Didn't say it was better, but that age doesn't necessarily mean it's inferior. Analog devices have some advantages. Digital devices have a tendency to lock up with power surges, but analog devices will normally recover quickly.
Analog devices tend to be superior in pretty much all settings except for cost-wise and inherently digital applications. Computers MUST be digital, they COMPUTE s***. Your car stereo? You'd be far better off exchanging that DSP for a real volume and tone stack, complete with pots and capacitors and all; though a 20-bit DAC instead of a 1-bit DAC would improve sound quality dramatically.
The big thing about digital devices is they rely on absolute precision, on programmatically predictable cases, and on stable and consistent internal states. Digital devices in mechanical systems attempt to emulate analog devices without analog loss: they measure a temperature or voltage or pressure at a source, encode it, then transmit it; while the analog device produces a voltage anomaly which decays in transmission (longer line, electromagnetic interference, etc affects the precise signal).
Let's take a car for an example, and let's take something simple: a viscous coupling all-wheel differential system (the general design of a torsen works the same way, but with different mechanical basis). This involves two axial differentials (front and rear) and a center differential. There's a lot of ways to do this, and depending on topology both analog and digital systems can do axle-first (rear left slips, rear right gets the power) or can do opposite-wheel (i.e. rear left slips, front right gets the power) power transfer.
In an analog system, you have a viscous coupling fluid which resists deformation. Under light stress it flows, and under heavy stress it solidifies--think silly putty, which flows when under weak force (even under gravity pressure) yet bounces like a set elastic (or in some formulations, shatters like a hard ceramic) under sudden impulse.
When all four wheels are on the ground, the front and rear differentials each have internal connectors churning this fluid at the same rate, so no energy goes into deformation, and no solidification happens; similarly, the front and rear wheels turn at the same rate, and the center differential doesn't experience anything either. Simple so far, right?
If the rear right wheel slips, though, the increase in power there puts more load on the visco-elastic fluid in the differential, which causes it to react by partially solidifying. Because it's now somewhat solid, it cranks against the other wheel's drive shaft, transferring the power from the rear right wheel to the rear left.
So far, we're getting 100% power to the rear wheels, and 0% to the front. What if both rear wheels slip? Well then the power going to the rear spins them, and as they start to slip, power goes to the center differential (as its input shaft--from the rear input shaft--is no longer sending power directly to ground). The rear wheels are limited from spin by the resistance encountered by the center differential, which starts to solidify from the force applied, turning the front wheels--power goes up front, and your car is now partially front wheel drive. Only enough power will reach the rear wheels to drive them without slip; if they slip at all, then their power is not being transferred to the ground (no load on the system), and instead goes into the center diff, which transfers the load to the front wheels, which of course takes the power that is spinning the rear wheels into slipping off the rear wheels.
This system works. It is 100% reliable as long as nothing outright breaks. Even when it is unmaintained and starting to fail, it does what it's supposed to do. When a sudden shock occurs (complete loss of traction to the rear wheels, etc), it quickly reacts because the laws of physics demand a certain mode of operation.
It is also not exactly 100% efficient. Torsen is more efficient (and somewhat better, as it always transfers power: 25% of the power goes to each wheel by default, rather than being 100% RWD until the rear slips, meaning you can shove about twice as much power against the ground BEFORE any loss from slipping wheels occur).
Even more efficient: slap slip sensors onto each wheel. When a wheel is detected to be moving faster than another wheel while it's moving, it's slipping (and when it's detected to be moving slower when brakes are applied, it's slipping and your anti-lock brakes kick in and release the brake on that wheel). Then the computer manipulates a differential to adjust how much power goes this way or that.
Obviously, the computer is trying to remove the "deformation of fluid" or "heat lost in gear system" part of the equation and instead use a less lossy gear system on a computer control to provide a more efficient differential. Rather than laws of physics, just a bunch of calculations go into deciding how to shift the power and to what wheel. This takes, in the end, actually less energy than using a viscous or even a Torsen system; and in fact, a profile can be produced that starts off with a more efficient layout (Torsen starts 25% to all wheels and redirects it when a wheel slips; but putting about 95% to the rear wheels is more efficient at launch) and even selects different layouts depending on vehicle dynamic rather than on "which wheel is slipping right now?"
The downside is, of course, if the digital electronics experience any minor flaw--decaying chips, decaying solder joints, localized static build-up, electromagnetic interference, anything--that changes a bit inside a chip, weird s*** happens. If the computer detects input that doesn't make sense within its algorithms, weird s*** happens. The system is not durable. It can respond faster based on known data rather than natural physical consequences of physical events; it can make predictions and pre-adjustments and such; but it can't handle something going vaguely out of spec.
Analog devices are quite functionally superior. They also quite often drain a decent amount of extra power out of the system, and use more costly or bulky components. Don't take that to mean that analog is the answer to everything, though: some things can [em]only[/em] be done digitally. Analog computers include the Japanese Soroban--lies, it's digital! Beads are set and cleared, and if they're dangling in between then you have a nonsense state (unless you assign a separate meaning to a "hanging bead" or whatnot). Hand-written word is digital: the letter "a" is an "a," and the vague variations in handwriting style and shape are meaningless (this is OTHER information, which is analog).
Plenty of systems to control nuclear reactors and the like are digital because they are computers, they compute things, and they are not a digital analog to an analog system that does something else. It's not a simulation of an oil-filled chamber that's heated by the flow of pumped coolant and somehow measures whatever by expanding and flowing up a graduated tube or whatnot; servos are controlled by systems where we say, "Put this here," and we are sure as hell not going into the nuclear reactor core and pulling a lever to do it. Lots of the automation is logic, not consequence (and thus, on a computer, logic implementing consequence).
If you really want to, maybe you can design systems for that; but are they academic "analog computers" or are they actually systems that are self-stabilizing? Pebble bed reactors are self-stabilizing: when they get hot, they become self-limiting, squelching the rate at which fission can even occur. You take away all coolant and all safety systems and dump the raw fuel into a big, uncontrolled, exposed pile and it will get hot and stay that way; it will not experience runaway fission past its design specifications, and there aren't even any moving parts to control this.
Can you create a system around a BWR that does the exact same thing? No computers, and when something is quite wrong it's okay because the entire design reacts to failure by shutting off elsewhere not because of fuses and triggers, but because some fluid was expanded or some metal flexed or whatever simply because everything was in proper operation, and when that went out of spec those things moved into a different, safe state by design as dictated by the laws of physics? It would be a system where the computer controls activate something, but that something relies on pushing i.e. a fluid around that's only in place because the reactor is i.e. getting proper coolant flow, and when that flow fails something deforms from the heat (temporarily, like a metal that flexes and shifts hydraulic fluid around) and drops the coolant rods into the fuel even though the computer says to raise them and go full-on hot? "Analog safety system" (with a manual override, an actual lever that requires a human to crank it over to supply the torque to undo what it's done)