What are speed factors?

At their most basic, speed factors rank how fast one driver is compared to another. They are very useful in determining the start order of the event, and also for ordering a reseed.

Typical examples
A super fast racer could have a 95.
A mid pack racer could have an 82.
A new novice racer in a slow car might have a 68.

How are the numbers calculated?
Speed factors are essentially a percentage of your speed compared to the fastest racer on that stage. Your worst stage is thrown out, and the rest are averaged together. Here's an example:

Stage 1. Length, 8 miles. Time for the fastest racer, 8 minutes. Time for you, 10 minutes.
Fastest racer's speed: 60 MPH
Your speed: 48 MPH
Your speed factor for stage 1: 48/60= 0.80 = 80% Your were driving 80% as fast as the fastest person on stage 1.

Then let's look at the rest of your stages:
Stage 2: 83%
Stage 3: 41% (you had a flat)
Stage 4: 89%
Stage 5: 90%

Your lowest stage speed factor is dropped (regardless of whether it's obviously low, like this 41%, or just 0.1 below the rest) and the rest are averaged together. 80 and 83 and 89 and 90 average to 85.5. That would be the speed factor for the event. Your last three events are averaged together, and that's the number used to sort you into the start order.

Notes:
-They are a relatively new invention, created in the mid 2000's, when computing power had become widely available. Prior to this, a much less computationally intensive seeding system was used, base on participation and finishes.
-Their sole intended use is for determining the start order.
-Since three events are averaged, these numbers change slowly, even if a racer suddenly got much faster.
-A variety of drop methods are used, the lowest one, two, or three stages might be dropped, and the highest one or two stages as well, depending on how many stages are being calculated. The goal is to find the most representative average.
-There are two possible reference points: top-referenced, where the computations compare you against the fastest racer, and mid-referenced, where the computations compare you against the "most average" racer, the person in the middle of the field. The example above is top-referenced, which is the simplest to explain and calculate. Mid-referenced is a statistically more stable calculation, as the numbers for all people wouldn't be affected by a super-super-super fast driver showing up making your percentage be lower even though your speed was the same. Mid-referenced has more scaling and computations.