Indeed, I fully expect to see some mechanical signal boxes still in use on the national network when I retire in 40 years time. The cost of resignalling is far greater than any savings in manpower will achieve over the lifetime of new equipment.
Extending ATO to the remainder of London Underground (say 150 miles of route not already done) would cost £7bn. How much will it save in staffing? By my reckoning, converting LU to the same Grade of Automation 3 as DLR would save £55m/year. In other words it would take more than a century to repay the capital. If you did go the whole hog and implement Grade of Automation 4 (unattended trains, you'd better pray that no passengers electrocute themselves if they have to evacuate unsupervised), you'd still only save £200m/year and take 35 years to repay the capital.
Extrapolate that to the 20,000 miles Network Rail control, only a few hundred of which are currently earmarked for fitting ETCS, with the added complication of needing to make level crossings and fences completely idiot-proof, and you are coming close to £1tn in capital expenditure.
If the government is planning on spending a few £bn in capital in an effort to reduce operating costs and maybe boost income too, they need to electrify. Diesel trains are maintenance intensive, guzzle fuel, pollute, and accelerate slower. Electric trains on the other hand are fast, clean, and economical. The Treasury needs to pull its finger out of its arse and start spending. One of the few good things to come out of Holyrood is the plan to electrify all Scottish main lines (basically everywhere except Fort William/Wick/Stranraer). Time for Westminster and Cardiff to follow.