I agree everything you've stated would be desirable, and if there was a strong culture and policy of supporting bounty programs from the CEO on down, this could potentially be achievable. However:

dupes - On the bounty side dupes are extremely common and buddies telling buddies about their finds is going to drive fraud up quite a bit. In my triage work I saw very clear attempts at this regularly.

wontfix - This one is largely due to the fact that many bug bounties don't have authority over or even shared reporting structure with the product teams. There's probably room for a consolation prize as long as the bug is in scope but that's about it. The fact that the bug goes away later could be a fix or could just be part of a new release. This should be extremely rare though and is worth following up with the program (again, as long as its in scope).

sharing issues - This is going to struggle mightily with legal without good contracts and NDAs for each researcher.

status updates - Agree its frustrating but is challenged by the product team/bug bounty alignment noted above. Most bounty programs don't get info from devs either and if product teams don't listen about fixing bugs the likelihood that they are going to regularly report on fixes is almost nil.

shared comp - Unless I'm missing your point you can self-organize outside of the bounty program (and many do) for this.

Phrased differently, internal organizational challenges should never be a valid reason why a bug is disqualified. It’s completely irrelevant from an outsider’s perspective.

The challenge, however, is constructing a sustainable model to incentivize product teams to reciprocate this closer working relationship. I would say all but the smallest of companies running bug bounties also have an internal security function that is already doing reporting on vulnerabilities, time to fix, etc. etc. So whatever internal 'reputation' there might be across product teams is well established (and in my experience the culture around bug fixing across product orgs is consistent from both internally discovered and externally reported bugs)

Another thing that happens is that there is typically a backlog on bugfixes, so if a researcher reports a new bug that's Medium priority, it's not going to get prioritized against a backlog of Criticals or Highs. Most of the lack of feedback from devs is simply the fact that there's been no action. Once the bug is front and center, the fixes are extremely simple and done within a few days and rolled out.

* properly qualifying a bug bounty submission

* actually fixing the thing

From what I understood from your original comment, was that sometimes things get qualified as “wontfix” because of internal struggles. I understand that these struggles exist, as with any large org with conflicting priorities, especially when the issues are not 0day severity.

I think, however, from an outsider’s perspective, acknowledging it’s a valid bug, but also saying “this will not get fixed any time soon” is infinitely better than just marking it as wontfix.

As such, I’d argue that the minimum effort the product teams would have to put into the bug bounty program is properly qualifying the submissions. If that, for some reason, is too much to ask, I submit it’s a higher level, leadership problem that they want to have a bug bounty program without the necessary resources to qualify the bugs.

One pretty interesting example is GitLab's bounty program because you can see both sides, both the reported issues and how they are tracked on the engineering side:

https://hackerone.com/gitlab/hacktivity?type=team

https://gitlab.com/gitlab-org/gitlab/-/issues?label_name=Hac...

There's very good organizational alignment here from what I can see, and overall the program seems to be quite adaptive and well supported (e.g. the comment about eliminating the 50% bounty here https://hackerone.com/reports/1154542#activity-11394543, 'unduping' a report here https://hackerone.com/reports/402658 ). It's about as best-case as you can get for a bug bounty...relatively small, single-product company that exposed to the sunshine effect of a public issue tracker.

That said, if you peruse every one of those issues you're going to find instances where they drop the ball, take too long, make mistakes, etc. etc.

So I would just say that if you ask anyone that has run a bug bounty, they will likely say they were surprised how difficult it was. You'll also generally find that they are advocates for the researcher community and want to run the best program they can. Everything else is largely a function of the dark art of business prioritization and the quality of the experience will be a function of how well they and their team is able to navigate it.

Thanks a lot for your insights, they were very valuable.

Completely off-topic comment follows, but I find lingustics interesting, am a native (Br) English speaker, and think it's worthwhile to reassure someone they're foreign language skills are fine.

Yes, it's good. (It's extremely valid :wink:). It has a slightly comical flair to it - not sarcastic, just certainly not formal. You could push it further to 'I extremely agree', which is more debatable but as long as they realised you were 'being ironic' nobody would say 'hey that's extremely invalid English' or anything.

Only one thing wrong: characterise (oxendict be damned!) and incentivise are spelt (not spelled) with an 's'. :British-trollface:

"extremely valid" is slightly off.

"extremely agree" is comical but definitely not standard US English.

It's just an unusual pairing, and 'extremely' is so much of a 'stronger' word that it's got that humourous edge.

Not in formal writing, sure, but it wouldn't make me think the speaker's struggling with English, if anything the opposite - a good command of it and able to twist it in fun ways.

wouldn't looking at the logs of when the reports were taken pretty much clearly show the first person to make the report? how is this a thing that gets confused to be an issue?

Duplicate reports are likely the most frustrating thing most security researchers will encounter. They put in a ton of work into finding the bug, developing a proof of concept and writing up a detailed report with the hope and expectation of being awarded for their effort. So when the triage team comes back and say it’s a duplicate and there will be no award, it's incredibly maddening. If someone needs to vent in my direction because of that, I can totally understand. The *problem* was trying to remain diplomatic with people who would sit there and repeatedly claim they were actually first, or that it wasn't a dupe, or that all of this was far too sophisticated for me to understand. Then, finding no sympathy from me, go to Twitter to wail and moan and bash the program with impunity because they know that the org won't respond in kind.

Maybe they should be allowed to. If it is someone that is seriously being that petulant about something, then the Org could post dates of correspondence, and even quote petulant tempertantrum once it escalates beyond civility. Once the user name gets out there, other bounty programs could just put a blanket ignor and drive the petulant person into obscurity. But of course it is the today&now, and nobody actually believes facts anymore.

But the transparency you're asking for is difficult to actually provide. Meanwhile, for a vendor at Google's scale, there is essentially zero upside to screwing over bounty hunters. At any realistic valuation for a vulnerability, these are rounding error sums to the business. In fact, the exact opposite incentive exists: these bounty programs are deemed to be performing well when they pay out more money, not less. The people managing these bounties aren't paying with their own money. They'd rather make you happy and encourage you to submit more stuff.

The two phenomena you describe here are real and common. But it's just simply the case that vendors are generally working through backlogs of issues, triaged by severity. If you're told your finding is a dupe of a private issue, it is overwhelmingly likely that it is. If you pay for every independent discovery of an issue, people game that; worse than the dead weight loss of the bogus bounties, you set up crazy incentives on your dev team to fix marginal issues because they're being gamed, rather than triaging according to real severity.

And, bounty hunters do turn up real bugs that aren't real security issues, but are still bugs. If those bugs are easy to fix, they're going to get fixed! You have the same weird gaming and precedent issues if you start paying out non-exploitable bugfix findings; you encourage people to find and report non-exploitable bugs, and you screw up the team incentives on what to fix.

I don't expect people to like any of this logic, because it boils down to "you should just trust the Google VRP people". But: you should. This isn't worth the cortisol. If it's driving you up a wall, maybe don't participate? There are other ways to market security bug finding skills. :)

If you've never worked triage on a bounty before, my guess is that you can't really imagine how terrible the median interaction is. Maybe the next evolution of these programs will be long-term contract relationships with trusted, successful vuln hunters that address some of these concerns by separating out the people who are good at this stuff from the median submitter (the median submitter invests 2 weeks trying to litigate whether copying a cookie out of the Chrome inspector and pasting it into curl constitutes an account takeover vulnerability).

For what it's worth, when I was setting up the culture and values of Google's first bug bounty programs, I hammered "be magnanimous" into the reward committees. i.e. look for reasons to reward more, not less. Find the value in the information provided, even if the person is being a jerk. etc. I don't think this culture has changed. There are teams of people rooting for incoming reports to succeed, and they get excitement and joy from issuing large bounties (because this means Google security is getting stronger).

I deleted three sentences about this very topic in my earlier comment because it turned into an ugly rant, lol.

>Maybe the next evolution of these programs will be long-term contract relationships with trusted, successful vuln hunters

I'm actually somewhat surprised that bounty programs missed the whole 'gig workers are employees' issue.

The refusal to pay while fixing the bug in 48 hours shows a power inbalance. I fear this is where we are headed in the future as AI pumps out most code all of the remaining work comes in the form of contests where thousands spends 80 hours a week and the company (in the best case) will award one person with a cash prize small enough to not hit minimum wage.

Here's the "20 factor test" (condensed from Oregon's condensed version) the IRS uses for employee classification:

* Does the company direct where, when, and how the work is done? (You can hunt for bugs from any country in the world, whenever you want, using whichever tools you like.)

* Does the company require company-provided training to provide the service? (The first contact Google has from a bounty hunter can be the bug they submit, and Google will pay).

* Is the service directly integrated into the business the company provides? (No: Google doesn't provide bug bounty services to other companies, and is not in fact a bug bounty with a small search engine attached.)

* Does the company insist on a particular named person to do the work? (No; in fact, until they're actually paid, many bounty hunters are psueodonymous.)

* Does the company control who the provider hires to assist in the work? (Nope, you can hire your own team of people to assist with your bountying, and Google does not care.)

* Does the company maintain a continuous relationship with the service provider? (Nope, you can submit a bug in June and then come back next April with another one, or come back never).

* Does the company determine the schedule on which the work is done, or is the schedule flexible? (Google does not even slightly care what schedule you find bugs on, or even the timeline you use to report bugs on once you find them.)

* Does the company demand full-time hours? (Google doesn't set any hours for bounty hunters whatsoever.)

* Does the company demand that the service be provided on-site? (Bounty hunters work exclusively from their own facilities.)

* Does the company demand that the work be performed in a specific order or sequence? (No, you can look for any combination of bugs in any set of services using whichever set of tests you want to use, be they fuzzing or mitmproxying or scanning, in any order you like.)

* Does the company require regular periodic written or oral reports? (Indeed they do not.)

* Does the company pay hourly, weekly, or monthly, rather than by the project? (Bounties are paid by the bug.)

* Does the company defray expenses for the work? (Nope.)

* Does the company provide the tools and materials needed for the work? (Nope, bounty hunters BYO laptops and Burp Suites.)

* Does the company provide facilities, office space, etc for the work? (Google doesn't provide even so much as a Slack account for bug hunters.)

* Are the earnings from the work predetermined, or can the workers realize greater or lesser profits from the amount of work they put on? (Regardless of the hours committed to bug-hunting, bug hunters earn revenue based on the value of the assets they generate; if you come up with a technique to find dozens of XSS vulnerabilities in a couple hours, you'll make a zillion dollars per hour.)

* Is the relationship exclusive? (Of course not, bounty hunters work with multiple companies as a rule.)

* Can the worker make similar services available to the public? (Not only can they, but independent consulting is the norm for this kind of work.)

* (This is 2 bullets, condensed) Can either party terminate the relationship independently? (Yep.)

Bug bounty participants are not, as a general rule, classifiable as employees. People make noise about it because there's drama about Uber --- where the company exerts substantial control over how employees do the work, when and where they do it, where they check in with their cars, &c, and do so while providing the core service the company offers. Someone on a message board is going to try to suggest that Google is in fact a bug bounty program with a small search engine attached to it, but I mean, come on.

Well, I felt pretty screwed over after Google decided not to reward me for discovering CVE-2021-30560.

It's more helpful to analyze the situation from a game-theoretic standpoint. Bug bounty programs are a typical example of a cooperative game. If someone reports a security issue, they start the game by cooperating. Google can then choose to cooperate by paying a bounty or they can defect by refusing a reward. Like in the classical prisoner's dilemma, both parties can typically maximize their rewards by choosing to cooperate.

My case is a bit special. While I don't know the exact reasons why Google chose not to reward me, it's possible that they thought I would continue to cooperate even after they defected. But like most rational players, I chose to stop cooperating, which ultimately leads to the worst possible outcome.

Same goes for the dialogs with unhappy hunters who like 'proof' for the arguments that a bug / vulnerability is not there.

There is literally no financial incentive for me at all to not reward, au contraire actually.

This kind of issue is rampant and opaque among bug bounty programs.

IMO if a company says a bug is a wontfix, that's an immediate moral justification for public disclosure.

If they say it's a known issue, they don't give a timeline of when it will be fixed, and it's still not fixed in a week, that's also moral justification for public disclosure. If multiple people have reported the same vulnerability, there is a high likelihood that the bug is known by even more people. Bugs that are not hastily addressed are wasting participants in your bug bounty's time.

Companies with bug bounty programs need to treat security researchers with more respect, and when they don't the moral imperative shifts towards public disclosure so that others are warned that vulnerabilities exist and are not being addressed. Too many companies set up a bug bounty program as a box checking exercise and then have a lackadaisical attitude about addressing reports.

I found a pretty obvious XSS on Tesla's website. Submitted through bugcrowd and got no information besides "marked as duplicate". Publicly disclosed, bugcrowd temporarily suspended me for disclosing, but it was fixed within a week. Nothing lights a fire under people's asses like airing their dirty laundry. If they had told me "we are working on a fix and expect it to be live in 3 weeks" I would have respected that and held off on disclosure.

about having access to private bugs, we don't want to share vulnerabilities with others without the researcher's permission, but the original researcher can make bugs public on the new website, you can see some of them here https://bughunters.google.com/report/reports

Other people have covered most of the rest of your comment, but there are things to say about this one specifically. The reporting platforms support it. It's a very frequent request from researchers who file duplicate reports. It's rare for a company to do this, because (1) it doesn't do anything to help with the issue being reported; (2) it doesn't change how the report will be handled; and (3) researchers hate it. They don't hate it when they file a duplicate report and want to see what scooped them. But they hate it when their report gets shared with someone who filed a duplicate finding.

Sharing duplicate reports causes a lot more fights than it solves. And it tends to piss off the higher-productivity researchers in an attempt to soothe lower-productivity ones, who generally aren't soothed anyway.

The ticket I saw the most outrage on over this issue was one that I duped to another ticket with a higher number. (HackerOne assigns ticket numbers serially. Fun!) How could I justify calling this report a duplicate of one that, as anyone could see, was filed later?

Well, the second report began

> Hi, I reported this through email and was told that in order to claim a reward I should open a ticket in the HackerOne program...

(This situation is odd enough, and easy enough to explain without sharing private information, that I could explain it to the guy on the lower ticket number without causing problems (or sharing the other ticket directly. That's a big no-no.). I present it here as an example illustrating that, even if you think you have ironclad evidence that the program is out to screw you over, it probably isn't.)

I definitely saw companies doing things that were unfair considering the program as a whole. That generally happened as part of an effort to preserve a relationship with a researcher who frequently filed useful reports. I never saw a ticket duped inappropriately. If you want my advice on how to get the most money out of your bug reports, it's this:

- Don't antagonize the team handling you.

- As much as you can get away with, demonstrate what you can do with the issue you report. The program will say "we investigate the impact of every issue and pay out according to the highest-severity potential impact." They are sincere. But they don't have the motivation you do to find the highest-severity potential impact. Any impact you actually demonstrate will automatically be considered, because nobody had to notice it was possible.

- Sometimes an issue might ambiguously fall into a low-paying category, or maybe a high-paying category. Try to characterize it as belonging to the high-paying category.

- Decide whether you're a "deep" guy who finds complex issues on a handful of platforms or a "shallow" guy who finds the same set of related issues everywhere he can with minimal effort. Both approaches work. If you're a "shallow" guy, make sure that when you report an issue, it's really there, and don't argue too much over what you think it should be worth. If you're a "deep" guy, you have more scope to develop a personal relationship with the team who handles you.

- Watch for existing programs to add new platforms. This often happens when a company with an existing program makes a new acquisition. The new platform probably has a lot of low-hanging fruit.

- Clean up after yourself. The most outrage I've ever seen on the company side was someone who demonstrated full-on RCE on a company server. He installed a webshell. And his webshell was still up when he filed his report. Don't be that guy. You can be disqualified from a bug that would have paid tens of thousands of dollars.

The number of reports that eventually become fully public is nearly zero. Most don't become public at all, but of those that do, a lot of content is generally redacted.

All report hashes would have to go public as soon as the report is accepted. The hash input could go public once the bug goes public, so the duplicate reporters can then finally see proof that the bug had already been reported.

In what cases would companies be unable to publish generic descriptions after the bug is public? I'm not in the industry so I have no idea about this.

You're also still failing to account for the fact that reports rarely become public. I can refer you, again, to a random number just as easily as I can refer you to the calculated hash of my bespoke summary of an issue that was reported eight years ago.

There would be no benefit except for a tiny amount of goodwill, so it's almost certainly not worth it. This is simply a method to address the duplicates issue. Nothing else.

It makes sense if you're vetting people beforehand to make sure that giving them this level of access and communication is worth the effort, but that's what a job interview is.

As for getting a prize, it goes back to the dupe issue which is the source of a lot of abuse. There's no way to prove you also worked on it or if you just got the info from your friend and want to double your winnings.