I wonder if it makes sense for ad platforms to enable a "clear history" option just like our browsers do? That way you can say very clearly "the current model is wrong, please restart and try building a new one for me". That might also help this particular case: rebuild the model from scratch including only content going forward rather than backward. Should eliminate the baby ads effectively, right?
Google at least let's you specifically go into ad settings and check/uncheck specific categories of things that you get ads about (that's an oversimplification of the process behind it).
I'm not sure how but one time I started getting a few ads in Spanish about toothpaste. I had also just found out about the ad settings page, and after taking the minute to remove the handful of wrong interests, the incorrectly targeted ads stopped.
Edit: Facebook allows it as well, here are the links for anyone that wants to do it now.
I mean right at the top of the page is a single switch that lets you turn off all targeted advertising if that's what you want.
But personally I like targeted advertising. Ads let me use services for no monetary payment while still providing income for the content creators, and I like seeing ads that are targeted towards my interests more than ads that aren't. And while it's not ideal that you might have to go in and change what your preferences are in cases where they get it wrong, I much prefer it over the alternative of all ads being "wrong" for me.
That'd work in this case, but would destroy lots of value when people like me would go in every few months and reset all my ad profiles. Then again, the total percentage of people who would care enough to go in and do that is probably pretty small.
I'm going to start sounding like I work for them, but Google already allows you to disable targeted advertising entirely, and Facebook at least let's you disable targeting for the vast majority of things (the major exceptions being age, gender, location, and the content of the page if I recall correctly).
I would immediately try to figure out a way to automatically clearing my advertising history as often as possible. So would other people; at least one of us would build an easy-to-use tool for other people to do the same.
Interesting idea, but I'm bothered by the idea that social media platforms would be effectively monetizing grief in this solution.
This sort of thing, to me, points at the deficiencies of appealing to purely statistical patterns without any guard rails placed on top. After all, statistically, the most likely relevant categories after "pre-natal" are going to be surrounding children who _were_ born safely. So if you appeal just to data alone, you won't likely solve this problem because it is relatively rare.
The biggest failure, to me, is the response of the system when the user goes out of her way to say that the "pre-natal" ads are not relevant. Statistically it assumes successful birth, but that doesn't reflect her actual intent. A simple dialog tree would maybe suffice: "not relevant" -> "this specific thing is irrelevant" vs "suppress this and all related content". One signal says maybe don't show that particular brand again, and the other says to pivot the relevance model significantly away from that topical section of the ad space entirely.
They already monetize joy. Is monetizing grief worse? Why? What about monetizing neither? What's the signal to pivot the relevance model significantly away from a world full of robots constantly self-optimizing to exploit emotions they cannot understand or share?
(Or if we can't manage that, maybe at least we could somewhat curtail the misbehavior of the ML systems they foist on the rest of us.)
The correct response is not to try to advertise targeting the traumatic event, but to pivot and avoid it and related subjects entirely. I doubt people will complain if ads about babies and parenting were to disappear after a still birth or miscarriage.
If this had occurred in the US (and the mentioned confidential conversation did in fact allege sexual misconduct), DJB would be in trouble:
> Within the University of Illinois System, ALL employees, unless specifically exempted, are “Responsible Employees” with the responsibility and authority to report sexual misconduct to their university's Title IX Coordinator. The only employees who are exempt from this reporting requirement are professional or pastoral counselors who provide work-related mental-health counseling, campus advocates who provide confidential victim assistance, and employees who are otherwise prohibited by law from disclosing information received in the course of providing professional care and treatment. Student and graduate employees are handled differently at each university. Please reference the Responsible Employee Resource Page under the "Portfolio" and Resources tabs. Please remember that all references to Responsible Employees are references to YOU and apply to you in your capacity as a university employee.
To me, this would mean that he is a mandatory reporter, and I am unaware of any scenario where you are freed from that obligation because it was a "confidential conversation".
The weird part comes in when you realize that (a) this is happening outside of the US, but (b) DJB likely has NSF grants, which require adherence to Title IX (this is what the author is referring to when he brings up Title IX training). But how does one enforce Title IX outside of the country in which it was passed?
While its different in each country, I know that teachers and counselors can be required by law to personally report such crimes to the police. If I remember right, this is true for Sweden, which would in this case result in a police report and then no further actions or communications from the university (in order to allow the police to do a proper investigation without interference). If it is a student that is accused, then the university might not even be allowed to suspend the student, through the police can of course put the accused in holding if the police suspect a continuation of crimes or interference of the investigation.
Compared to the US system, I actually prefer this way since it puts the whole process into its proper place as soon as possible, and puts a form of common-sense approach when a university employee hear or witness a crime.
I know that education institutes in Sweden sometimes move students if they consider that the person is continuing disruption the education, through as with all of this, there need to be documentation that they tried multiple methods to correct the situation and still failed. Moving students is seen as a last-attempt.
In the case of Jacob, we don't see any of those actions. No police report or investigation. No claim that he is continuing acting disruptive to the university, nor that they have tried and failed to correct that behavior. Basically no events or documented actions of jacob after the point he left the tor project.
Not only are you correct, you allowed me to find an equivalent feature in CommonMark! If there is an empty line between some content and its wrapping HTML tags then that content will be interpreted as markdown. So
I absolutely love this project. I've been making bindings for a C++ library using it recently and have yet to encounter a situation that was too tricky to bind properly. I'm loving the ability to take an abstract class from the C++ side of things and derive and override virtual functions in Python. Super excited about how much easier that makes it to do exploratory programming to get something that woks, and then worrying about writing it in C++ when I need that level of performance.
The lack of a dependency on Boost only sweetens the deal for me. I love what Boost has set out to do, but it's just such a huge dependency with a ve4y strange (for me) build system.
He starts the talk by assuming monosyllabic words as his primitives and builds up the words he needs to use to give the talk by providing definitions for them first.
