What determines sync speed

[#1]

If I've read Kitz correctly syncing at peak times should result in lower sync speeds than say late at night or very early morning. I've experienced the very opposite, under 4000 at 6 in the morning and around 4500 mid morning. There can't be that number of I-need- a little-walk-sp- may-as-well-stay-ups around.

[#2]

For a given attenuation (notionally cable-run distance from the exchange) sync speed depends on noise margin, and line noise (the zero-margin point) depends on the line condition and environment.  When people are about (which tends to be peak time) electrical equipment is used and this can increase line noise and reduce noise margin - hence reducing sync speed.
However everyone's environment is different so the correlation between peak time and lower sync speed is not absolute.  Generally though noise margins are smaller when it is dark, and often drop markedly when street lights (and/or security lights) come on.

[#3]

From that I read that I need to lie a bed a little longer, until the street lamps are off, but probably before the business traffic gets going.
Digressing another, to me oddity, is that sometimes a decreased download speed is paired with an increased up load. Naively I'd thought that busy one way bust the other.

[#4]

If by "business traffic" you mean that in big lorries passing the door that could be true.  A busy Internet shouldn't affect noise and sync speed (but it does affect throughput).
On the Internet "busy" is just that, it doesn't matter which way the traffic is going.  Apart from transit capacity, problems can occur at junctions  where the traffic has to be collected and forwarded (parallel routes help here - but the exchange can be a bottle-neck) then at the ends where requests and responses have to be processed and originated.  A farm of multi-processing computers helps at the far end, though a bigger farm might be needed to cope with overload.
The copper wire to the exchange is full-duplex, download and upload use different frequency bands so shouldn't interfere with each other (and the micofilters one hears a lot about are meant to stop cross-talk between voice and broadband frequencies).
I hope that makes sense.