in previous videos, we spent some time looking at Bell-LaPadulla and Biba, and I explained to you how they were the basis of computer security, now cybersecurity, for so many years. And, it's because they had a very compelling view for how you might deal with disclosure and integrity properties. And I think you may recall that we looked at how the integrity properties of Biba could be used maybe to protect like an operating system kernel, so really compelling issues. The problem is that, the theory turned out to not match a lot of the experiences that people started to have with real operating systems and networks that were built to try to respect these properties. And again, I've said this and kind of touched on these themes in a lot of prior videos and I want to do it again. The idea of scientist creating theory is wonderful but the theory has to match the experiments, or it's wrong. Something we've all learned from Richard Feynman, if you watched his videos or read his books, we've learned it from Einstein and from all the great scientists, would say you can develop a beautiful theory, you can have impeccable credentials, you can do the mathematics, you can write all these wonderful papers, you can win awards, but if the observation that people make, the experimentation, the real world implementation, does not support the theory, then it is wrong. So now, let's look at a couple of problems that emerged, with respect to both Bell and LaPadulla and the Biba model. And, we'll start with something called, the Blind write. So the way a Blind write works is that, if I'm allowed for example, to write something up like let's say, unclassified, and I'm writing up into a top secret document, which we all agreed is no big deal, if you want to put unclassified stuff in a top secret document, go ahead. So, you can write that, but the corresponding operation of reading what you just wrote, is not allowed in Bell and LaPadulla. That's kind of weird, right? Like, if you're developing a Unix system and you're using a command line interface of the type that you might be familiar with, if you're a Linux hacker or something, you know what a CLI, Command Line Interface, is, while you're typing a command where you're putting something into a file and you're dropping it off into some top secret directory or something, and then you want to go read what's there, even see if it's there, and you're not allowed to see it. So file not found or something. Doesn't that seem weird to you? Would you like to work in a system like that? Most people came to the conclusion they didn't. They thought it's weird. I love Biba but, really? That's how my system's going to work? So, that's one of the first hints that something was not right. And here's the second problem and this, I think, is the final nail in the coffin. It's called the distributed read problem and here's the way this works. Everyone knows, I hope you do, that in the TCP/IP protocol, what happens between say, Alice and Bob, packet's fly back and forth. That's TCP/IP. Alice says to Bob, "Hey, I'm here." Sends packet and then Bob goes, "Oh I see." sends the packet. And then Alice goes, "Ah, here." Sends the packets. So you get this bidirectional traffic and once those packets fly back and forth, an operation can be implemented on top of that between Alice and Bob, like sending a file, writing to something, reading, whatever those operations, those application level operations are implemented after the TCP protocol has actually been established to do the handshake. So, let's say you have a top secret Alice on a network, who wants to write to Bob on the same network. That would be a write. You would know that there would be all this bidirectional stuff happening underneath. Similarly, if I have, I don't know, a top secret process that wants to read something across the networks and a secret file or something, we've said before, that that would be allowed, right? You're okay with somebody with top secret clearance, reading something that's secret across a network. Bell and LaPadulla would say, "That's just fine." So, top secret wants to read secret, do it. But, how do you feel about top secret writing to secret? We said we didn't like that. But it's like, wait a minute. I'm going to allow you to read. Some information can flow this way, but to initiate that I had to send the packet this way. And then pack et this way and then packet this way. And you go,this can't be right. Like, the model doesn't make sense.Do you follow? I hope that that's sort of coming through in the discussion, that the way science works is smart. Computer scientists, smart scientists, physicists, chemists, whatever, they propose theories about the way the world operates. We all look at those theories. And the good ones make sense and we think this is going to be useful over time. They either stand the test of time or they don't. And it turns out, that the Bell-LaPadulla and Biba models did not. And other things have emerged, which will show. But now again, just to sort of test our understanding here, just to do a little testing to make sure that you're sort of following it, we do a little quiz. And it turns out that the answer is C, which is sort of the definition of this blind operation. When you read or write, you can't correspondingly write or read. So, that would be the criticism that we would have of any of these types of models. So, I hope that's useful for you. We'll see in the next video.
