Such connectedness can be done (and should be done) such that the system doing the 'connecting' is separate and independent of the Flight Control System.

That is easily done using monitoring devices that can only monitor the signals and data of interest. It is then that monitoring device/system that is 'internet connected' to report back to home.

For such a safety critical system as the Flight Control Computer, there simply should be no physical way to endanger that system in flight...

Never going to happen. One upload from the ground in flight that happens now is a change in flight plan. Also the fuel weight and balance, weight of cargo and which cargo hold has how much is updated usually in taxi not airborne. All of this feeds into the FMS. Others are all the same ones for any computer such a a BIOS flash. You simply can't allow a rouge ground person the opportunity to plug a carefully crafted update USB into a jack on the flight deck where trust would be implicit. No NO NO! That update needs to be verified by a ground computer and then loaded via satellite radio. There are far too many ways to compromise ramp personnel and there are lots of them. Or that will be the thinking of some safety person and they might be right. The maps that are part of the navigation system are valid for only 28 days. Updated a little more often than your Tom Tom.

That may well be so for the Flight Management System (FMS).

But for the critical Flight Control Computers (FCC)?...

So, where is there any documentation for what is actually implemented and what is actually done?

Martin, how is the FMS going to tell the FCC to turn left at the waypoint and descend if the FMS doesn't talk to the FCC?

Avionics network, fully multiply redundant. Spans the entire aircraft for flight sensors, flight actuators, and cockpit displays.

 |
One way push connection to
 |
 V

Flight Management Network, should include intrinsic or automatic redundancy. Spans the cockpit area for the FMS and associated displays.

 |
One way push connection to
 |
 V

General purpose network for the entertainment systems and internet.

They are one single system.

The maintenance net needs to read the outputs of the FCC and FMS to know what things need to serviced and what error codes they may have thrown. They are one single system. Never mind the maintenance net is likely used to upload software to the FMS and FCC when an update is needed!

In flight entertainment should be separate. However it likely has to share a single (air to ground) receiver transmitter with the other networks. Depending on how that interface works, it may not be TCP/IP over eithernet, the systems may have to talk to each other to sort out who has the bus.

When this stuff was designed, no one thought about a hacker onboard. How many miles of wire in how many planes from how many manufactures will need to be replaced? How many billions of lines of code will have to be rewritten?

The question is: Does Boeing use that technology?

The Boeing 737 MAX story gets yet ever even yet worse:

To my view, a continuing damning story that should never have happened...

I don know that at least two of Boeing's suppliers use such technologies in some of their products, so there is a chance.....
But if the reported attitudes demonstrated recently by Boeing I wouldn't be surprised if they didn't.

... The 787 has a core network cabinet system on-board that includes multiple network modules that segregate and provide network interfaces among the sensitive avionics network, the passenger information and in-flight entertainment system, and the aircraft maintenance system used by engineers, crew, and airline employees...

... Santamarta also spotted two cases where proxy servers used by airlines to communicate with their 787 aircrafts on the ground via GateLink were exposed on the public Internet. "So it was possible to compromise those servers," which could allow an attacker to reach the plane's network over the Internet, he says...

... At the heart of the firmware issue, according to Santamarta, is that the Honeywell firmware was based on a version of VxWorks that was not certified for use in avionics...

... The Boeing 787 core network security controls includes IP table-filtering in the Ethernet gateway module of the core network, where different rules determine which traffic goes from the open data network to the internal data network, for example. The aircraft also runs a firewall packet-filtering function based on a VxWorks library and employs system rules in the network interface module that help isolate the networks, Santamarta says.

Santamarta says that both Boeing and Honeywell confirmed the flaws in the 787 firmware. "However, Boeing did not share with IOActive the version of the CIS/MS firmware they were using in their testing, despite the fact that this information was requested several times. So technically, all of the 787 currently in production contain the vulnerabilities, but Boeing denies those vulnerabilities are exploitable," he says...

Or is that moving to turn the thumbscrews on a monopoly or 'leveraging' proprietary lock-in?

Hmm, here's one big organisation that will ditch technology.
USN to ditch touch screen controls

"Who's to blame? It's the fault of those who should have got their hands dirty and intervened in the bridge's condition but didn't, those who should have spent the money, but didn't, those who should have checked it, but didn't," d'Ovidio says.

Sorry, airbus A321 is an all glass cockpit, so the software and technology was flying the plane. (It just didn't disagree with the pilot.)
Sorry, airbus A320 is an all glass cockpit, so the software and technology was flying the plane. (It just didn't disagree with the pilot.)

Sorry, airbus A321 is an all glass cockpit, so the software and technology was flying the plane. (It just didn't disagree with the pilot.)
Sorry, airbus A320 is an all glass cockpit, so the software and technology was flying the plane. (It just didn't disagree with the pilot.)

Interesting thought process you have.

Perfectly logical one. As said, the software did not disagree with the pilot*. In the case of the crashed planes the software did disagree with the pilot and ignored pilot input.

And Sirius, you are the one with a comprehension problem. Software and technology doesn't mean fully autonomous, or are you one who would get in a Tesla and engage autopilot mode and go to sleep?

Never said the software in the A320 series even has the ability to disagree with the pilot. Boeing built that into the MAX.