At least as a statement about how autofocus cameras work in general, I believe the quote is basically wrong.
Older System
Let's start at the beginning of AF: the Minolta Maxxum/Dynax/Alpha 7000 from 19851. In the original Minolta lineup, there are lenses in two varieties: some send 32 bytes of data, and others 45 bytes of data to the camera.
This does contain the lens focal length, largest and smallest maximum aperture, minimum aperture, a lens ID number, and a few other bits and pieces (e.g., a bit that indicates a manual focus lens).
It clearly does not, however, contain a table that any reasonable person would probably describe as even close to "huge". For example, we're talking about less storage than you'd need to store the first sentence of this post.
As far as telling the lens to move a given percentage toward infinity, or anything similar: again, no. In these old cameras, the camera itself contains the motor to focus the lenses. To focus, the camera applies power to the motor. Although I haven't looked specifically at the motor for the Maxxum 7000, in the cameras I've looked at the focusing is done using a stepper motor. The processor in the camera sends a number of separate trains of pulses over separate wires to the stepper motor, and the direction it turns depends on the relative phase of those trains of pulses.
What this means is that the camera is turning a screw, and sending a separate pulse to the motor to turn each (for example) 1/10th of a turn.
While it's doing that, the AF system is watching the result of the phase detection. When the phase detection system sees the picture getting close to focused, it slows down the pulses going to the motor so it turns a bit more slowly, to (try to) prevent it from over-shooting, and needing to go back the other direction (much anyway).
More Recent System
On a more modern system, his claim at least can be reasonably accurate. Since I started with Minolta's AF system, I'll now consider the most up-to-date version of roughly the same line: the Sony E-mount.
The Sony E-mount uses electronic communication between the camera and the lens for focusing. It's basically a serial port sending commands and data. For focusing, a few commands are supported:
- relative position: focus N% closer or further away than you currently are.
- absolute position: move to N% of the way between closest and infinite focus.
- Hunt: autonomously move continuously at N% of top speed, reversing when/if you hit the limit (and stop when told to do so).
Contrary to what you might initially seem obvious, it appears that when an E-mount camera is talking to a "native" E-mount lens, that last mode (just move until told to stop) is actually the most commonly used mode. The other two are used primarily when speaking to non-native lenses via adapters. As a simple rule, when talking to lenses originally intended for contrast-detection AF, it uses relative positioning, and when talking to a lens originally intended for phase-detection AF, it uses absolute positioning. It can also use absolute positioning with native lenses, but this appears to be relatively uncommon.
New lens/older Camera combo
As far as an old camera with a new lens goes, it's pretty much irrelevant. The data comes from a lens in a specified format, so a new lens mostly just has to fill in the relevant data (focal length, speed, etc.) at the right locations. The fact that this specific lens didn't exist when the camera was designed does't change much.
That said, it appears to be true that at least some modern cameras do contain tables of characteristics of lenses based on the lens ID they receive. This appears to be updated as newer models of cameras are released, so a newer camera may be able to make better use of the capabilities of a lens. I don't have much information about exactly what the camera really does with this information though.
At least in the Minolta/Sony system, there is also more or less the reverse though: newer lenses have added "extended" information that the camera can receive from the lens. For example, newer cameras support "Advanced Distance Integration", where they can control flash power based on the distance to which the lens is focused.
Newer lenses can supply this distance information more accurately than older ones, so a new camera with a (relatively) new lens can control flash power more precisely (under the right circumstances) than an older camera could or than the newer camera can with an older lens. If you do mount a new lens to an old camera, the camera just gets the same data in the old format that it would expect from an old lens. It doesn't "know" how to ask for the extended information that the lens can provide, so it remains "unaware" of its existence (not to mention its content).
I'd add, however, that many more advanced photographers probably don't care much about ADI. Since it's computing the flash power based on focus distance, it only works when using direct (not bounced) on-camera flash. I suppose it probably reduces some of the "deer caught in the headlights" look common to many shots with on-camera flash, but most people who care much about picture quality are likely to use other flash setups for which ADI wouldn't work anyway.
Summary
So, the quote is true to the extent that with some systems under some circumstances, things can work the way he's outlined--but some systems don't support that mode at all, and others that do support it don't necessarily use it all the time.
1. Technically, this isn't quite the beginning. There were (at least) the Nikon F3AF and a Pentax whose model number I don't remember before Minolta introduced their AF system. Neither of these, however was really a complete system (e.g., the F3AF had only two AF lenses, and the Pentax only one), and neither had any significant commercial success.
