Could you discuss what is important or relevant on what the reviewers seem to omit in their reviews.
A useful review would include things that are operationally relevant. For example, short all outputs together when the supply is powered. How much power does it then output (and, of course, not fail)? Does its current foldback limiting work? And do those numbers meet ATX standards? How long does its fully loaded output provide power after input power is removed? How low does AC voltage drop without any affect on DC (output) voltage? How fast can loads change with no adverse affect on output voltages? For example, how fast can load on 5 volts change without even affecting the 3.3 volts? How well does the supply perform with various reactance (not resistance) loads? Sometimes a supply can fail intermittently, not power up, or oscillate when load reactance is too extreme. How extreme?
What is ripple voltage under various loads? How stable are all voltages when individual loads vary? Can it output a full load even when room temperatures well exceed 100 degrees F (what is a normal temperature for all PCs)? How well does the other supply inside a PSU (+5VSB) also meet these same requirements for short circuit, load variations, etc?
How robust is the supply? How many thousands of volts can its input suffer without damage? How well is its line filtering? How severe can AC line noise be without affecting DC output voltages? Does no noise appear on any adjacent longwave or AM radio when located adjacent to that supply? Does it really conform to FCC and other international standards? Does it exceed its power factor numbers? How 'dirty' (ie %THD) can input power be without causing operational problems? How high and low can AC voltages actually be without causing any output power interruption? What are the actual threshold voltages for the Power Good signal? Or does that power Good signal even report anything useful? Does the 3.3, 5, and 12 volts always remain higher to each other as power starts and during power down under various loading? For example, should the 5 volts fall faster than the 3.3 volts, then eventually semiconductor damage can result. No supply must even cause damage to its load (ie motherboard).
Does the supply actually have overvoltage protection? What it that threshold voltage for each output? What are the actual efficiencies at various loads. Is its maximum efficiency at full load or only at half load? What is a acoustics noise in dBs? Does it use smart fans? Does the supply protect itself when no air flows through its box? How high must room temperature actually be before that supply safely powers off? What safety compliances does it meet? How does it conform to various international standards from EN61000?
What is its rated MTBF for full load? Just some relevant parameters for all supplies. How to select a supply based upon what is important. Visual inspections report near nothing. What does the supply actually do? For example, how good is surge protection that is found inside all appliances? Do all reviews report these basic operational specifications?
So many know so little about computer power systems (which is more than just a PSU) as to automatically blame surges for failures traceable to manufacturing defects. Therefore the PSU market is ripe for dumping inferior products in a market dominated by computer assemblers. Who select a supply only on dollar, watts, and hearsay. Why do so many buy 800 watt supplies for a computer that does not even consume 350 watts? That draws well less than 200 watts most of the time? If a supply is grossly oversized, then one need not learn relevant and simplest facts such as how many amperes for each voltage.
If reviewers were discussing these relevant operational requirements and numbers, then most of what is in this post is familiar (if not understood) by the reader. Good reviewers should be informing a consumer of factors that determine PSU operation. Some power supplies may be so poorly designed as to not even provide a Power Good signal in response to a power system failure. How would we know? How often do reviewers test the Power Good signal operation? Does it do what it is suppose to do?