I think that the date usually listed for "last frost date" is misleading. It's usually the AVERAGE last frost date. That is, 50% of the last 30, 50 or 100 years did have a frost after that date.
And of course the average is formed by looking at historical temperature records, so it's less reliable as erratic climate changes increase.
Who wants to plan a garden so that their first sowing is killed by frost five years out of ten?
The "10% frost date" is a safer date to plan with. That's the date after which the temp went below 32 F in 10% of the last many years.
Of course it's different for every microclimate, but right where I live, my "10% safe date" is around three weeks after my "50% average date".
If I was ambitious and had twice as much garden space, I'd plan for two sowings - half on the 50% date and half on the 90% date.
