The "seat of the pants" stuff isn't as reliable an indicator as you might suppose.
Predictably, there is a tendency for people who have saved money, planned carefully, chosen the device they want and finally driven it
not to say "Well, that was money flushed down the rat hole. I'm an idiot, and this was pointless."
"Seeing" only what you expect to see (and if you're an optimist, what you want to see) is not only normal, it's universal - everyone does it.
The body is only capable of perceiving changes in input (vectors, temperature, weight, loudness) of about 5%. Below that threshold, you may notice a vague change but how much, in which direction? 5% is when you know what you have, + or -.
Testing a car after every small change will not be very satisfying, because you're only comparing C (the most recent change) to B (the previous mod), rather than to baseline A (no change). If you change 6 things that give an invisible 3% each, you'll never "feel it" until you time it or dyno it (or some other non-subjective standard), but you still have +19.4% (+3%^6).
How do worthless bits get favorable reviews?
Especially for engines already highly tuned, any improvement is not only difficult and technically complex, but is generally a compromise (cams move the power upstream, exhaust loses some low speed, high compression retards the spark).
Read some tests of aftermarket chips - of 10 cars (provided by the chip manufacturers, suuure I believe they're stock):
2 of them will not run ("it never did this before")
2 do not complete the test
3 are slower than stock at every speed
1 is about the same as stock - after 6 hours work
2 are slightly faster, but get 15 mpg, overheat and have no torque below 3,000 RPM
Then how did some of those buyers say it worked great?
Very simple.
The car produces 90% of stock power until 4,000 RPM, then 102% above that. The net result is a loss (except for actual racing with full throttle upshifts), but it feels great because "you get kicked back in the seat at 4,000, then it really takes off!!".
The abrupt change from less power to stock power is what they expected to see, what they wanted to see - so they were happy when they got it. The transition was the only thing perceptible.
Noise, of course, makes everything more intense (you're going faster) and masks the other physical sensations (acceleration). A loud muffler + 95% power is always labeled a success.
Has anyone seen a quarter mile and top speed test using a box-stock low mileage car with no preparation for SRI, CAI, exhaust, chip etc. done separately?
A straight back-to-back A:B comparison?
This is why I've confined my actual research and mods to engines so primitive that anything at all is helpful.
Currently working on the 1937-63 Chevrolet "stovebolt" and G.M.C. L6 engines:
http://victorylibrary.com/235BK.htm
Already completed: 1955-67 Chrysler, DeSoto, Dodge, Plymouth "polyspheric" (semi-hemi) V8 engines:
http://victorylibrary.com/POLY.htm