To include the liability of the attackers, which I think will ultimately be more scalable and effective than punishing the victims. Not saying there aren't incentives for the victims to "do better", but I think that will only get us so far.
History paints a picture of societies evaluating the effectiveness of better walls vs. owning the landscape, and deciding on the latter as being a more beneficial approach. It's how we get the saying that "Rome conquered the world in self-defense". I would bet that's where this ends up after enough material losses.
Does this not also just kill tech? CVEs pop up decades after products have died. Now every tech product is just one unsupported CVE away from losing _all_ lifetime revenue. I just can't see how anyone would ever invest again...
edit: to clarify further the fact that any CVE triggers this, no matter how small, seems egregious to me. The idea of there being no lifetime on the liability seems wild given how CVEs are often the result of other developers breaking ABIs. Imagine a profitable product that was last sold 10 years ago having it's full lifetime revenue refunded because of some change in glibc.
Are they even covered within the warranty period? I never tried it, but I think I'd have an interesting conversation if I went to a shop and told them I want to return a product because while it works flawlessly it's got a vulnerability. The standard procedure is usually getting a replacement, but this isn't possible here as the whole product range is affected.
I had played around with the idea of requiring support for 3/5/7/10/whatever years after the cessation of sale, kind of like how car manufacturers are required to offer parts support for 10 years after sale, but I can see that causing enough overhead that many tech devices simply would never get made.
I think you underestimate the (a) greed and (b) capabilities of the people involved.
- How about 5 years minimum for hardware? And as much as the vendor wants to promise.
- How about that requiring that vendors at least allow for customers to pay for extended support for another 5 years by paying 20% of the initial price per year.
It is just ridiculous that currently many devices are insecure 3 months out of the gate.
We already know its neigh impossible to ensure software is bug and vuln free. We have to reckon with the fact that secure software is possible, but extremely hard and impractical to achieve with any regularity.
Because of well known computer science problems such as the Halting Problem[0] and more that I am totally ignorant of, the only alternative is extremely thorough verification like happens in planes[1]. The rub is that unlike aerospace and other critical controls, we cannot define the software security problem as closely as is possible in those other critical systems. Doing so will undo the benefits of our general technology solutions.
The reason I claim it is nonetheless not a software problem is: In case after case of these troubling examples of security failures, there are only a few organizational commonalities that link them all together. Time after time, there are shocking misconfigurations, corners cut, best practices eschewed, warnings ignored, processes disregarded or absent and swept under the rug.
C-Suite execs, as has been well noted, opt for the checkbox-style silver-bullet-whiz-bang, because they can understand the value proposition of one-problem-one-fix and it is easy to communicate to stakeholders. They are totally ignorant to and uninterested in the details of actually providing a quality product. Their product is often not what they are selling; They are getting their bonus, and they are gonna exercise those stock options.
The only way to make a dent in this security problem is to require adequate processes for infrastructure the way we require adequate lighting, or fire protection devices, or plumbing. Until it is a business requirement to provide appropriately managed processes, infrastructure, testing, and finally development practices to go along with it all, there is no hope.
Will your school or hospital be shut down next? Will it be your bank that is defrauded when your cardholder details are skimmed at the local gas station or stolen from a multinational retailer? Will it be your government's top secret clearance database, your municipal water, your child's photos from their phone or social media?
Our only real hope is that the next victims of this heinous, Steinbeckian, tragedy are the Congress, Senate, and whoever else it takes to get a fucking grip.
[0] - Computerphile https://www.youtube.com/watch?v=macM_MtS_w4
[1] - Examples of aerospace validtion https://scholarworks.sjsu.edu/cgi/viewcontent.cgi?referer=ht...
The only way to get secure software is to increase liability of parties involved.
My suggestion: Start with the ability of any customer to return any purchase (hardware/software) which contains software with a disclosed but unfixed CVE after 90 days without a patch. If this doesn't get rid of the Internet of Shit, I don't know what will.
Next, set a minimum damage rate of 100 USD per user for each data-breach that involves personal information and 1000 USD for any special kinds of personal information (credit cards, etc) and 10000 USD for any protected health information breach.