2YwaZHXV
2YwaZHXV on Hacker News
would certainly be interested to learn more about this
I haven't tried it much but this seems like precisely the problem gitbutler is trying to solve
Presumably there's no way to get someone at Google to query their database and find all the shortened links that go to godbolt.org?
well, I see a flash of a normal page and then it goes to 404
I'm using Chrome, with only uBlock Origin Lite but disabling it ~shows no change~ appears to allow it to work.
Looking at the requests with uBOL disabled, app.68194690.js and 5.5caaec0f.js both returned 404, everything else returned 200/202.
and with uBOL enabled https://plausible.hey.si/js/script.hash.outbound-links.js shows as blocked. maybe that's the issue?
makes sense, thanks.
the images on the page are all just base64 encoded right into the html
given that it seems like the entire thing is busted, can anyone explain how the unicorn page is being served?
While it is published on their site, it appears to really just be a summary of the book by Carreyrou? And I have yet to see anyone actually publish any of the dilution ratios and compare that to any existing process, so it seems like an easy-but-easily-incorrect leap to conclude that it is "way more than usual".
Until seeing your comment I had never considered that someone might actually want any of that stuff, so I had never considered selling or otherwise parting with it since it is fun to keep all that stuff around. I'll certainly let you know first if I decide to part with it, though.
One thing that seems to be forgotten or lost in all of these armchair analyses is that while it is true that something like 3mL of blood is collected for a conventional lab test, the machine itself when it runs the test only takes a few microliters (maybe 10s of uL, all of this is depending on the test of course). The rest is frozen for follow-up or repeat testing or used for other tests or just discarded after some time.
EDIT: And my favorite part is the first step the Siemens machines they used for testing is to "dilute" the part of the sample it is going to use (addition of DI water, to get the blood to the desired concentration). That is hard-coded into the protocol, with a specific dilution amount, and it is done for all the official Siemens tests, as well, hah.
As someone who used to work there and has one of those water bottles and likely knows the person selling that one on ebay (based on the username) I find this hilarious...
While not quite as big as other pumped storage systems this is one system that is looking to avoid the exposed area and dependence on natural reservoirs:
https://heindl-energy.com/
From my experience buying/getting quotes for Hamilton/Tecan machines, they're more likely in the $150-250k range. At least for the ones much wider and capable than that OpenTrons appears to be.
The interest level in achieving the best result (and the required associated time, patience, planning, skill, and potentially cost) is likely the more limiting factor.
I"m saying this as someone that works with vision systems for industrial settings. Often I see garbage images with garbage logic and they complain about garbage output. Then the camera is removed because "it doesn't work".
As someone who works with modern industrial robots, like you'd find in a car factory, the sensors they're using for the arm positioning are unbelievably precise. I can have the robot stopped, with the brakes on, and push on the end of the arm with my fingers and see the joints registering the deviation (and not by 1 or 2 counts, but many more), while I don't perceive any significant movement of the robot.
They published a video of the manufacturing that showed those CNC machines, injection molding, and robotic assembly.
It's a lot faster to iterate if you have in-house machining and molding; you can get parts made or modified in <day vs weeks with an outside shop. You can also work closer with the machinists to iron out challenges.
I know people that worked there and having Theranos on their resume caused them to not get an interview at at least one company.
OK, so these are more business reasons for bigger samples than technical reasons.
1. Ok, so that can still be done with small samples, it's just a matter of collecting enough. The link I gave showed 15 tests in 30ul of sample. So, 60ul would be enough for your 30 tests. The amount used per test (for many many tests) is still very small compared to the amount drawn.
2. I imagine if a tube is dropped and some sample is lost they consider the whole sample lost. Since if sample got out that also means that stuff can get in, possibly contaminating it. The poor customer experience of having to go back in to be re-drawn is a poor business choice, but again, not a technical hurdle with being able to perform tests. Having enough sample for re-testing a week later is also just a matter of collecting sufficient sample. So, you would need 120+ul of sample (from previous example of 30 tests) to meet that requirement (plus some amount for loss in equipment, etc). But, it is still a far cry from the 2000+ul typically drawn.
3. You actually often have more than 2 drawn, since each tube has a different kind (if any) anticoagulant and if they get sent to different labs they may need duplicates of the same anti-coagulant. This also doesn't have anything to do with the size of the sample required for a test. So, you can collect a bunch of small samples instead of bigger samples. That's not a technical hurdle for being able to perform a test on a small volume of blood.
Ok, so you're giving a business reason to not collect smaller samples. Which is fine and totally valid. Theranos was under the impression that there was an untapped opportunity for smaller samples.
But, the wording of the parent comment was saying that it was supposedly clearly technically impossible because of the small sample size. The OP said "I just never, ever understood how the sampling rate could ever work" and "and this should've been screamingly obvious to everyone since the beginning." That seems very different from "this is a very hard problem due to X, Y, and Z", especially since commercial machines that everyone here gets blood results from are already only taking small samples from the collected vials so the biggest challenge just seems to be in the collection and transport.
I agree that everything is easier with a bigger collection size, however "hard" is different from impossible.
Existing analysers already use very small samples.
See Siemens Advia 1800 specs under "Microvolume Technology"
https://www.healthcare.siemens.com/clinical-chemistry/system...
It only uses 30ul of sample to run up to 15 assays.
How is that any different from what Theranos was doing?
Only difference is that you collect a bigger vial when using the Siemens machine. And then don't use most of it. So why not collect a smaller vial?
It was actually a Wall Street Journal building before it was torn down to build the building Theranos moved into.
Theranos was previously in 1601 California Ave (the old Facebook building), which was previously HP/Agilent.
They didn't actually build it. Stanford owned the land and building (and built most of it) while Theranos paid for internal improvements. Stanford also recently sold the property (maybe a year or so ago) and 1701 is available for lease now.