# What is the measured distribution of block times since Homestead?

The variance of block times should arguably follow a Poisson distribution. But by casual observation I've seen blocks fast as 4 s or 1 m 15 s. The average is around 14 s since Homestead.
If it follows a Poisson distribution it would have not that huge variance I observe:

What is the true distribution of block times in Homestead, derived from empirical observation?

First of all it seems that I am too ignorant to understand Poissons Lamda!
At least I have some R skills left:

``````Summary:
Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
1.00    5.00   10.00   14.35   19.00  155.00
``````

Interpretation: if you take the mean and forget the stupid outliers we count with only 10 s block time!

The good news: 99% blocks are under 1 min

`````` quantile(blocktimes\$blocktime, c(.9,.95,.99))
90% 95% 99%
31  40  61
``````

The peak is around 2.9 s

Source code on Github

Please see these statistics from etherchain.org: https://stats.etherchain.org/dashboard/db/average-block-time?theme=light :)

• Sorry but it shows only the daily average block time. I'm looking for the statistical distribution of block times. A man with one hand in the freezer and one in the oven can enjoy room temperature on average ;-) Jun 10, 2016 at 15:41
• added more details Jun 10, 2016 at 16:05