Why not transofrmer (whether 50Hz or in switching PS) with some massive shielding to conduct any Magnetic fields around it and out of the way ?

At one moment the shielding saturates as well and so becomes inefficient. Already proven, the thieves had strong enough magnet...

Won't a permanent magnet saturate the core of the current transformer so prevent making the measurement anyway ? (or will it appear as high current, so cause the meter to record it as max power consumption ?)

It will. It won't appear as high current, but as nearly no current (only the AC part is evaluated, as the DC carries no power).
But this is the reason, why meters use shunt resistors only, and not current transformers. Yes, it yields to 10's uV signals on the analog input, therefore the need for the dedicated front end IC (and why the generic ADC present in most micros is not enough - the microcontroller just generates way too much noise on the chip, so the sensitive front end must be on a separate die).

The small users dont usually have as much buckets

The bigger users (that also have big electrical installation) or utilities do realise that the extra current does not stop at the breaker, it have its share through all the upstream grid as well (the 2A for ballast at 120V translate all the way up to 1.5mA at 161kV on the big across the state power lines, and further away on bigger higher voltage lines... that may not look like a lot, but think how many of those buckets are in the entire state). So it does make perfect sense to want to correct the power factor "at the source of the problem" and not carry it all the way to the power plant



Then add a big "magnetic reed switch" that will cut the power if somebody goes too excited with the magnet..

Then add a big "magnetic reed switch" that will cut the power if somebody goes too excited with the magnet..

There are many "tampering clues", which are detected and recorded by the meter (it starts from opening the covers, unusual power up/down sequences, unusual communication signal,...), so quite many ways of tampering attempts are detected.
But the best protection against any method is is making that method just principally not effective at all (no magnetic component in the whole meter design means no magnet, whatever strong, may influence the meter functionality at all).

Making power contacts for a power input is quite difficult task (mainly dealing with the high current arc,...), so that means it is just not possible to integrate them into devices like meters, power line modems and so on. They are used only on devices which principally need them (protection devices and power switches)

Many smart meters allredy have contactor to enable remote shutdown of non paying customers. Besides, for meter that is installed in the user's premises (so is not subject to possible vandalism by others), the tamper device can be made as a single use device, requiring a visit from the power company worker to restore power

Well, it is always way better when people just do not attempt to try it in the first place. Here I'm not aware about any use of such disconnect feature inside of the meters (even when they do feature remote readouts and settings - some do display the cost in money, so for sure the unity cost is being updated remotely as well), I think the present laws here do not allow that. But I've noticed a debate about changing the law in that respect, allowing more "flexible" tariffs for customers with more problematic credit...

Why not make analog meters with digital transmitters?

Why not make analog meters with digital transmitters?

Why? You can get precise readings with a digital meter with a PC output.

Why not make analog meters with digital transmitters?

To combine the disadvantages of both worlds? Being complex, mrchanicaly sensitive and expensive mechanics, high power consumption low accuracy metering system, prone drifts causing phase shift errors,...

The analog meters were just fine. The real "problem" that the smart types solve is to make the meter reader workers redundant, so they can be unemployed instead, and paid by our taxes instead of by the power company

So digital meters are just for cutting costs.
I have another "energy saving" question, what about surges when things are turned on? There's a myth that tells you to keep your lights on to avois wasting electricity on startup surges.

So digital meters are just for cutting costs.

The remote readout is not only cutting costs, but as well for users comfort and security: You won't be bothered by the readout visits, yet the meter could be behind your door, so not accessible for strangers.
But there are many meters in the field featuring an electronic measurement "core", but the display and registering counter is in the mechanical form (set of gears and dials, actuated by a kind of stepper motor, getting pulses from the microcontroller). The main driving force for these is a combination of the lower consumption of the system (just below 1W, instead of 5..10W of the classical induction motor based system), meter accuracy (mainly at lower power levels) with still acceptable cost (the mechanical register means there is no need for things like high cycle endurance EEPROM, nor for accurate precision mechanics)

I have another "energy saving" question, what about surges when things are turned on? There's a myth that tells you to keep your lights on to avois wasting electricity on startup surges.

Practically all meters do register the exact energy it consumes, include the inrush current spikes or so. But the thing is, the energy in these is usually equivalent of the device operation for a second or so. The only exceptions are the power tools with huge inertia, but intermittent load pattern and HID's. There the run up takes quite long (a circular saw takes two seconds with about double the rated current input, while the average consumption is about 1/2 of the rated power; the HID needs minutes of nearly full power warmup) before being ready for work.

Why make mechanical display device in an otherwise digital meter ? In case of power outage the readout data can be saved with some NVRAM or battery backup RAM so this must not be a problem

Some point in the far past (i think before you joined LG WattMaster) we had a discussion here about startup energy consumption of a Fluorescent lamp. Back then i showed that in the time of starting the lamp takes actually less power than when working normally for the same time length

Why make mechanical display device in an otherwise digital meter ? In case of power outage the readout data can be saved with some NVRAM or battery backup RAM so this must not be a problem

The problem is, the only NVRAM technologies are based on a floating gate charge and these have limited write cycle life (the tunneling oxide degrades as it is stressed by the erase/write cycles).
The system needs regular data saving, because the RAM register content is prone to data damage by disturbances, power brownouts and so on.
There are many wear leveling techniques, but these are able to boost the 100k native endurance to some 2M or so. If the meter shall work 10 years, it means the data should be stored in a rate less than 2M/year. That means barely 10 seconds per single write - quite long time, even when using quite modern EEPROM. Only the last few years the CMOS technologies allowed to reliably build thin enough oxides, so the native EEPROM endurance become in the 1Meg range, together with the a bit higher EEPROM density allowing wear leveling gains in the order of 100 or so, allowed the manageable 1second write intervals, yet with longer than 10 years life requirement.

Using a battery backup RAM has the battery as the weak spot: There are batteries, that are able to work for 10years, but at room temperatures. The problem is, the meters very frequently work in rather hot environments, where the battery will degrade way sooner. Or the meters are exposed to freeze, when the battery does not work either. This makes the battery really not usable.

So to solve the problems with the induction meters (power consumption, accuracy, tampering resistance against magnets,...), the mechanical register just solves the issues (mechanics allows only increments so playing with magnets means you only add up, the high accuracy and low consumption comes from the electronic front end, the registered value could be read out independently on the presence of the power, there is no data retention problem,...) of the induction systems.

The bidirectional communication ability of the smart meters makes the requirements for the EEPROM way easier, as the data is backed up remotely, so even when the EEPROM fails, the daata could be recovered into a new unit.

Some point in the far past (i think before you joined LG WattMaster) we had a discussion here about startup energy consumption of a Fluorescent lamp. Back then i showed that in the time of starting the lamp takes actually less power than when working normally for the same time length

Exactly. So when in numbers:
The preheat takes about two seconds, the consumption is about 1/3 to normal, so the starting energy loss (it consumes power and does no work, so it is loss) is equivalent to about 0.6 second of a runtime power. Keeping the lamp ON for 10 seconds when not in use means loss (no use => loss) equivalent to 10 second, so already 10x more energy loss than when switching it OFF even for 10 seconds when not in use...
Obviously there is the starting wear and comfort loss related to such frequent switching, but this was just exclusively about the energy use.

So the myth about "Keep the lights on to avoid surges!" is false.
What If the power company had meters at the power generation site/substations instead of having them at each house?

If you didn't have house meters how would you work out what to charge each customer? Consider that you mite have say only a TV and lamp on and your neighbor runs a furnace a range and all his lights how would you divide the bill? With fluorescent lamps I was told only turn off if your going to be out the room for more than half an hour that's just a rough guide its what I got told