While it is hard to imagine the mechanical "toilet lid" part of the K-Jetronic CIS system working well during high 'g' manouvres and having a generally long service life, it was obviously well enough engineered to satisfy many "sporting" customers.
In Porsche and BMW land, Kugelfischer mechanical systems made superior performance possible, but even the renowned German engineering did little to give long or satisfactory service. I believe that regular service for simple cleaning and fastidiously clean fuel, air and oil would keep the system performing well for over a hundred thousand miles, but most owners would not oblige.
The simpler Lucas system was worse for road car use, but was chosen for many serious Continental and even some Oriental racing engines as well as the expected British examples. Bosch took over Kugelfisher's injection business mid-70s and could provide more precisely tuned mechanical systems than Lucas. Electronic systems were well accepted in road cars long before the pure racing industry perceived a need for greater sophistication.
Three facts underlie this.
1) IC engine output changes very slowly over quite a broad A/F range
2) The possible failure paths increase with the number of components in a system, so these additional components have to have greater reliability for the system as a whole to not lose reliability.
3) Performance gains were possible with superior fuel atomisation from high pressure mechanical systems. Even today, most production EFI systems are low pressure and prepare fuel less well than a carburettor. Of course, poor manifolding can undo the carburettor's good work.
The reduced aerodynamic clutter in the induction tracts speaks for itself.
Although the weight of fuel is always significant in high performance racing machines, reliability risks could be avoided until fuel consumption regulations were introduced.
The need for more detailed and tighter control was also pushed by special fuel brews, rising RPM and advantages found controlling fuel temperature.
Tight, closed-loop control was needed in production cars far sooner for catalyst operation. Even before complex integrated circuits, microprocessors and digital control significantly increased reliability and reduced costs, a simple electro-mechanical system like K-Jetronic was probably cheaper to manufacture than a complex carburettor. Thus, while emissions were the original bogey, in mass-production high-tech actually helped profits, when the technology was mastered.
However, I sympathise with Pento's Jetronic drivability issues. Most examples I drove (never on the track, and did not include VW) felt like the throttle was connected with a soft rubber band. Notwithstanding, the contemporary Japanese carburettor cars were worse with tardy dashpots and air-control valves. Nevertheless, I find many recent 'Drive-by-Wire' systems are little better.
As Tristan appreciates, all quite easily fixed with a DCOE (or IDA) choke per cylinder, but not exactly cheap these days. Tuned well, they will be almost as economical as a similarly well tuned EFI system, unless the engine is wild. In that case, you would need to spend a fortune mapping a complex EFI sytem to be superior anyway. After-market injection is probably cheaper in the long run, especially if you want to tune it yourself. Make it like the Pug.
Manufacturers have very little interest in making their cars perform like new at ten years. Most of the desirable and affordable RWD cars for club racing/track cars are over fifteen. It is unfortunate that RWD is becoming more exclusive/expensive or historic.
If I hadn't written mine off, it would be thirty-five
Now I have a family, I think a lower powered but newer FWD might one-day be affordable and an interesting project.
As Turby states, there are sophisticated and affordable logging A/F meters available. If you have RPM and throttle position logged as well, you have the basics to do some serious fine tuning, but Nb 1) above.
All academic, of course, unless you have a tunable system.
In a turbo-car, or an atmo unit that is making a lot more power than the original, induction air temperature becomes more significant. Reducing and controlling it is necessary to find and maintain performance. As power-to-weight rises, more installation variables must be considered part of the tunable system. Just changing the fueling on the basis of an A/F reading can send you 'round the twist.
This is a good reason why a logger like the DL1 is perhaps better value than an A/F logger, because we produce a speed, position and lap number reference that is much easier to interpret than a time based record, even if it includes wheelspeed as well as RPM.
After this epistle, the simple summary should be made:
It's very much 'horses-for-courses'