It's surprising and disappointing to see this point of view here, particularly with so little evident dissent.
Of course you can "break into" a typical computer system by gaining physical access to it, for example by breaking into the house that it's in and unscrewing the computer's case; but that's only metaphorically connected to what's going on here, which is that criminals are sending data over the internet to the computer systems in question. The software the owners previously installed on those systems then responds to that data by giving the criminals complete control over the system, as long as they care to maintain it, or unless the system is destroyed. This is dumb.
Writing software that does not behave in this fashion is not only physically possible; it's actually the majority of software. Even in typical software, there is only one deployed exploitable security hole per thousand lines of code or so, and, until only about 25 years ago, it was reliably possible to recover from such an invasion by reinstalling the OS.† The best software, like seL4 or qmail, has orders of magnitude less, though we can quibble about whether the actual number is 0 bugs or 1 bug.‡
The problem is that our systems are architected so that even one exploitable bug anywhere in hundreds of millions of lines of code enables total and irreversible subversion of the system; our system complexity is growing much faster than existing code is getting audited and fixed; much of the code is not even open to auditing; and the people with the power to fix it have no incentive to do so, instead spreading pernicious misinformation claiming that usability and security are unavoidably in conflict (a concept obviously absurd to anyone who has had to use an OS without memory protection) and bulletproof security is impossible anyway. So, at any given time, there are somewhere between thousands and hundreds of thousands of exploitable vulnerabilities in our systems, any one of which is sufficient to enable the implantation of a persistent backdoor that cannot be reliably detected or removed.
The solution to this has been known since the 01970s. At the systems design level, minimize the complexity of the trusted computing base (the hardware and software whose integrity every program in the system relies on) in complexity, audit it rigorously, and freeze it. At the hardware level, provide an easy incorruptible way to restore a known safe state. At the social level, ensure that the people who rely on the integrity of the computer system have the authority to audit it and fix any problems they find, and the technical competence either to do this themselves or to delegate these tasks to people who are competent to do it, rather than to charlatans. At the user-interface design level, ensure that users can understand the information they need to assess the risks they are taking in relying on any given piece of information, and decouple the system to eliminate their incentives to take risks, for example with memory protection and petnames. We know a lot more about how to achieve these things than we did 45 years ago, and in some ways we have enormously more resources. We have seL4, Bitcoin, ssh, Monte, elliptic-curve cryptography, BLAKE3, NaCl, LUKS, decades of SOUPS proceedings, RISC-V, yosys, and 16-MIPS microcontrollers§ that cost 3¢.
But that future is not merely "not widely distributed"—it has become inaccessible except in isolated cases like Trezor, as economic incentives have driven our hardware and software down a path of boundlessly ballooning complexity and diminishing alternatives, while proprietary software licensing eliminates any possibility of assessing and controlling the risks. Meanwhile, the shallow pop culture of computing reduced users from creators to mere customers, and then "eyeballs", while conflating hacking—the only way out of this mess—with computer invasion.
So, I fully expect that if I live long enough to need a pacemaker, I won't be allowed to secure it against ransomware, which will be rampant at that point.
It doesn't have to be this way. This can all be made better.
Ready? Begin.
______
† In fact, shortly before that, on most PCs you could recover from any kind of system corruption just by taking the floppy disk out, resetting the system, and inserting a new, uncorrupted floppy disk. Better hope that one's not stoned too...
‡ You might argue that the possibility that there's an undetected security bug in seL4 means that complete computer security, even against carefully crafted data sent over the internet rather than some dude running off with your cellphone while it's unlocked, is still impossible. But in fact I think there's a very real difference in kind between the possibility that I might currently have presymptomatic covid, and the certainty that I have a small amount of covid. Systems like qmail are analogous to the first case, because they might be secure or might contain an undiagnosed flaw; systems like Linux and Chrome are analogous to the second case, because they are certain to contain a small but fatal fraction of flaws, which are inexorably multiplying.
§ Unfortunately the whole line of Padauk microcontrollers is out of stock this week at LCSC.