So, my thoughts are that I don't like the terms load or rate control,
because they can make it seem like the objective is to optimise the
loading or the rate - whereas the success of TCP/IP seems to rest on not
causing **excessive congesion** as a result of times of overload.
The difference in what I call congetsion control concerns the way in
which a person looks at the impact of excess traffic on other traffic.
If the average link capacity increases, and someone designs only for low
latency, or only for high throughput or maximum bottleneck utilisation -
the test cases could tend to look at these most common paths and optimise.
This is where congestion reaction methods become harder to evaluate,
because if you look at self-benefit or equal sharing with friendly
protocols, someone can so easily miss that it horribly impacts something
different (like the impact of long lived flows on video streams....)
Whereas, as a safeguard measure TSV has in the past thought it to be
much more important to know what the impact is on the most impacted
paths. i.e. What is the impact on sharing traffic, when that traffic
has a very different path or a different dynamic variation in the
protocol design?