Hacker News Reader

Comment by jakelarkin | Hacker News Reader

If 23andMe is eventually acquired/reaped, what is the worst thing the new owner can do with the data?

to ensure long-term stewardship of the data, a company that holds genetic info should be a non-profit owned by the DNA providers, along the lines of a mutual insurance company.

View on HN

Sell it to insurance companies, schools, employers, or the government who will grant and deny access and determine cost to you and your descendants inside or outside the law.

Business and politics is becoming more and more about using analytics to manipulate you and your fellow citizens. Your genetic makeup along with other data about you could be combined into optimal targeted campaigns tailored to just you to get you to behave in line with somebody else's goals. Knowledge about you correlates with ability to control you.

There's a law against this. The anti genetic discrimination law passed by Bush 2.

Sure, and employers aren't legally allowed to discriminate based on criminal record unrelated to the job, but that has never stopped them.

I think they are allowed to discriminate based on felony convictions etc... at least in the states.

A blanket policy of excluding all felons without a legitimate business justification may violate Title VII of The Civil Rights Act of 1964 due to disparate impact.

https://en.wikipedia.org/wiki/Employment_discrimination_agai...

Can you get around the law by phrasing a question to 23andMe. A health insurance company could call them up and ask if so and so have genes that has an increased risk of contacting x disease ? Technically they are not selling the information so they are not breaking any laws but it's still valuable to the insurance company.

Laws come and go. Your data stays.

GINA law already prevents insurance companies from acting on this data.

It seems unlikely that this sort of genetic data is really a risk. To test that, I had my genome sequenced and published it with a reasonable license. Anybody is free to download my data.

How would you know if that data was used in a decision?

Great question. Ultimately, I can't know for certain, but I can make some reasonable guesses.

I think GINA only prevents health insurance. Life insurance would probably be interested in genetic indicators of early mortality.

I've put my raw data on GitHub in case anyone wants to see what it looks like or make a PR. This is the V3 chip (V5 is the current), but the formats should be the same.

https://github.com/blopker/DNA

so, uh, if I send you a pull request, do you CRISPR it into your genome?

I'm just waiting on Travis CI integration.

Sell it to an insurance company, who in the future may be able to discriminate against you due to your genetic characteristics.

Sharing fundamental characteristics of yourself with random cloud service providers is probably one of the dumbest things I can think of.

If insurance companies are able to take advantage of genetic information, won't they just pay people to gather it?

Like, for $500 someone could follow you around for several days waiting for the 1 lapse where you leak a little DNA into the public environment.

Or you know, they will just discriminate against people that refuse to share their DNA. The narrow wedge where they are forbidden from requiring DNA but free to discriminate based upon it is still a pretty fucked up world.

It's not really plausible that they would hire people to secretly gather it from you. Maybe they just make it a requirement for getting the insurance at all, that seems cleaner and more scalable.

> who in the future may be able to discriminate against you due to your genetic characteristics.

Look, I know it's fashionable to disavow discrimination in all its forms and for all its reasons, but businesses built on evaluating risk, like insurance companies, NEED to discriminate.

It's impossible to price and discount risk without access to accurate information.

I'd be FAR FAR more worried about governments misusing my genetic data than insurance companies using it to determine an accurate risk profile.

The insurance company can price risk at the population level, there is no need for it to be able to price it individually.

This can be "less fair" to the lower risk members of the population, but the other side of that is whether the insurance company is actually capable of doing the individual pricing in a fair way, or if it is just being presumptuous.

The problem is adverse selection. If insurance buyers have information that insurance sellers aren't allowed to have or act on, high risk people will buy, while low-risk people won't. As losses go up, so will the price and when the price goes so high that only the highest risk people will buy, the market will collapse.

Giving insurance companies more information simply reduces the scope of what they will insure. Highly predictable risk (I will be dead before I turn 150) is not really insurable. Unpredictable risk (I will be struck by lightning) is. Genetic screening just moves some unpredictables to the predictable column.

Is it fair? No clue. If it turns out that there's a genetic predisposition to being struck by lightning that modern genetic testing uncovers, then all those "struck by lightning" policies will die - they were subsidizing the highly conductive to the detriment of the non-conductive. Hope you're not highly conductive!

TIL using statistical data to make actuarial data more accurate (e.g., insurance) is discrimination.

Yup.

"Oh, you're of African background and have sickle-cell anemia. Your sore throat isn't covered as your spleen dysfunction is a preexisting condition."

I dunno, I think it's pretty trivial to back into life expectancy and genetic risk factors from publicly available ancestry information combined with death certificates. We have ~200 years of pretty good records here in the US and many more if you go into Europe's records. From that data you can get a really good idea of longevity for a given person.

Ancestry.com even has a public tool for doing this at a high level: https://www.ancestry.com/name-origin?surname=trujillo

Seems the cats been out of the bag for a long time.

Maybe you shouldn't throw around 'dumb' quite so liberally.

They're already discriminating. It's a large part of their job. I much rather they do it accurately than inaccurately.

I read this last year and it is scary

https://www.washingtonpost.com/news/worldviews/wp/2016/09/14...

It is never a good idea for a single (for profit) entity to have that much data on anyone.

I am sure Facebook would love to connect that data to theirs. Now you have full surveillance: behavior, face recognition and genetics.

I think I will request my data be deleted. I have the raw data archived anyway, and run it through Promethease.

Good luck. There will always be a copy somewhere. That's why I would never use a service like this.

The benefits trump the risk for me.

I hope you also considered the risk for / asked all the people related to you before handing over significant parts of their DNA as well.

Not the worst thing, but they could make it possible for distant family members to locate you based on DNA similarity. This isn't always desirable.

They could clone customers... I guess it could be a good thing in some cases, if they don't modify the data.

Sell to law enforcement to find relatives of suspects.

Reply on news.ycombinator.com