Hacker News Reader

Comment by justinator | Hacker News Reader

I believe they also reverse engineered a lot of the third party machines, in an attempt to make them work the way they wanted to. That's... that's bad.

View on HN

Emphasis on attempt.

After failing to produce their own reagents and hardware, Theranos pivoted to using off-the-shelf Siemens hardware and reagents (purchased through a shell company) and dramatically diluting the blood samples to make good on the "single drop of blood" customer promise.

This would've been an impressive feat of reverse engineering, software eating the world, IF IT WORKED. But it didn't. Results couldn't even pass internal audits and the FDA forced Theranos to invalidate ~1 million test results when they found what was going on.

Reverse engineering is an important right that is protected by centuries of patent and trade secret law in the US and many other countries. It's fundamental to innovation, even though non-innovative bad people like Theranos also do it.

Reverse engineering is not bad, and it is dismaying that someone posting on a site called "Hacker News" would think so.

Reverse engineering for shits and giggles, good. Reverse engineering for profit by modifying a device regulated by the FDA, bad.

I think you mean, reverse engineering to create new, better machines: good. Reverse engineering to game a medical testing device: bad.

Reverse engineering is not bad, and it is dismaying that someone posting on a site called "Hacker News" would think so.

Personal attacks aside,

I'm just saying that if I get a blood test and it says I've got AIDS and I actually don't -

and that's all because some sociopath grifter with less of an education than what's been proven with my own Bachelors of Fine Arts realized that to swindle millions of more dollars from her geriatric backers she's been manipulating (and ruining the lives of their own family members),

she needs to fake the test results from an impossible dream machine that does NOT exist and couldn't ever - because it defies the laws of physics,

and does this by running modified machines of her competitors that are now basically broken,

in secret rooms she doesn't tell journalists or inspectors about,

well: that's truly fucked.

Yes! And reverse engineering is not bad. Everything else you mention is bad.

I do think there's nuance when the product is a medical testing device, and the company doing the reverse engineering is a competitor of the company who made said device and that engineering was faulty and could have cost people's lives.

But I don't know if the context of any of this is what some really want to discuss. It's just the slogans. You know: "hack the planet" and all.

I also don't think I'd be really OK with working for a company that told me to do this reverse engineering, but if your morality says differently, then: hey.

I don't think you should be criticizing reverse engineering here any more than you should be criticizing programming, science, skill, studying, engineering, logic, thought, changing your mind when you're wrong, or any of the other basic aspects of hacking. You're as out of place here as someone complaining about how nerds study all the time, get better grades than you do, and know how to fix their computers while you're stuck paying someone else to do it. You come off as a troll posting deliberate flamebait in hopes of getting a rise out of us.

Of course badly done reverse engineering can be bad. But that's not because it's reverse engineering; it's because it reaches incorrect conclusions that hurt people, just like not doing any reverse engineering at all. The problem that put people at risk in this case wasn't that Theranos did reverse engineering; it's that they didn't do enough reverse engineering.

> I don't think you should be criticizing reverse engineering here any more than you should be criticizing ...

No, sure, real reverse engineering is fine.

But when "reverse engineering" is used in the sense Theranos (and/or Holmes' defense team) apparently did -- i.e. tampering with a competitor's machines so they sometimes don't work properly, and passing off the sometimes-invalid results as super-duper hot stuff created by your new and superior WonderTech™ -- then yes, it darn well should be criticized to Hell and back.

When "reverse engineering" is used to mean all that other shit, then that "reverse engineering" is bad.

Well said, a shameless grifter needlessly put people at immense risk because ego is a thing, and showed craven desire for money and celebrity.

Agreed. My criticism is that these problems should not be blamed on reverse engineering any more than they should be blamed on histology.

I think "reverse engineering" in the gp was just an ill-considered choice of word. What they are describing doesn't really have anything to do with reverse engineering -- it's just using the devices in a way that will not produce the desired reliability of results. The fact that their methods differed from what the manufacturer recommends would not be a problem if they could prove that the false positive/false negative rates were up to the relevant standards.

> The fact that their methods differed from what the manufacturer recommends would not be a problem if they could prove that the false positive/false negative rates were up to the relevant standards.

Well, better than the relevant standards. Otherwise all the "reverse engineering" -- i.e, tampering -- is fucking meaningless.

It's meaningless if nothing is better about how you're doing it. Presumably in this case the thing they wanted to improve was the amount of sample you need to use, not the false positive/negative rates.

But they knew at the time they were peddling it that they didn't yet have that advantage. So selling it on the premise that they did was... Well, fraudulent, in my book.

Their use of the term "reverse engineering" is perhaps not ideal.

If they were taking apart and studying the hardware inside competitors' machines, then that would not have been a problem. It's something which many companies do. But they would not be using this for patient testing, they would use it as input to product development and engineering.

Theranos were modifying the machines to operate outside their design parameters. Medical diagnostic equipment is validated at great expense to operate within given parameters to give a diagnostic result which is known to be accurate with a very high probability. They are carefully calibrated and tested to guarantee the accuracy of every test. Tampering with them makes them invalidated and destroys any guarantee of diagnostic result accuracy. The FDA approval for the equipment is only valid if they are operated as intended.

Agreed. My concern is precisely with this conflation of Theranos's negligent misconduct with the exercise of the fundamental rights which allow us to fix our cars (if you fix your car without a shop manual, you're doing reverse engineering --- and both Chilton and Haynes do reverse engineering to write their shop manuals), fix our laptops (see Louis Rossmann), uncover Microsoft's dirty tricks to break competitors' software (reverse-engineering was crucial to the antitrust trial), preserve our information (both PDF and Word formats were reverse engineered before they were opened), use IBM-compatible PCs (the clone industry was able to launch because Phoenix reverse engineered the BIOS and wrote a cleanroom clone), and generally do anything with a manufactured device that the manufacturer didn't intend.

For this reason there are carve-outs in copyright law where something is illegal except when done for the purpose of reverse engineering; for example, in the US, you have Sega v. Accolade.

But not illegal, except under very specific circumstances.

I think that's entirely fine - both for patients and investors (Siemens might be a bit Irate, but I don't know if I would put that in the "Bad" category). The issue would be if they informed their investors that tests were being done on Theranos technology, but were actually using siemens technology. With regards to the patients - all they should care about is accuracy of test results - whether Theranos was doing it with Theranos tech, Siemens Tech, Reverse Engineered Siemens Tech + dilution of the blood to get enough volume is mostly irrelevant to a patient - they just want to get the highest quality information possible, regardless of the route taken.

> Reverse Engineered Siemens Tech + dilution of the blood to get enough volume is mostly irrelevant to a patient

Those machines weren't designed to work that way though, so the test results were wildly inaccurate when they did dilutions. And they fired people who wanted to delay rollout in order to ensure that they had consistently accurate results.

The point I was making was that it was irrelevant how the original designs were intended - if Theranos had been able to successfully reverse engineers the tests, and determine a way to make them work reliably, while it would have made those of us who believed Theranos had some incredible new technology, and that Elizabeth Holmes was the next coming, no patients need to have been concerned. It was the fact that the tests were unreliable, that made Theranos guilty.

Lying about it to the Investors, and possible intellectual/contract issues with Siemens are another issue.

> It was the fact that the tests were unreliable, that made Theranos guilty.

Very few tests are 100% reliable, so I would expect that Theranos claims about how the tests where performed would play a significant role in determining the guilt. People where told they would get the perfect magic pixy dust test that worked on one drop of blood instead they got one of the early covid tests that needed at least four repeats to give a "most likely negative" result, unsurprisingly the covid tests also resulted in confused tweets by the one or other VC.

Well - some of the tests have to be 100% reliable, because medication decisions are made on them, and the incorrect dosage can be harmful. It's not like a cheap $10 antigen tests that has sub 100% specificity and sensitivity - a bunch of the tests were trying to measure particular levels. Theranos was just overall horrible. They were trying to do some tests that were impossible based on the physics of what can be measured by pinprick blood.

Yeah, I think we're actually in agreement here. Sometimes I think about how, if Theranos had figured out how to make things work over the last year before the WSJ article, all of their horrible and unethical practices would have been swept under the rug even if they were essentially lying for years to investors and patients.

Unfortunately, it turns out that "fake it til you make it" doesn't work with some problems that are seemingly unsolvable (at least by the approach they took).

> Unfortunately, it turns out that "fake it til you make it" doesn't work with some problems that are seemingly unsolvable (at least by the approach they took).

IOW, it turns out that "fake it til you make it" doesn't work if you don't actually make it.

Reply on news.ycombinator.com