Blog/2026-01-27/The Obvious Isn't Obvious: Difference between revisions
No edit summary |
No edit summary |
||
| Line 27: | Line 27: | ||
}} | }} | ||
The first and last comments are the most popular on Hacker News. Popular sentiments among programmers (or at least the kind who visit Hacker News) are therefore aligned on the Get It Right First Time and No Failures mindsets. What Amazon was doing is pretty standard in refinement of ML techniques. You build an imperfect program that gets you some of the way, then have humans review the output of the program, and that is used to refine the program | The first and last comments are the most popular on Hacker News. Popular sentiments among programmers (or at least the kind who visit Hacker News) are therefore aligned on the Get It Right First Time and No Failures mindsets. What Amazon was doing is pretty standard in refinement of ML techniques. You build an imperfect program that gets you some of the way, then have humans review the output of the program, and that is used to refine the program<ref name=palm-scan>. | ||
[[File:Amazon One Palm Authentication Discontinuation Email.png|frameless]] | [[File:Amazon One Palm Authentication Discontinuation Email.png|frameless]] | ||
| Line 67: | Line 65: | ||
== Notes == | == Notes == | ||
<references/> | <references> | ||
<ref name=palm-scan>Unfortunately, my favourite feature at Whole Foods - palm-scan for payment and Prime membership - is going to die along with this. | |||
</ref> | |||
</references> | |||
Revision as of 00:41, 4 February 2026
When you live in San Francisco and around the Silicon Valley ecosystem, a mantra that everyone repeats is 'iteration'. Attempting to reach product-market fit is traditionally, here, described as an iterative process where you put out a minimum viable product and then work on it in response to customer feedback. This is so ingrained in the psyche here that if you were to say it, you sound like a freshman in Mathematics explaining this crazy thing called a set of simultaneous linear equations: it's far too basic to be talking about.
That is true. Here. But that's why so many startups come from here and so few from elsewhere. The traditional-for-elsewhere approach is far different and as Hacker News enters its Eternal September, non-startup software engineers who believe in another tradition dominate the audience. Every encounter with them reveals that the Iteration Dogma is something actually unique to startup culture, and that despite there being great successes from the approach, it may never actually break out into the mainstream.
Here are a few comments that illustrate this:
On April 4, 2024, it was revealed that Amazon's "Just Walk Out" technology was supported by approximately 1,000 Indian workers who manually reviewed transactions. Despite claims of being
fully automated through computer vision, a significant portion of transactions required this manual verification.
This was proven to be false on the WAN show. Only 20% of transactions were low confidence and handled by mechanical turk.
It's great that they faced essentially no consequences for this. A sure sign that we have a functional and sane market.
The first and last comments are the most popular on Hacker News. Popular sentiments among programmers (or at least the kind who visit Hacker News) are therefore aligned on the Get It Right First Time and No Failures mindsets. What Amazon was doing is pretty standard in refinement of ML techniques. You build an imperfect program that gets you some of the way, then have humans review the output of the program, and that is used to refine the programCite error: Closing </ref> missing for <ref> tag
</references>
