The current vote percentages at your site, Harris/Trump/Stein 63/18/18, suggests that your sample so far is quite unrepresentative. Is there a reason to expect that to change?
Needs more participants. :) If there were thousands of votes I expect the percentages would align with traditional polling, although if they didn't that would be super interesting.
I think polling has gotten weird because 20-somethings (whatever the correct term is for that generation) don't answer the phone for numbers they don't know. Polling is blind to a whole generation, a generation that can vote in greater and greater numbers. That means that the difference between polls and reality is growing over time.
As far as this poll goes, it's a self-selected sample, which has its own problems.
> 20-somethings (whatever the correct term is for that generation) don't answer the phone for numbers they don't know
Not just 20somethings. I'm in my 50s and lots of people in my general age group also refuse to answer the phone when an unknown caller calls.
Even fewer will consent to taking part in a poll even when they do answer the call, in large part because so many of those calls are push polls and other political marketing rather than real polls. That means they're no different than other scam spam calls.