<div dir="ltr">Nah, I *think* that that is an acceptable number -- this is an almost perfect example of a birthday attack.<div><br></div><div>Birthday attacks are weird -- During discussions on the IEEE Mac Randomization experiments I had a flaming row ^w^w polite disagreement with Dan Harkins about the likelihood of collisions of multiple stations each choose a MAC address at random. I asserted that if there were 24bits of random space (>16million), 1000 machines each choosing an MAC address at random would very seldom collide.</div><div><span style="line-height:1.5"><br></span></div><div><span style="line-height:1.5">Turns out I was completely wrong -- there will be a collision once every ~34 times.</span></div><div>This was, um, surprising to me.</div><div><br></div><div>I ended up writing an appspot app to allow people to play with the numbers: <a href="http://mac-collision-probability.appspot.com/calculate">http://mac-collision-probability.appspot.com/calculate</a></div><div><br></div><div>'m too lazy to redo the code to calculate in the opposite direction, but with a 64K space (2^16), you only need 302 keys to have a >50% chance of a collision, and things go rapidly downhill from there (as you fill in the space).</div><div><br></div><div>Birthday attacks are weird.</div><div>W</div></div><br><div class="gmail_quote"><div dir="ltr">On Mon, Nov 30, 2015 at 11:25 AM Jim Reid <<a href="mailto:jim@rfc1035.com">jim@rfc1035.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
On 30 Nov 2015, at 15:56, Warren Kumari <<a href="mailto:warren@kumari.net" target="_blank">warren@kumari.net</a>> wrote:<br>
<br>
> Currently I have generated 9209, and 7285 of them are unique.<br>
<br>
A 25% collision rate is nasty.<br>
<br>
Looks like your /dev/urandom is not much better than Dilbert's. :-)<br>
<br>
<br>
<br>
</blockquote></div>