Electronic – Why are modern telephone lightning arrestors less prone to (rectification) RFI problems than old ones

I was reading about telephone lightning arrestors in my ARRL RFI Book (first edition, copyright 1998), in chapter 9 "Telephone RFI" (by Pete Krieger, previously WA8KZH, now K8COM) and on page 9.6, the following is stated (emphasis mine):

The Lightning Arrestor

Telephone service enters a house at a grounded, fused lightning
arrestor located outside at the house end of the telephone company
drop. Years of exposure to weather or moisture can cause corrosion or
discoloration (onset of corrosion) of wires, junction boxes or
components inside the lightning protector housing. If the arrestor is
accessible, a good visual inspection may reveal potential problems.
Lightning arrestors, especially those that have done their job a few
times, can become nonlinear, acting like a diode, rectifying any RF
energy present on the phone lines (just like the crystal radios many
hams built in their early years). If you discover that a lightning
arrestor is creating RFI, the fact that it's rectifying RF is one
indication that it needed to be replaced anyway! Modern arrestors
are less prone to RFI problems than older ones.

Just a few notes about this quote… This is the entire (small) section, so this is likely all the information that I have behind my question. Also, the context does indicate that the lightning arrestor is standard telephone service equipment, and therefore part of a standard land-line telephone installation; therefore, this question may be limited to only standard telephone company equipment. Quite possibly, it may be a broader generalization — I just don't know.

So, it's that last sentence that piqued my interest. Why, exactly, are modern lightning arrestors less prone to RFI problems? Is it more even "wearing" so it doesn't become a diode? Is it a better design, so it lasts longer and fails more gracefully?

EDIT:

I emphasized the "diode" aspects of the quote from the book a little more in preparation for making the following request:

I would like to argue that for me to accept an answer, your answer has to show either how semiconducting happened more frequently with the old style, or alternately show how semiconducting happens less frequently with the new style. This is because I can't really decide between the two really great answers already given. How did a lightning hit (or moisture, or time) cause carbon blocks (the old style) to semiconduct?

Best Answer

The carbon block becomes exactly like a high power TRIAC (with a trigger voltage (<250V), when sufficient voltage and time duration for ionization exists.

There is an incremental negative resistance when triggered (avalanche effect) and is inversely proportional to the current.

Line inductance limits the Re-radiated EMI as it spews an impulse of low current yet widespread on the Telco line. The Broadband EMI is approx. L/-ESR = omega -3dB in a radial direction along to the CO. (Central office)

The rise time dI/dt=V/L is approx 0.35/f bandwidth using 80% of the risetime.(10~90%)

This parallel gap is closer to 3K/mm vs. a sharp edge or point =< 1kV/min dry air. This BVD threshold also degrades with air quality and higher humidity and surface contamination from sputtered carbon.

the above contributes to a much lower BDV threshold.
When contaminants are involved , there is always a canary in the mineshaft effect called partial discharge which sounds like static lightning when there is none. This also has a Relaxation Oscillator effect with random intervals due to ionization and ozone effects. So it might sound like a Geiger counter with self discharges rather than a linear effect of rectifying AM RF broadcasts, but this diode rectifying effect is also possible.
thus large epoxy coated metal oxide varistors are better, but also have a limited lifespan of conductions, so Floridians may need to have theirs replaced on a routine basis. This is based on past experience reading specs not handy at this time.

Using @Mike’s data 0.076 mm = 76V for 1kV/mm and 228 V @ 3kV/mm when new for the -48 to -54V telco DC voltage. thus to me is a good design value for an arrestor and also for a TVS but for an exposed carbon gap not so reliable long term. The relaxation pulses might not occur when the phone is off-hook low impedance but only when high impedance so the capacitance charge can build up from ingress of lightning, then partial discharge egress on the phone lines picked up by AM and HAM radios in addition to real lightning EMI hundreds of miles away which would be much stronger when close and weak far away. (Friis Loss)

If corona discharge on wet powerline bushing is say 100kV triggers the creapage moisture and dust at lower levels, so that when you walk near HV power lines and transformers, you may hear these in the morning with dew, like snap crackle pop which are harmless surface Partial Discharges (PD) But when sufficient dirt accumulates, it can lead to a failure just like carbon gap arrestors if not self cleaning from rain. I have heard HVAC lines crackle overhead in golf courses and back lanes of apartments. when they monitor this with circuits and determine a HVDC or HVAC line is at risk, they then fly helicopters and techs in faraday cage suits to spray clean the insulator bushings.