The crowd roars and Ruckus joins in with 802.11ax kit

If that held us back, we'd still be using 802.11b and 2G. Phones and laptops which support 802.11ax will begin appearing before long, and after a few years more devices that support it will be in use than devices that don't.

The nice thing about having multiple radios on an AP is that some radios can serve efficient 802.11ax, and others can serve inefficient 802.11ac and older, and as 802.11ax takes over fewer resources will be required by the legacy stuff.

Which is like not sending a response packet to a DoS.

They've still used up the airwaves, fought with existing clients, and spoke over them to request anything. Sure, you're not propagating that situation but without protocol changes there's no way to say "shut up and don't ask again" or isolate such requests from the parts that actual data-transferring clients are using.

Additionally, what you're doing then is ignoring random "who's there" probes, which is going to affect auto-join of all kinds (remember - the clients are dumb and may just be trying to connect to favoured network while connected to an unfavoured one, which they can't because you ignore their probes).

At best this is a minor tweak, that will impact legacy clients (maybe in protocol-breaking ways?) and not actually help all that much (e.g. if you have even 11Mbps clients, the probes are an incredibly TINY fraction of the data that they would transmit just to stay online once connected, and mostly passive - SSIDs are broadcast quite openly and clients pick up, they don't really transmit until you join - this is how the old WEP-cracking tools of old worked, they could determine the SSID and WEP key without broadcasting a single byte of data over the airwaves. It's the "thousands of clients" bit that's the problem, and ignoring a portion of them still doesn't make it any better - they're old so they're likely to re-transmit more often to get an answer!).

This is hype at best. If you are so congested that can't fit in a client scanning for SSIDs it might want to join, then you don't stand a chance of transmitting any kind of useful data to any connected client anyway.

10,000 clients sensing networks at even 11Mbps (i.e. taking up the most chunk of spectrum, while also taking the greatest portion of their allocated data to do so) is literally lost in the noise.

The problem comes not from the responses given, but the sheer "waiting time" for the airwaves to be clear before it's safe to broadcast any kind of request at all, and that's determined by the protocol of the client, not the AP.

Re: 1024 QAM?

Just because it supports QAM 1024 doesn't mean it is required. Like previous wifi standards it'll step down until it has an acceptable SNR.

The headline feature of 802.11ax is the subcarriers allowing multiple clients to be transmitting and receiving at once in the same channel, not the usual wifi blather about how many Gbps it can theoretically do.

Unless you count QAM 1024, and it sounds like you agree with me that you shouldn't, wifi speeds haven't increased at all in the past five years. They've just make access points able to use progressively larger and larger chunks of radio spectrum at once. Which is nice for marketing folks trying to sell them, or people who live where they don't have neighbors close enough to interfere and have gigabit internet speeds, but useless for wifi in public places, businesses, etc.