Hacker News Reader

Make it absolutely, unavoidably painful to be caught with private data, to the tune of increasing percentages of the total value of a company, like 3% per day of offenses. Make it retroactive - get caught running illegal data harvesting for a month and you're liquidated. Make 80-90% paid directly to the victims and 10% to a whistle-blower if there is one, and the enforcement agency. Require jail time for egregious offenses, and include any managing bodies and boards. Make liability apply to individuals regardless of any other consideration so the owners can't just bankrupt and run.

Make it "fuck around and find out" painful.

We'd throw the book at people creeping into our homes and vehicles and places of work to gather data on us in person. The fact that they're using barely perceived surveillance apparatus shouldn't factor into how we treat it.

Except I think there's a pretty good track record of companies knowing exactly how to get around these with shells, sub-contracts to overseas companies, too many hops and hand passes, loopholes, etc.

Or you get fraudulent 'whistle-blowers' inserted by the competition to reduce said competition.

It would be better for it to be painful for people to actually enter their data, so that they really must want to, further then have to manually accept for each individual party that will hold that data. But of course that is unpalatable to most any degree because convenience is king...

We keep wanting to fix the exit gate because that's 'easier' than the entry one.

> Recommendation algorithms hold far too much power and influence in our society to remain as secretive corporate IP.

I don't think that simply making the algorithm public would do much. Okay, now you know it's an ML algorithm applied to X data, what does that change? It's just as toxic as before. Maybe it should be required to show a _non_-recommendation feed as an option? Recommendation feeds create self-reinforcing filter bubbles by design.

Oh I don't mean to publicize just the algorithm.

You would need to publicize the weights and input data of the algorithm. You can't just say "see this whitepaper of the textbook for algorithm."

Obviously there's a lot that goes into this. But the flip-side is that the recommendation algorithms are so crucial we have to tackle this.

> You would need to publicize the weights and input data of the algorithm

Again, I don't think that really tells you anything, and I'm telling you now there is _zero_ chance that these companies open source their training data (that's their secret sauce). But even if you could, what can you possibly do with this information that would be useful? It's completely inscrutable for laypeople, which means it will only be analyzed by ML-competent engineers and mathematicians. And if it's a deep-learning algorithm, it's completely unexplainable and will yield practically no useful insights.

I don't think there is an "ideal" recommendation algorithm. There is no way to say "ah if we just tweak these weights, everything will be fixed!" Every single one has tradeoffs and will direct you towards filter bubbles. The problem is that many sites give you practically no other way to see content, which means you're constantly being manipulated by an algorithm that is itself learning from how you interact with what it proposed to you. That feedback loop is toxic, and steals user agency.

The problem with publicly disclosing a recommendation algorithm is that it makes it easier to game. But I guess everyone would be able to game it equally.

"Any other disclosures of Personal Health Information require the covered entity to obtain written authorization from the individual for the disclosure" -HIPAA

Wouldn't companies just update their terms of service to say "you agree to let us buy and sell your data?" Not allowing that would go up against freedom of contract.

What even is "freedom of contract"? Do you need to allow and legally enforce the ability of website owners to bury "by browsing this website you forfeit your eternal soul and also your firstborn" in a ToS? It doesn't seem like the kind of core human right that you need to allow a free society.

In my state there's limited freedom of contract in, for example, payday loans:

https://www.michigan.gov/ag/0,4534,7-359-81903_20942-171016-...

Companies do not have the freedom to offer a $100 payday loan with a service fee of $16. The state says that 400% APR is usurious; by law it's restricted to $15. There's no doubt that without this law, unscrupulous and predatory companies would try to craft a contract which demanded more than that, and we've decided that our society is better off without that unlimited greed.

> What even is "freedom of contract"?

When people organize and collectively set limits to their exploitation, in the form of legislation (child labor laws, workplace safety rules, ability to join unions and discuss salary, limits on non-compete and similar clauses, etc.), corporations will try to frame this as illegitimate, infringing on their ideal society where the only law is contract law, and they're the ones writing the contracts.

Because organization and collective action on the basis of democracy is the counterweight to the organization based on capital concentration, the latter tries to cast the former as illegitimate.

The linked site is European, where data transfers for non-essential purposes are already prohibited.

HIPAA is a good chunk of the reason medical software and practices are so dysfunctional. Medical practices are some of the last places that use faxes because of it.

It used to be the patient's charts were hanging at the foot of the bed. This made it easy for the doctor doing the rounds to get accurate information and for any observers, such as family and friends, to verify that the doctor was actually aware of the details of the patient's condition. Nowadays computer charting is so unusable that patients with drugs allergies are surprised if their physician doesn't try to prescribe a drug that will kill them.

Is it more of a problem of regulation not keeping up with modern developments? Or draconian requirements that got implemented once several decades ago to spec and now no one wants to poke it with a 10ft pole? I guess both apply to most heavily regulated industries that have any level of technology involved.

It’s neither, just old technology that worked and a system of incompatible electronic health record software vendors which are extremely slowly developing interoperability (there’s not a lot of motivation to make this happen)

Faxes are used because they are __exempt__ (not secure) and really easy since the UI is simple.

Contrast this with any method at all of trying to shove a document over the Internet, which will invariably result in something like a scanned attachment email of a PDF that bloats the mail queue and ends up in legal archives and inboxes forever causing problems.

Yes, we've all dreamed of better ways, but that's just like the XKCD standards comic ( https://xkcd.com/927/ ) and ultimately all solutions will require user training and effort.

Every other industry, even old school ones, uses email and digital filing just fine. It's only hard for medical organizations because of HIPAA.

They likely do use them, and get work done, and from a productivity standpoint might be faster without the red tape.

Discussions like this make me wonder if as a society we'd be better off not trying to perform the impossible: have OpSec that is flawless and never leaks data that might be misused; but to not allow the misuse to happen.

Medical information misuse touches at least the topics of discrimination, selecting victims of misinformation campaigns, exposing personal issues and preferences (E.G. did someone get an abortion; is a topic ripe for abuse, as are medical histories related to other issues.)

Living in a glass house is far from ideal, but in a world where individual rights are better defined and protected, data security could be more standard.

"But at the end of the day, any company that has your name/address/credit card number because you bought something from them is going to know. A lot. It's unlikely that any data inside said company isn't going to be de facto visible to them for whatever purposes that company so desires."

Hate to single out any one comment, but this "point" gets made quite often on HN. Unfortunately there is always something missing from the ensuing discussion. The customer did not buy anything from a "tech" company. Yet the people who think the "tech" company intermediary is a legit business model want us to believe they have the same right to name/address/credit card number. That does not make any sense.

If a merchant where a customer bought a product has her personal information, let us assume that is acceptable. (Without "tech" companies, only 20 years ago, this was never a major prvacy concern.) A merchant could transfer the information to another party, although she may allow the customer to decide whether that is permitted.

If merchant A does transfer the information to merchant B, its is likely that merchant B believes it shares customers with merchant A, i.e., customers of merchant A are likely to be interested in products from merchant B. We could argue that this is a major privacy concern, however besides the "junk mail" aspect of it, we should be able to agree that relatively few consumers were ever bothered by this sharing.

If the merchant transfers personal information against the wishes of the customer and this annoys the customer, then the merchant could risk losing the customer's future business. IOW, there were and still are some explicit and/or implicit limits on whether and where a merchant will transfer personal information about its customers.

However, the "tech" company intermediary is different from a merchant and is subject to no such limits. The customer has not purchased any product from the "tech" company. The "tech" company's customer is not the person purchasing a product from a merchant, it is the merchant, the merchant's competitors, any other merchant on planet Earth, anyone who is interested in influencing a computer user.

The data collection by the "tech" company intermediary is a new privacy problem. It is not the same age-old problem as merchants sharing customer lists. This is a new, additional privacy issue, with far more invasive potential. The structure, size and scope of the problem is entirely different.

Suffice it to say, there does not seem to be any logical connection between the age-old fact that a customer's personal information may be held by a merchant and the common HN comment thread suggestion that this somehow obviates any issue a customer might take to have her personal information held by a "tech" company. The later is a novel privacy threat with much greater invasive potential. Certainly, the fact that no one today has "perfect" privacy (they never did) is not a justification for sweeping, unbounded personal data collection by "tech" companies.

A merchant needs certain personal information to complete a transaction and deliver a product to a customer. The "tech" company wants the personal information to sell ads or take a percentage of sales based on use of its computers. These are very different scenarios and do not amount to equivalent problems.

The truism that no one can ever have "perfect" privacy does not logically support a conclusion that therefore unlimited "tech" company data collection is acceptable. It feels very much like people who have bet their career on the continued legality and effectiveness of the "tech" company business model want privacy-conscious commenters and readers to just "give up". Credit card users cannot have "perfect" privacy therefore they should be content with no privacy at all on the internet (the type of world where where "tech" companies can stay alive). That does not make sense.