And, the airport configuration.

Well, the airport configeration has been like that for a long time, what is changing is the volume of traffic and possibly the competence of the crews flying in. If for some reason a particular flight doesn't manage their spacing as anticipated then the crew in the tower have to deal with it.
Because of the closeness of 28L and R it is always a bottleneck for traffic flow when the weather is bad.
Back in the days of "steam" gauges the crew were paying attention 100% of the time, now, it seems, the FMS is programmed and if there is a last-minute runway change then rapid typing results, as the magic "follow the line" has to be displaced.
There is a youtube vid out there showing a Lufthansa 380 flight into SFO, naturally there was a runway change, the comment (from a Snr. Captain) was "typical"... so why was he surprised?.
Fly the aircraft, be prepared for the conditions and pay attention.
SFO isn't Denver.

If you listen to the ATC tower recording, AMX 668 clearly contacts the tower stating he is inbound at "duyet." And the tower clears him for 28R. As we know, however, "duyet" is not a point on the ILS for 28R, but for 28L instead. (The correct point for 28R is "axmul.")

Why ATC didn't notice his call and make sure he was headed for 28R or give him the go around then is a question to be answered.

Additionally, two seconds after clearing AMX to land, ATC clears a Delta to land on the same runway.

Finally, about 15 seconds after clearing AMX, there is a blocked transmission on the recording where the tower and a pilot are talking at the same time. Could AMX have been blocked when asking for clarification on the runway assignment? (There is a call of "blocked" on the Norcal Approach recording, see below, but not for this one.)

The Norcal Approach recording is no more reassuring. The transmission instructing AMX 668's turn onto its final runway heading is cut off so that the airline and flight number is not heard. And the repeat omits the runway designation.

ATC: "...maintain 4,000, turn left heading 3-1-0, intercept 28R localizer."
Unknown aircraft: "Blocked"
ATC: "Aeromexico 668 turn left heading 3-0-0."

Although later AMX 668 is directed to report established on the 28R localizer and the read back is correct. Then AMX 668 is cleared to 2,500, but after read back that is corrected to 4,000.

This exchange didn't help matters any:

AMX 668 " ..."are we clear for the ILS?"
ATC: "Aeromexico 668, uh, stand by....
AMX 668: "Roger."

ATC: "Aeromexico 668, four miles from DUYET, cleared to runway 28R
appr...correction, four miles from AXMUL, cleared to runway 28R
approach."
AMX 668: "okay the ILS runway 28R approach."

If the controller said DUYET rather than AXMUL because he noticed AMX 668 was lined up for 28L, he should have given a clearer instruction to get on the correct localizer. (SFO's own noise-abatement flight tracker indicates AMX 668 was lined up for 28L from when it first turned onto heading 290, at least 10 miles out.)

Finally, I note that at least two prior arrivals had trouble capturing the localizer and had to continue turns to re-intercept.

Combined with the tower recordings, it seems there were multiple clues for ATC that AMX 668 was not aligned properly for the approach and runway assigned, despite read backs to the contrary. Why these clues were missed is perhaps something that warrants investigation.

Agree, especially that SFO isn't Denver.

all those long, straight lines on the ground, close up, against each other.

atlanta comes to mind, dfw..but those app's are generally strung out, progressive, turns to final, (strung way out), gently turned into with mothering radar vectors.

sfo can be way different

now that i think about it, thats why i liked it. it was just a little different, not always the same.

sfo required a little more attention.

otoh, pilots been landing on taxi ways ever since there has been taxi ways. not to mention wrong runways, wrong airports, wrong airports in cities, hell, wrong countries.

another blind spot in the runway monitoring system?

The close spacing of the 28's plus the distance from the tower and sight angle make it very difficult to determine the runway the arrival is lined-up upon.

A video camera situated between the 28's at the approach end and tower monitor would be cheap, easy and effective final check on arrivals.

Without some change this will happen again.

Or when a plane for 28R reports he's at DUYET, the controller could pay attention and redirect the wayward aircraft before a go around is necessary.

and just who would monitor this?

There is already a system in place to monitor the approach, it is the first installed, and was to be used as the pilot for other airports. This system was detailed in the AC thread.

According to the FAA, the system was successful and was going to be implemented at other airports.

The first 'blind spot' found was the AC that tried to land on the taxiway...now this?

It does not appear that the system works, or is ready for distribution

No idea how it's called but you have a system that monitors the complete approach. Too low and off track gives a warning. Should be a nice idea at SFO to use it after 3 times wrong run or taxi way and 1 time the T7 being too low.

So in this case, did it or didn't it?

Either way, questions need to be asked.

No idea how it's called but you have a system that monitors the complete approach. Too low and off track gives a warning.

Sounds like an ILS to me. I wonder if a ground based system, so close to the runway, to tell the airborne guys that they are off track is the best way to go to solve the problem. There are enough bells & whistles on board, plus 2 sets of eyes, that should be enough. We are sometimes finding that the more back-up error beeping systems there are the less the pilots do their job of monitoring. I saw this attitude in the 80's when an operator graduated from a 3 crew cockpit, needles & dials, to a 2 crew LNAV/VNAV EFIS a/c. We had many cadets coming on board and they were being drummed to "follow the Flight Director" and "there's no need to keep scanning the panels as there are bells & whistles & beepers and flashers to tell you when things go wrong". Oh dear. I tried and failed to stem that heresy. That was 30 years ago & things seems to have evolved to worse not better.

How about the advanced system monitoring the flight called a pilot? They were given a clearance, read it back correctly several times, and ended up messing up. Simple confusion and lack of situational awareness, simple as that, has happened before, probably will happen again.

And the controller DID notice what happened and called for a go around, what else is there to see?

I'm fairly certain EGCC has an electronic 'gizmo' that sets off a hooter if a/c deviate from the ILS centre-line. Although the runways are staggered by 1850m, they are much closer together than KSFO. Apologies, but I can't remember the technical name of the piece of kit.

Sounds like APM (Approach Path Monitoring).
Our system has it (as well as DPM- Departure Path Monitoring) but not activated yet.
Basically 3-D polygons adapted that trigger an alert if the a/c strays laterally or vertically from the defined path.
Would have thought very difficult to adapt separate approach paths with such close spacing as SFO has, that didn’t set off nuisance alerts often.

Errr, quite the opposite, SFO runway centreline separation on 28 L/R is on the order of 750 feet...Manchester is more like 1250 feet. Operationally they are completely dissimilar

Absolutely! And well said. It was a simple case of the Aeromexico crew not doing the 'pilot thing'. SFO literally has hundreds of movements a day that go without incident. Professional pilots in a multi-crew environment should be capable enough to follow ATC instructions and clearances without incident.

Qiute easy really. Rename 28R as 29R or 28L as 27L. Similar concept to not having RWY 02/20 such at YPPH.