I have spent the better part of 15 years looking at curves daily from 8 to 5 and often until 2am at home due to my work. A lot of this was to determine speed.
Just as a fine point here, during the catalytic work, one of my co-workers developed a normal emulsion, then amplified the image by catalysis, then did another experiment. He exposed the film and then removed all silver halide and then he amplified the latent image. He got the same inflection point in all cases but the contrast could be made to vary by means of amplification time. The effective camera speed varied due to contrast but the inflection point stayed the same.
By reducing silver halide by 90% he saw the same speed in the inflection point but without amplification was unable to reveal the image speed correctly, even though the contrast varied.
By variations in processing time, he was able to achieve useful camera speeds and contrasts.
So, the point being that even latent image can be made to show the true speed. But toe governs the overall camera speed and contrast varies but is best at about 0.6 for neg-pos materials, and at this value gives the best image at the true camera speed.
This was, in part, presented at the ICPS by myself in 1979 and the two other major workers on this effort. The paper was heavily redacted , and never fully published except as the abstract, but most of it is in EK patents.
All of the reading in the world does not prepare you for doing this daily for years and then seeing the results produced as a product.
PE