivalent mechanism for your web activity and every site on earth.

Yes there are. The two parties need to agree on a common source. It can be a file somewhere on the web (an image) or a something that doesn't exist yet.

That's what happens with passwords.

How are the two parties supposed to agree when they've never talked to each other before?

If I connect to https://www.SomeWebsiteIveNeverVisited.com/, how is the web server supposed to tell me where to get the key? Or if I, the client, am choosing where to get the key, how do I securely tell the server where to get it?

Passwords work because they're being sent over TLS which we've decided is "good enough".

Those problems exist with current systems. There is a phase where the two parties must recognize themselves and agree they are legit.
Yeah and you need a new one for every message.
No. You can have a single secret covering many messages.
And totally throw out the "one-time" part of "one-time pad".
No. The secret can be longer than the sum of all messages.
How is a one-time pad going to fix the issues in TLS?

Honestly, it feels like you're treating "one-time pad" as a buzzword without understanding what it actually is. It's just an encryption technique. It doesn't fix the PKI problem. And your one-time pad key needs to be sent over a secure channel. How do you suppose that happens?

I'm not into crypto. Reply yourself to your own questions. You're patronizing.

If I need encryption for one of my projects, I'll try that.

You admit that you're not into crypto, yet above you tried to propose a solution to the problems with PKI, as if the people that ARE into crypto hadn't thought of it.
You show your values and you prove nothing with that sentence.

Experts are often wrong. They exist because because we don't know. When we know something we don't need experts anymore. We just know and apply our knowledge.

Keep in mind the context of this whole conversation. You suggested one-time pads as a solution to PKI and the problems of OpenSSL's large code base being added to projects that need encryption. I don't know how to put this nicely, but it just shows you really don't know what you're talking about.

Yes, sometimes experts get it wrong. Yes, non-experts can sometimes find solutions that the so-called experts couldn't find. I'm not arguing against those claims.

But suggesting one-time pads as a solution to PKI is like seeing someone on the side of the road with a flat tire and suggesting they refill their gas tank.

People have the right to criticize whatever they think is a problem. They don't need to be competent. It's just their applied freedom to think. I just mentioned my lack of interest in crypto to prevent what happened but I'm not surprised that it was useless.

IMHO most people defending HTTPs do that by loyalty because they invested so much time on that and not because they understand all the details of the crypto behind.

My message is just: "It's overcomplicated. I quickly found an alternative. I don't buy the meme".

https://en.wikipedia.org/wiki/Shooting_the_messenger