I speak to a lot of enterprise buyers ands startups and it's definitely a major issue when it comes to buying products, especially from a startup.
There's a big difference between what a product does in theory vs in practice - and social proof is a big guiding factor for what it can do in practice.
If you want to use it for family tree research you basically have to go to the companies that have the largest databases, because that's where you'll find familial matches (e.g. 23andme and AncestryDNA).
23andme has a better track record of refusing governmental requests than AncestryDNA.
In terms of genetic information sharing, none of the providers shares non-anonymized data. 23andme allows an opt-out for sharing anonymized data.
You can also do the tests pseudo-anonymously, while they ask you to provide your real name, etc. you don't actually have to do so. You can purchase a test kit at a retail store for cash and register online with a fake name if you wish to do so.
I think you're confusing the common-usage meaning with the legal definition.
Generally in Europe the market share threshold for monopoly investigations is 25% - Apple's market-share for mobile app sales is well above that threshold.
> The Commission considers that low market shares are generally a good proxy for the absence of substantial market power. The Commission's experience suggests that dominance is not likely if the undertaking's market share is below 40 % in the relevant market. [1]
Historically the European Commission's antitrust cases against tech giants have largely been against companies with overwhelming market share:
European Commission vs Google: 90% market share in search [2]
European Commission vs Intel: 70% market share in x86 CPUs [3]
European Commission vs Qualcomm: 90% market share in LTE chipsets [4]
Apple's 25% is nowhere near the same level and I expect the Commission's investigation will conclude that they do not, in fact, have a monopoly in the EU.
In terms of medical data the amount of leakage is somewhat limited by the nature of DNA, e.g because you get a random mix you can't conclude anything about parents, etc medical status.
By far the biggest practical knock-on effect is if you match someone who's doesn't know their parentage (adoption/illegitimate children/etc) who can figure out their parentage as a result of that match.
Familial DNA crime searches are probably the next biggest, but they're still very rare at the moment and many of the DNA platforms don't allow them (GEDMatch was one of the few that do).
> I’m assuming that the stolen information doesn’t have this limitation
Stolen information has provenance problems that make it difficult to use as evidence of any crime other than theft itself in any system with even rudimentary due process protections and presumption of innocence.
I mean, it's hardly as if you are going to be able to get the people who handled the data between the people who had it lawfully and the time it got to the police on the stand to attest to it's integrity.
(That doesn't prevent its use in investigations, but it means that it would only lead to convictions in a contested case where the police used it to locate proof that was legally sufficient without the use of the DNA as evidence.)
Most law enforcement typically use a handful of commercial agencies (like Parabon NanoLabs, etc) for these kind of searches, it seems highly unlikely that any of them would risk using illegally obtained data because it would put their entire business at risk.
(obviously if your threat model includes intelligence agencies, etc. then your calculus might be different)
It's not about law enforcement using the data, it's about the viability of running a business that provides illegal hacking services for law enforcement.
I’m cackling at the idea of GDPR compliant data thieves. “This data is only to be used for the purposes of: anything. The data controller is: whoever.”
> You know what parent a male's X and Y came from.
You can identify which parent any chromosome came from. They're all marked, and the same genetics may do sharply different things depending on whether it was inherited from the father or the mother.
Inability to recover this data has nothing to do with "the nature of DNA" -- the data is very much present in the DNA. It's unrecoverable because when we summarize DNA, we leave it out.
> You can identify which parent any chromosome came from. They're all marked, and the same genetics may do sharply different things depending on whether it was inherited from the father or the mother.
I did not know this. This sounds interesting! Can you provide any google search terms (or a link) where I can read more about this? (e.g. a name of what they are marked with)
This surprises me. I thought that there was a process by which portions of the two copies of a chromosome get switched between the two. Is that right? How does that fit together with these markings?
(If these questions would be answered by searching for whatever search term or reading whatever link you provide, I would consider providing said search term or link to be answering these questions)
> Can you provide any google search terms (or a link) where I can read more about this? (e.g. a name of what they are marked with)
The term I know related to this is "methylation". https://en.wikipedia.org/wiki/DNA_methylation . I don't know all that much about it; I would not want to claim that methylation is the only such mechanism, or that this is the only information expressed by DNA methylation.
> I thought that there was a process by which portions of the two copies of a chromosome get switched between the two. Is that right? How does that fit together with these markings?
Yes, that's correct. "Crossing over" does not occur during ordinary cell division ("mitosis"), in which one of your cells divides into two of your cells -- your chromosomes should stay the same (except for new mutations) through your life.
But it does occur during meiosis, the process by which one of your cells divides into four sperm or four eggs (these are "gametes", and in terms of chromosome content they are only half-cells, not full cells). Your children's chromosomes may therefore differ from yours.
So the interaction between parental marking and crossing over would broadly look like:
1. You are going to produce four gametes.
2. Remove the parental marking (indicating the sex of the gamete's grandparent) from the cell undergoing meiosis.
3. Do the crossing over.
4. Apply parental marking indicating your own sex (the gamete's parent, rather than grandparent).
5. Divide into four cells.
I don't actually know where the unmarking and remarking occur in the process; maybe reality is more like 2435, or 3254. But both crossing over and applying correct parental marking are part of meiosis -- since meiosis produces a cell that belongs to your child rather than a cell that belongs to you, it's easy to know what kind of marking should be applied.
Yes, but there's generally not much medical data you can infer from those.
You're right that you could reconstruct parental haplotypes, but that reveals a fairly limited amount of data, typically you'll share haplotypes with many millions of people.
What would insurance companies do with the data though? If they knew you were predisposed to obesity and cancer due to this data, would they be kind enough to ignore that info?
If we’re hypothetically considering what they could do if they weren’t one of the most regulated industries, they have exponentially better options for limiting their risk than requesting DNA.
It’s not only medical data that’s of concern, but also nation states Could try to use the data to identify embedded foreign agents/spies implanted in their country. Those are the ones without diplomatic cover
Would recommend the book The Complete Guide to Capital Markets for Quantitative Professionals, it's more an intro to how finance works rather than quant dev but will give you a sense of what it involves at a high-level.
Fundamentally it's hard to see how the current situation is any different from the situation in which anti-trust regulators moved against Nintendo and Sega in the mid-90s.
In both cases you have a duopoly which essentially controlled the market and used access to the platform to force fees and conditions onto developers using their gatekeeping abilities.
One difference that comes to mind relatively quickly is that you didn't need a gaming console to live in practice.
Try being part of today's society without a smartphone running either Apple or Google software.
Apple's goal to shove off 30% of subscription fees, if successful, most likely would lead to more expensive subscription fees for end users. I wonder whether that's finally a case that has a chance to stand in court w.r.t. anti-trust laws. (Driving up end user prices through anti-competitive behavior.)
As someone who previously competed on Kaggle, this seems a reasonable decision. In previous contests it was pretty clear if you wanted to do something that used third party data you should get pre-clearance for it from Kaggle/contest organizers.
The disqualified competitors here seem to have assumed that CC-BY meant you can do whatever you want with data, when actually that's far from true. CC-BY is solely about copyright and doesn't address other rights (e.g. model release, gdpr, etc.)
Blossom Capital | London | Software Engineer in Venture Capital
We back Europe's top startups at Series A (including YC almumi Duffel, Fat Llama and Sqreen) and building technology is a core part of how we operate, from sourcing deals through to automating due diligence.
Roughly a third of the companies we invest in are sourced through our technology platform and we're looking to hire someone who will help take it to the next level.
Our current stack is Python + Postgres based, but beyond engineering skills we're looking for someone who'll enjoy spending time with the investment team really understanding what we look for investments and thinking about how to translate that into technology!
If that sounds interesting drop me a line at imran@blossomcap.com
Blossom Capital | London | Software Engineer in Venture Capital
We back Europe's top startups at Series A (including YC almumi Duffel, Fat Llama and Sqreen) and building technology is a core part of how we operate, from sourcing deals through to automating due diligence.
Roughly a third of the companies we invest in are sourced through our technology platform and we're looking to hire someone who will help take it to the next level.
Our current stack is Python + Postgres based, but beyond engineering skills we're looking for someone who'll enjoy spending time with the investment team really understanding what we look for investments and thinking about how to translate that into technology!
If that sounds interesting drop me a line at imran@blossomcap.com
There's a big difference between what a product does in theory vs in practice - and social proof is a big guiding factor for what it can do in practice.