This is what happened: he tweaked settings galore, hoping to squeeze out more performance somehow. Nerds tend to do this; it's natural.
However by removing the automatic settings and forcing it into one or more configurations that may have been suboptimal, performance actually decreased.
When he returned the settings to "fully automatic" everything was fine and dandy and maximal.
Moral of the story (and I quote): "Resist the urge to tinker with these settings."
You can basically ignore everything except that last line—and if you're familiar with the excellent AirPort Extreme, you probably already knew that.
However, it's possible that the flexibility of the settings toward other cases (like using the router with older hardware, or less compatible or buggy devices) necessitate their presence and complexity. Seems likely to me that each of these settings has been placed due to a specific support case with the plethora of device data and testing cases that they must have from years of customer support.
I think, then, that the settings are not for the maximal optimal case, and the misunderstanding here is that the router is somehow not initialized for maximum performance. This may be true for other router brands, but in my experience, is not true for Apple.
Changing it to a single name for both networks leaves my iMac always connected via 2.4GHz and no obvious way to switch to 5GHz.
I can't say that this configuration change will help for everybody - I'm just reporting what I'm seeing.
Now I'm on the 5GHz again I seem to be getting a more consistent 450Mbps link speed although actual throughput seems the same.
I think I might just leave it on fully auto now, it seems to work pretty well.
* Connect to the 2.4Ghz network
* Hold "Option" and click the wifi icon in the top menu
* Note the "Transmit Rate"
* Repeat this for the 5Ghz network
When I first setup my Time Capsule I noticed that my Mac often times was attaching to the 2.4Ghz network. At that point I separated the networks and did the above.
Low and behold, each time the Mac picked the 2.4Ghz network it did so because it had a higher transmit rate than the 5Ghz one in a given location.
So now I keep separately named networks and can usually connect at 450 mbps (mbp) or 150 (mba). At least I believe that is the case, I can double check at home.
This is a classic case of someone geeking out over settings and over-optimizing themselves into a corner. Hence the last line of the article, and an excellent moral.
Saying that, I see no downsides to leaving it fully auto and I think I'll leave it at these settings now.
1. Disable your other airport expresses and the 2nd airport extreme
2. Place your iPad as close to the airport extreme as is reasonably possible (the one connected to the cable modem... just to try to get the max speed – it would be interesting to see what the speed dropoff with distance is in a second, separate test)
3. Try the different options again (especially connecting on 5ghz using a different SSID with the airport extreme's 5ghz network set to n-only)
It's not just the wifi interference (which shouldn't be a problem, but could you also post how many other wifi SSIDs are visible from your iPad?) – but multiple airport expresses mixed with an airport extreme falls outside the realm of what 99% of users do (I think it could be a bug when the airports interact).
As other posters have said, this kind of performance drop is abnormal.
I don't think the Airport Expresses have anything to do with it - I've had some reports of other people being able to replicate my results without that.