Protect Python Source

[Bengt Richter]

On Fri, 01 Nov 2002 08:57:35 GMT, Alex Martelli <aleax at aleax.it> wrote:
[...]
>
>It is conceptually impossible to stop a halfway-competent 'cracker'
>from disassembling any compiled form of your program if they have
>any interest at all in it (or even if they don't: they DO do it just
>for fun even for programs they have no earthly use for), as long as
>you distribute your program in a fully executable form for general
>purpose computers.  If the GP computer can get at your program, as
>it must to be able to execute it, so can the cracker.  In some cases
>you may be able to achieve a measure of protection if your program
>can only execute when connected to some Internet site: in that kind
>of scenario, you may be able to check that connections only come
>from validly registered sites.  But in general, you should rely on
>the law to protect your programs against unauthorized use, not on
>illusory "protection".  "Security through obscurity isn't".

I am afraid that the "GP computer" will soon have right in the CPU chip
the microcoded ability to input a byte stream from memory that is
digitally signed and encrypted with the CPU's public key, and
decrypt it into an on-chip cache memory that never gets written to
RAM (or if it does, through an xor prng encryption mask), and execute it.

I think they're moving the concept of a computer receiving trusted code
securely over an insecure network to a concept where the CPU chip
looks at the motherboard as part of the insecure network.

It's not a big conceptual leap. The details are just a bunch of engineering.
(However, no doubt this simple conceptual move from computer to CPU chip
will be loaded with patents on obvious ideas that have been in place in
bigger contexts for ages).

I think you can resign yourself to the idea that it will eventually
become as hard to decrypt a private message to a CPU as it is to do
the same to a PGP-encrypted email message to you. The message will
only be opened on-chip, using the chip's private key. It'll execute
as written or it won't execute at all. Either way only the CPU will see it.

Obviously you have to build stuff into an OS to make use of this stuff,
and no doubt it will start with the BIOS below that.

There will be a lot of opportunities to form exclusive clubs of hardware
and software vendors. Even if the technology itself is as open and royalty-free
as PGP, so that Linux/BSD/etc can also use it for bullet-proofing, it could mean
that the latter will not be able to interoperate except when others choose,
unless non-exclusion is guaranteed by laws with really big teeth.

The technology is neutral, but the temptations it affords are not.
Oh well, I guess it's all part of economic evolution, moving from
selling atoms to selling bits to selling permission ;-/

Regards,
Bengt Richter