Yeah, the hype machine is basically a feature of the industry at this point. What got me though is that the demos aren't even dishonest, they just measure the wrong thing.
True! But if you think about their success criteria, for some companies or labs, if it gets them that investor check or that grant funding, it served its purpose.
But most people forget that "we'll iron out the details later" is what gets you such a massive field like robotics, with 20 yo archaic approaches and stacks.
That's exactly the tension. The demo optimizes for the funding decision, not the deployment decision. Those are genuinely different problems. The scary part is that "we'll iron it out later" has worked often enough in software that people keep applying it to hardware, where later has a way of becoming never.
What's interesting about Sunday is that they've essentially restructured the incentive. Their pitch to Coatue wasn't a better demo. It was a Thanksgiving deadline and a data flywheel that only starts spinning when the robot is actually in someone's home. The check and the deployment became the same thing. Whether that holds is a different question, but at least the optimization target shifted.
Let's hope that this will improve with enough push from the community + big players making better moves.
Sooner or later, investors and early adopters will start to see the gaps in the storytelling, which will be patched by other startups and the open-source community working on the real problems!
Awesome article, but the Hype machine will always roar i guess.
Yeah, the hype machine is basically a feature of the industry at this point. What got me though is that the demos aren't even dishonest, they just measure the wrong thing.
True! But if you think about their success criteria, for some companies or labs, if it gets them that investor check or that grant funding, it served its purpose.
But most people forget that "we'll iron out the details later" is what gets you such a massive field like robotics, with 20 yo archaic approaches and stacks.
It will get better tho, Im sure!
That's exactly the tension. The demo optimizes for the funding decision, not the deployment decision. Those are genuinely different problems. The scary part is that "we'll iron it out later" has worked often enough in software that people keep applying it to hardware, where later has a way of becoming never.
What's interesting about Sunday is that they've essentially restructured the incentive. Their pitch to Coatue wasn't a better demo. It was a Thanksgiving deadline and a data flywheel that only starts spinning when the robot is actually in someone's home. The check and the deployment became the same thing. Whether that holds is a different question, but at least the optimization target shifted.
Facts!
Let's hope that this will improve with enough push from the community + big players making better moves.
Sooner or later, investors and early adopters will start to see the gaps in the storytelling, which will be patched by other startups and the open-source community working on the real problems!