Why k-mer=31 was not considered in kraken2 db?

[LilyAnderssonLee]

Hi, thanks for providing this very valuable resources. I am wondering why default k-mer = 31 was not considered in kraken2 databases. To avoid high false positives?

For Illumina short-gun metagenomic data, I guess 50 and 70 are two common ones used during the classifications. Do you think k-mer = 31 will improve the performance of the classification?