News for hackers. Stuff that matters.

A comment on Theranos devices ran “null protocol” to skip actual demo for investors. Parent comment

Re:Theranos devices ran “null protocol” to skip actual demo ... by PragmaticPulp on Wednesday October 20, 2021 @06:28PM (#28937644)
> Having scripted demos that hide error messages is not really sketchy. It could be part of a fraud, or it could not, but it's not really telling.

They pricked investors fingers and implied that a Theranos machine would process the sample, but then they secretly ran the sample on a competitor's machine in another room:

> In demos of the company’s technology, investors would have their fingers pricked for testing alongside one of Theranos’ proprietary devices, but the actual test would be run in a lab away from prying eyes.

This would be the equivalent of a SaaS company setting up an API that just passed the calls through to a competitor's API, but then telling investors that their own platform was operational. You can run the pass-through API long enough to fake demos and raise money, but it's still fraud if you mislead investors to believe that you have the platform operational.

They weren't just hiding error messages or smoothing over rough edges. They were architecting fraudulent claims to investors.

[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by the_jeremy on Wednesday October 20, 2021 @07:18PM (#28938121)
The first iPhone demo involved bringing in a portable cell tower, hardcoding signal strength to always show 5 bars, and repeatedly swapping out phones throughout the demo because using one through the whole presentation would have caused it to crash[0].

I believe Theranos was worse in many ways, but running misleading demos to drum up hype sounds pretty par for the SV course.

[0]: https://www.nytimes.com/2013/10/06/magazine/and-then-steve-s..., https://www.techspot.com/news/54246-former-apple-engineer-re...

[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by Aloha on Wednesday October 20, 2021 @07:35PM (#28938270)
Bringing in a COW is standard course when you have a large number of people in one place, even in 2007.
[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by eli on Wednesday October 20, 2021 @06:56PM (#28937914)
Pretty sure a non-zero number of AI/ML startups were actually just feeding data to humans, at least at the start.
[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by spiffytech on Wednesday October 20, 2021 @07:14PM (#28938096)
There was a story earlier this year about a startup that had a government contract to detect crime using AI, but turned out not to actually have any AI technology, instead appearing to pass tasks to humans: https://www.vice.com/en/article/pkd7pk/banjo-ai-surveillance...
[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by woeirua on Wednesday October 20, 2021 @07:34PM (#28938267)
This is not a problem if you're transparent about what you're doing. If you say, the AI/ML will do this, but we're still perfecting the algorithm so we have humans assist it right now. No problem.
[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by HWR_14 on Wednesday October 20, 2021 @08:05PM (#28938528)
There was just a report on HN about Expensify doing that. I don't know if they switched to AI, or if it's still just people.
[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by spaetzleesser on Wednesday October 20, 2021 @06:59PM (#28937944)
“At least at the start” is possibly ok if you have a path forward to get to your goal. Theranos had no path.
[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by spiffytech on Wednesday October 20, 2021 @07:21PM (#28938149)
> “At least at the start” is possibly ok if you have a path forward to get to your goal.

Only if you're not misrepresenting your capabilities to investors or customers.

[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by AndyNemmity on Thursday October 21, 2021 @12:25PM (#28946853)
It's tricky. If you're saying what you will do, and show it as it should look, while you continue to develop it really working, and it's not endangering health like in the case of Theranos, I have a hard time seeing the issue.

I've never been in a situation where it literally didn't work though over time, and before customer usage.

[ Parent | on HN ]
Re:Theranos devices ran “null protocol” to skip actual demo ... by AndyNemmity on Thursday October 21, 2021 @12:24PM (#28946830)
can confirm a few examples. the key is at the start though, they all made it there over some time.
[ Parent | on HN ]