> Expanding this business to the private sector might have positive effects. Corporations could buy the "zero days" in their own products in order to fix them.

I should point out that some companies are already getting vulnerabilities reported to them for free (in the best case, for the price of a symbolic bounty program) and fail to fix them. See http://reverse.put.as/2014/10/31/patching-what-apple-doesnt-... for one recent example that comes to mind. Giving them the opportunity to pay market prices for the vulnerabilities is not going to improve the situation.

Everything else remaining identical, the only thing the emergence of a free market for vulnerabilities will make efficient is the exploitation of victims that aren't informed of and do not have the capacity to fix them.

The only real fix to the current situation starts with legislation that aligns the interests of the companies that rely on vulnerable code and those of their users. Forbidding the trading of vulnerabilities won't help, and encouraging it won't help either. Recent legislation in the EU forcing companies to disclose security breaches they are aware of is a first, baby step in the right direction. There are many essays by Bruce Schneier on this topic, here is one picked at random: https://www.schneier.com/blog/archives/2004/11/computer_secu...

The fundamental problem with trying to retard vulnerability markets is that vulnerability research is expensive. There aren't that many people who can do it(†).

You could conceivably ban overt vulnerability sales. But that doesn't actually fix anything, because the firms that want to buy bugs now will simply retain researchers directly instead of licensing their work product.

You could even conceivably ban commercial vulnerability research. ACLU's vulnerability markets guy has come close to suggesting that in the past. But that also won't fix anything. Firms that are using vulnerabilities illegally would continue to (now illegally) fund vulnerability research --- an activity that is far harder to monitor than actual hacking, which is already hard to monitor. The net effect would be to clear the industry of everyone doing "benevolent" research --- no more Metasploit, but lots more VUPEN(††)-type stuff.

† Where "it" means the kind of vulnerability research that commands 5-6 figure premiums, virtually none of which is web app security work

†† Full disclosure: I know fuck-all about VUPEN and am just using the name as a shorthand for "the vulnerability boogeyman".

As another slightly less draconian approach than actually trying to ban vulnerability research, what about mandating a market where researchers have to sell to the developers and developers are obliged to buy from the researcher, with a middle-party setting the prices.

That might serve to provide incentives to the vendors who aren't already working actively on their security to do so, and also reduce (although obv. it wouldn't eliminate) the number of vulns going straight to offensive products.

So I have two options: I can sell work product in a market controlled by "the FCC of vulnerabilities" with capped upside, or I can work directly as a 1099 contractor at an enormous daily rate for firms that will pay to get vulnerabilities before vendors do.

Why would I take the market route?

In this case what are the firms who pay these rates doing with the information?

Defensive work (e.g. IPS vendors) well after they've got their early day protection they can just sell on to the vendor.

Offensive (e.g. "cyberweapons" ugh I had that term). well that's the point I'm making that whole industry is bad for defensive security as it involves keeping vulns secret as long as possible, so they can be used by them.

Gov's have to make a choice whether that's an industry they want to encourage, be neutral to, or discourage.

But given this line of thinking is one you'd disagree with, what option for addressing the problem do you prefer?

What difference does it make what they do with it? Stipulate for now that they use them to hack Russian and Chinese computers. That is, stipulate that there is a good public policy reason to regulate that kind of work. How would you accomplish that regulation? What, exactly, would you ban?

If you can't articulate a reasonable and effective regulation that would control vulnerability research, regulation will do more harm than good: it will wipe out beneficial research and drive talent towards malicious research.

It's not on me to come up with a way to "address" the "problem". Doing nothing seems like a more credible response than trying to outlaw specific kinds of computer programming.

Indeed, however in the same way as it's not up to you to "address" the "problem" nor is it up to me :)

Doing nothing would seem like a losing response given the current swing of events but I'll defer to your greater experience.

BTW I'd respond to the other threads but HN seems to object to deep threads.

On "flaws fixed rather than concealed" sure in a perfect world they do, but limited resources == prioritization and some systems and packages inevitably get left behind, insiders know which those are..

What do I, as a user and customer of software, care why companies harbor known security flaws (that, after all, is what you're talking about)? Hope is not a strategy. Those flaws are going to be discovered whether or not insiders leak them.

I'm not sure where I said hope was a strategy, I said insiders might have information that's of value to attackers (unless I got my threads mixed here)

The solution is obvious: all software development needs to be classified as a utility and regulated under Title II.

It's a systemic problem. One option is to decriminalize cracking, and then to associate a fine with creating vulnerabilities. It's straightforward to find the original author of some code. $500 maximum fine, or something on that level.

Changing the system (obviously) has a bunch of other obvious (and a bunch of unforeseeable) consequences. However, code would get a whole lot tighter.

It would also help if we got OSes that do not give applications blanket rights to read and write everywhere and use as much processing power as it likes. Abandoning the shared security model would help a lot.

I'm having a hard time coming up with an example of a critically important vulnerability that relied on permissions models. Arbitrary code execution is usually game-over no matter what privilege level you have.

The exception to this is sandboxing, which is effective (but unreliable) in limited, specific scenarios but not at all effective for the general problem of controlling real, full-featured user programs. Compare the Chrome content sandboxes to the Apple application sandbox.

> Corporations could buy the "zero days" in their own products in order to fix them.

The implicit threat that if you don't pay enough to secure exclusive access to the zero-day, it will be sold to others who may use it against you - that seems a lot like blackmail. It's different with bug bounty programs, which are not explicitly set up to compete with the price a vulnerability would get on the black market, but rather to incentivize civic-minded responsible disclosure.

Assuming the exploitation of a vulnerability is considered illegal, selling the vulnerability to someone who you have reason to suspect will exploit it should also be illegal, surely?

interesting idea, to an extent this already exists with bug bounty programmes that a number of companies have in place.

The risk is where either the company doesn't have a programme of that kind, or where the value of the bug is substantially higher to the offensive industry, than the product company...