[Cryptography] Photon beam splitters for "true" random number generation ?

John Denker jsd at av8n.com
Sun Dec 27 13:31:14 EST 2015

On 12/26/2015 09:53 PM, Dave Horsfall wrote:

> A good hot flame?  What's the melting point of gold and silicon, anyway?  

material   MP / °C
--------   -------
Aluminum:    660
Silicon:   1,414
SiO2:      1,600
Al2O3:     2,072
Gold:      1,064  (irrelevant, but since you asked)

> What do jewellers use; just a hand-held blow-torch, isn't it?

chemistry      T / °C
---------      -------
Air-MAPP:       2,020
Oxy-propane:    2,253
Oxy-MAPP:       2,925
Oxy-acetylene:  3,500

One amusing low-temperature option is to drop the chip into a pot
of molten aluminum, Terminator-style.  Silicon will /dissolve/
into molten aluminum, much as sugar dissolves into water, at
temperatures well below the MP of the solute.

*** Discussion:

I suggest that a belt sander or even a simple disk sander makes a
more convenient solution.

In any case, the physics problem is relatively easy to solve ...
but that leaves us with other problems.

For starters, a chip can hold many gigabytes of data.  Most crypto
operations, even one-time-pad operations, don't need that much, so
it is wasteful to destroy one chip per operation.

Even more serious is the user-interface problem.  In this forum
we bemoan the fact that users all-too-often choose a low-entropy
password, and re-use the password across multiple sites, because
it is "more convenient".  Therefore it strikes me as unlikely that
ordinary users can be trusted to annihilate one micro-SD card per
message, or one USB stick per message.  It's just too inconvenient.

It seems to me that for user-interface reasons alone, we really
need a flash memory with good crypto-erase performance.  Features
should include:
  -- Can erase small chunks (not just the whole drive).
  -- "Spare" copies of the data are never left lying around on
   the device.  If data is moved for wear-leveling, the old version
   is immediately obliterated.  If blocks need to be moved to the 
   bad-block list, they are immediately obliterated.
  -- Simple high-level interface.  In particular, overwriting a
   logical block should suffice to obliterate the previous contents.
  -- Good efficiency for normal operations.

This is all eminently doable!

A lot of the required pieces are already lying around.  See
and references therein.

More information about the cryptography mailing list