So in 14k rolls you found two dodgy sequences. Are you ever going to make 7k dicebot rolls? If one of those came along every few hundred rolls, I'd be alarmed.
I've done analysis on sequences of low/high and high/low (etc.) rolls, and there just aren't enough 2s and 12s in the list to get a good view.
Snakes were rolled 396 times in 14,000.
A pair of snakes occurred just 6 times (you'd expect exactly 11).
Snakes followed by boxcars occurred 22 times (again, you'd expect 11).
But 396 isn't a big sample to base an opinion on.
I'll try to summarise the results of my analysis.
Because of the sample size being too small to look at the extreme ends, I looked at 'high' (10 to 12) and 'low' (2 to 4) rolls.
Where the initial roll was Low (2,356 times)
Another low roll followed immediately: 380 times (expected: 396)
A high roll followed immediately: 426 times (expected: 388)
Where the initial roll was High (2,307 times)
A low roll followed immediately: 381 times (expected: 388)
Another high roll followed immediately: 363 times (expected: 380)
Note that my expectations were based on the actual distribution of rolls in the sample, not the probable distribution.
I did the same thing with 'very low' (2 or 3) and 'very high' (11 or 12), but I don't think the sample sizes (at c. 1,200) are big enough to draw any great conclusions.
FWIW, there might be an indication that very low rolls aren't followed by another very low roll as often as might be expected; also, very high rolls aren't followed by another very high roll as often as might be expected.
V-low/v-high and v-high/v-low combinations were pretty much in line with expectations, though.
Sam (if you're still with me here!), would you be willing to post my analysis on your site if I mailed it to you? (I think it would probably be best near the original data).
