Very expensive hardware, bad drivers, weeks of troubleshooting and your measurements being used in AI answers, while the source struggle to keep operating. The reaction to RTINGS putting its detailed results behind a membership struck a chord, so I thought my two cents would matter quite a bit. Let me tell you what goes into producing the information people expect to access for free.

You need the device itself which you either buy or get one from a generous brand. But the tester needs not be biased. Sure, there is a difference between supplying a device and paying for coverage, but relying on samples does mean somebody else controls which routers you get and especially when. Getting them too late can put you out of the business unless you offer something else, like a week of data collection...

If you actually get to test it properly, you will find bugs more often than not, so what do you do? You run the test again and again until you can clearly say that it's not some driver conflict between your devices, some bad firmware, some weird CPU behavior. If it's the router or access point, oh boy, you need to speak with the manufacturer and find out why together. Troubleshooting time which can take hours, days, weeks or more. If you don't some manufacturers can get very pissy and goodbye collab. I sometimes feel that it's wiser to just read the box numbers, smile at the camera and you're set! Heh.

Now let's talk about the hardware. I already have a capable server PC, a 10GbE switch that still offers a good performance and a collection of real client devices. This does allow me to do sustained throughput, MLO testing and the multi-client wordkload tests as well.
The problem is that I do not have the control I want over the radio environment and the obvious solution is a more controlled RF setup (duh). I know, I know, I knew years ago, but let me show you what it includes, and this is still very modest equipment:
-two large RF shielded enclosures, one for the router ot AP, the other for the client devices (again, bare minimum),
-programmable attenuation (like Vaunix LDA-802-8) to automate changes in signal attenuation,
-an 8GHz RF sensor + any calibration arrangements for checking the RF paths
-chamber antennas, SMA feedthroughs, coax, fixed attenuator pads and more. You get the idea.

And it's very expensive. My working estimate is around 15K euro using second hand equipment where it makese sense. And that's without the server and clients, since I bought them already over the years. This hardware would help remove roughly a week of setup, supervision and repeatitive manual work (moving devices around) which is fantastic for publishing more often.

Anyway, the point is that the testing itself is not the finished publication. I need to check the results, create the graphs, check various config, interpret and explain the data. Feed it into the Test Lab. It's a lot of work still, but less so with proper equipment which I currently don't have.
So why don't I just buy it? Right?

Historically, ads and affiliations would somewhat cover living costs, buying gear and testing it, but that contract was broken when the AI started eating everything and pushing it in a manner that the source no longer matters. So the revenue has collapsed for most publications, including my own as well. I know what you'll say: YOUTUBE!
I did go that route and an ASUS router review reached roughly 50k views which is good for this type of content. But, the production and the router took money from my pocket and I got 100 euro from YouTube! Well, that covers everything nicely!

So no, just make YouTube videos does not help finance anything, it's another black hole tbh. And for the operation I want to run, my estimate is roughly 50K-100K a year which includes hardware acquisition, testing equipment, operating expenses, producing the article/video and maintain the results. It is not my proposed annual take-home pay. That's why the reaction to RTINGS bothered me that much.

You can dislike a subscription, you can decide that the info is not worth the price to you, but from the production side, this is hardly a betrayal for their audience, quite the contrary because they still want to do things independently.

And what keeps bothering me is the expectations. Some people want the reviewer to buy the products, don't go the sponsorship route, use professional equipment (Octoscope would be ideal, hehe), investigate difficult bugs, test more devices, revisit firmware, retest the device, publish the data and offer everything for free and make sure it's accessible. I am sure there is a joke there about wanting more.
Is it understandable? I guess. Is it reasonable in the current conditions? F no.

If you believe that there are always plenty of free reviews elsewhere, I checked... Oh boy. If producing the in-depth detaield measurements becomes uneconomic, the coverage will shrink to what is cheap and quick to produce. I mean, I bet you can already see the enshitification in real time. The internet will not be empty, there could still be thousands of pages and videos dicussing a router, but the question will remain about how many contain measurements somebody actually performed, under documented conditions, not a blind reliance on specs and conclusions repeated from somewhere else.
Ask Gemini for an answer by all means. See if it can perform the physical test that nobody could afford to run.

So, before the next discussion asking where the independent reviewers went, which funding model are people actually willing to accept while those reviewers are still here? Ads? Sponsorships? Voluntary support? Private commissioning? None of these are consequence free, but understand that "all of the testing, none of the funding" is not a model.