As a big fan of Matt's writing, I think this is worth the (not too too long) full read, which elucidates exactly how iRobot transformed from innovative DARPA-funded former-MIT professor-run tech...
As a big fan of Matt's writing, I think this is worth the (not too too long) full read, which elucidates exactly how iRobot transformed from innovative DARPA-funded former-MIT professor-run tech company to... a drop-shipper of Chinese hardware and software. But the last few paragraphs serve as a decent TL;DR:
In other words, the story here is Wall Street destroying a promising robotics enterprise through financial engineering, aiding the Chinese in the process, and then demanding a bailout via amnesty from antitrust laws so that shareholders wouldn’t lose any money, while refusing to acknowledge that a key Trump ally of Wall Street facilitated the transfer of the firm to China.
Of course, this bad faith is routine. None of the critics of antitrust enforcement, including Furman, care if U.S. technology flows to China or if companies fail. They in fact celebrated offshoring when it happened to 90,000 manufacturing plants from 2000 onward, and they often make the point that failure is part of capitalism. But when it comes to one specific company, where they can cherry pick information to make a case against antitrust, well then, all of a sudden iRobot’s bankruptcy is a disaster.
All that said, there is an important lesson here for anti-monopolists. Antitrust is a useful tool, but it cannot substitute for a broader national economic development strategy. Right now, America, through a whole set of policy choices, from bailouts to government contracts to pro-speculation regulations to attacks on the rights of labor and creators, ensures that financiers get an unfairly high return on capital. We can see the consequences in everything from the collapse of iRobot to the destruction of America’s cattle herd to the erosion of capacity in Hollywood to the financialized AI data center build-out. The business of America right now is extraction, not creation.
To reverse this strategy, a more assertive antitrust regime is necessary, but it’s not enough. We also have to reduce the many other public levers of support for elevated returns on capital. Only then will it make sense for companies like iRobot to invest in robots instead of share buybacks.
Skimmed the piece, so maybe it got addressed, but it seems like a bit of yes and no. Stripping down the company and neglecting research was bad, but a lot of that research was for robot usage in...
Exemplary
Skimmed the piece, so maybe it got addressed, but it seems like a bit of yes and no. Stripping down the company and neglecting research was bad, but a lot of that research was for robot usage in the military. I think the idea was that you can't focus on both military applications and vacuum appliances simultaneously and do both that well, so the investors aggressively wanted iRobot to focus on vacuum appliances. Which isn't that unreasonable of a direction either.
That pivot might have gone alright, but here's where the article neglects to mention that iRobot spent a long time not innovating. There were cheaper competitors that had LiDAR mapping out floorplans for efficient vacuuming while, at the same price point, your stupid Roomba would just randomly run in a direction until it bumped into something and repeat. It fucking sucked. My parents had a Roomba and I had a Roborock. Worlds apart in functionality despite being about the same price.
No, Lina Khan did not kill iRobot, but it's not totally fair to pin the blame on the investors either. iRobot seemingly rested on their laurels as first-movers and stopped adding real value over the years.
Research that, according to an iRobot spokesman, was intended “to spur innovation in robotics and robotics-related areas.” Mesdag argued that the money should have been used for share buybacks to boost the stock value instead of for research…
“In my trips to Wall Street,” Dyer told the panel, “one of my analyst friends took me to lunch one day and said, ‘Joe, you have to get iRobot out of the defense business. It’s killing your stock price.’ And I countered by saying ‘Well, what about the importance of DARPA and leading-edge technology? What about the stability that sometimes comes from the defense industry? What about patriotism?’ And his response was, ‘Joe, what is it about capitalism you don’t understand?’”
The above is about the defense side of research. I don't see where this stopped iRobot from focusing on the most obvious of obvious which is: make the vacuum cleaner not just bump into shit randomly as your best algorithm for cleaning. They eventually added it.. late and more expensive.
EDIT: The first Roomba with LiDAR was released this year. The first Roborock with LiDAR was released in 2019. Roborock wasn't even the first to do it.
Oh yeah before anything else, what killed Roomba was Roomba. They weren't even mediocre robot vacuums, they were bad robot vacuums. Even their actual vacuuming left something to be desired and...
Oh yeah before anything else, what killed Roomba was Roomba. They weren't even mediocre robot vacuums, they were bad robot vacuums. Even their actual vacuuming left something to be desired and their pathfinding was even worse.
The robot vacuum market was dominated by ten other brands before Roomba. They wasted away when their name recognition was no longer enough.
I mean, it's really hard read "Joe, what is it about capitalism you don't understand?" And not blame investors. Being consumer focused doesn't mean focusing on selling cheaper crap and...
but it's not totally fair to pin the blame on the investors eithe
I mean, it's really hard read "Joe, what is it about capitalism you don't understand?" And not blame investors. Being consumer focused doesn't mean focusing on selling cheaper crap and undercutting employees, but that's "capitalism" when left unchecked.
And that's the charitable interpretation. Odds are they instead spent the pandemic preparing iRobot for a big fat selloff. And then it just didn't happen, for various reasons (I. E. They couldn't pivot fast enough to the AI hype train either). Americans companies this decade have not been innovation focused at all.
We just got a Roborock (based on Wirecutter's recommendation) for our new house and it kind of works, but it seems kind of mediocre at cleaning rooms with rugs. It tends to leave dust bunnies...
We just got a Roborock (based on Wirecutter's recommendation) for our new house and it kind of works, but it seems kind of mediocre at cleaning rooms with rugs. It tends to leave dust bunnies everywhere despite cleaning the brush often. Maybe it's because we also got new rugs.
I have a $1300 Dreame. Piece of shit at getting wood floors clean, especially edges and corners compared to my two older Roomba s9. I will say, the lidar is amazingly good at mapping (well, no...
I have a $1300 Dreame. Piece of shit at getting wood floors clean, especially edges and corners compared to my two older Roomba s9.
I will say, the lidar is amazingly good at mapping (well, no shit) compared to a camera system.
I'm so angry I spent money on a Dreame. I want a high quality robot vac that lets me run the OSS all-local backend.
You're going to be stuck with Valetudo for anything open source of self installed. Unfortunately it's limited to certain models only. For anything else I tend to check vacuumwars.com as they're an...
You're going to be stuck with Valetudo for anything open source of self installed. Unfortunately it's limited to certain models only.
For anything else I tend to check vacuumwars.com as they're an independent reviewer of robot vacuums that quickly updates their lists with new models, price changes, and other interesting changes in the market.
I chose mine based on vacuum wars. Its abysmal performance is why I have serious doubts about them going forward. Which is a shame, because I really did trust their process and valued their...
I chose mine based on vacuum wars. Its abysmal performance is why I have serious doubts about them going forward. Which is a shame, because I really did trust their process and valued their reputation.
That's genuinely surprising to me. I haven't seen any misplaced reviews myself, and I'm of half a mind to consider the incredibly frustrating and opaque naming schemes of the vacuum models as the...
That's genuinely surprising to me. I haven't seen any misplaced reviews myself, and I'm of half a mind to consider the incredibly frustrating and opaque naming schemes of the vacuum models as the main culprit, but hey they can be wrong too. Carpets are difficult to assess too, as how they're made differs so much that one easy to clean carpet doesn't mean the other one is.
Robot vacuums are in a weird position where they are overall worse than if you would do it yourself and their value derives from the ease of use and frequency of vacuuming.
I see it as an increase in baseline cleanliness, and a big one at that, while still needing to clean behind things.
I was shook, too. Had used them extensively for my two Roomba s9. Did a couple weeks worth of noodling before decided to pull the trigger on the Dreame X50 Ultra. At the time it was going between...
I was shook, too. Had used them extensively for my two Roomba s9. Did a couple weeks worth of noodling before decided to pull the trigger on the Dreame X50 Ultra. At the time it was going between #1 and #3 on their list as prices fluctuated.
Great piece of writing. One cannot simply trust the words of Wall Street analysts and financiers. They have too many financial incentives to advocate for specific messaging to ever have public...
Great piece of writing.
One cannot simply trust the words of Wall Street analysts and financiers. They have too many financial incentives to advocate for specific messaging to ever have public takes that are at all based in truth. In some sense, I think that even critiquing their “opinions” is not very helpful because of how shallow and fake those opinions obviously are to any informed person.
Even if the piece does feel like preaching to the choir, Matt does a great job breaking down how American capitalists will choose to compromise American interests if they will get these capitalists greater returns. While I’m not sure that the national security concerns are legitimate, the consequences are still clearly a result of Wall Street favoring profits over stability and expecting government support.
I paraphrase the argument because I’ve heard some concerns from people in the AI space about the extraordinary investment in companies causing perverse incentives; investors expect enormous returns on their enormous investment, and so companies may be pushed to make otherwise bad decisions in pursuit of these profits. I agree with the take, but these incentives really aren’t anything new, and would not disappear absent the AI investment numbers. This piece is a good reminder that those incentives will always exist in a financialized capitalist economy.
Great point, it's true that treating them as actual opinions rather than cynical strategies is a contradiction in terms. I thought the author ultimately did a fine job of making that clear... and...
In some sense, I think that even critiquing their “opinions” is not very helpful because of how shallow and fake those opinions obviously are to any informed person.
Great point, it's true that treating them as actual opinions rather than cynical strategies is a contradiction in terms. I thought the author ultimately did a fine job of making that clear... and of reminding us that, despite sounding like a dream come true in the current political reality, the Obama administration was nearly as co-opted as the Bush admin.
I think the national security concerns are legitimate, it's safe to assume that any data that goes into China, goes to the government, and Roombas are collecting a lot of sensitive data. Not just mapping homes physically, also network information and physical cameras. It's a mobile surveillance device in 10's of millions of households. At the very least such a sale deserves significant scrutiny.
A million years ago (maybe 5-10), I was briefly involved in a safety assessment for a robot vacuum (not this company). When assessing, we found they had no mechanism to detect negative obstacles....
A million years ago (maybe 5-10), I was briefly involved in a safety assessment for a robot vacuum (not this company). When assessing, we found they had no mechanism to detect negative obstacles. They said they didn't think they needed one. I said, "What about stairs and balconies with horizontal railings? If you drop this on someone, it would be pretty severe." .... Crickets .... Also no answer when we asked about eye safety rating for the laser emitters they has chosen.
The main objective seemed to be to complete the software in time for it to be loaded in the factory before they needed to be shipped from Asia in order to get units in stores for the holiday season. I think they were expecting the assessment to be largely a formality, but when it wasn't, we were dropped from the project.
This is bit of sideways to the article. I really do have to wonder what goes through the mind of someone willingly buying an internet required device they don't urgently need. It really goes to...
This is bit of sideways to the article.
There are about 20 million active Roomba vacuum cleaners in operation, and unless Trump regulators or antitrust enforcers act, now all the data harvested from our homes will go to China.
I really do have to wonder what goes through the mind of someone willingly buying an internet required device they don't urgently need. It really goes to show the absolute unwillingness to consider the broader impact. Not to say average person should have to do that but that is the world we live in.
It also highlights the glorification of computer illiteracy.
Amazon Sidewalk
A corporate mesh network for hire in a variety of products.
Another excellent argument against just don't connect it to the internet stance on smart things. It really is only a question of cost effectivity, public perception and legislation if forced connections are added to more classes of products(for now I am only aware of cars).
As a big fan of Matt's writing, I think this is worth the (not too too long) full read, which elucidates exactly how iRobot transformed from innovative DARPA-funded former-MIT professor-run tech company to... a drop-shipper of Chinese hardware and software. But the last few paragraphs serve as a decent TL;DR:
Skimmed the piece, so maybe it got addressed, but it seems like a bit of yes and no. Stripping down the company and neglecting research was bad, but a lot of that research was for robot usage in the military. I think the idea was that you can't focus on both military applications and vacuum appliances simultaneously and do both that well, so the investors aggressively wanted iRobot to focus on vacuum appliances. Which isn't that unreasonable of a direction either.
That pivot might have gone alright, but here's where the article neglects to mention that iRobot spent a long time not innovating. There were cheaper competitors that had LiDAR mapping out floorplans for efficient vacuuming while, at the same price point, your stupid Roomba would just randomly run in a direction until it bumped into something and repeat. It fucking sucked. My parents had a Roomba and I had a Roborock. Worlds apart in functionality despite being about the same price.
No, Lina Khan did not kill iRobot, but it's not totally fair to pin the blame on the investors either. iRobot seemingly rested on their laurels as first-movers and stopped adding real value over the years.
The above is about the defense side of research. I don't see where this stopped iRobot from focusing on the most obvious of obvious which is: make the vacuum cleaner not just bump into shit randomly as your best algorithm for cleaning. They eventually added it.. late and more expensive.
EDIT: The first Roomba with LiDAR was released this year. The first Roborock with LiDAR was released in 2019. Roborock wasn't even the first to do it.
Oh yeah before anything else, what killed Roomba was Roomba. They weren't even mediocre robot vacuums, they were bad robot vacuums. Even their actual vacuuming left something to be desired and their pathfinding was even worse.
The robot vacuum market was dominated by ten other brands before Roomba. They wasted away when their name recognition was no longer enough.
I mean, it's really hard read "Joe, what is it about capitalism you don't understand?" And not blame investors. Being consumer focused doesn't mean focusing on selling cheaper crap and undercutting employees, but that's "capitalism" when left unchecked.
And that's the charitable interpretation. Odds are they instead spent the pandemic preparing iRobot for a big fat selloff. And then it just didn't happen, for various reasons (I. E. They couldn't pivot fast enough to the AI hype train either). Americans companies this decade have not been innovation focused at all.
We just got a Roborock (based on Wirecutter's recommendation) for our new house and it kind of works, but it seems kind of mediocre at cleaning rooms with rugs. It tends to leave dust bunnies everywhere despite cleaning the brush often. Maybe it's because we also got new rugs.
I have a $1300 Dreame. Piece of shit at getting wood floors clean, especially edges and corners compared to my two older Roomba s9.
I will say, the lidar is amazingly good at mapping (well, no shit) compared to a camera system.
I'm so angry I spent money on a Dreame. I want a high quality robot vac that lets me run the OSS all-local backend.
It's just so infuriating.
You're going to be stuck with Valetudo for anything open source of self installed. Unfortunately it's limited to certain models only.
For anything else I tend to check vacuumwars.com as they're an independent reviewer of robot vacuums that quickly updates their lists with new models, price changes, and other interesting changes in the market.
I chose mine based on vacuum wars. Its abysmal performance is why I have serious doubts about them going forward. Which is a shame, because I really did trust their process and valued their reputation.
That's genuinely surprising to me. I haven't seen any misplaced reviews myself, and I'm of half a mind to consider the incredibly frustrating and opaque naming schemes of the vacuum models as the main culprit, but hey they can be wrong too. Carpets are difficult to assess too, as how they're made differs so much that one easy to clean carpet doesn't mean the other one is.
Robot vacuums are in a weird position where they are overall worse than if you would do it yourself and their value derives from the ease of use and frequency of vacuuming.
I see it as an increase in baseline cleanliness, and a big one at that, while still needing to clean behind things.
Unfortunate that yours is a miss.
I was shook, too. Had used them extensively for my two Roomba s9. Did a couple weeks worth of noodling before decided to pull the trigger on the Dreame X50 Ultra. At the time it was going between #1 and #3 on their list as prices fluctuated.
Great piece of writing.
One cannot simply trust the words of Wall Street analysts and financiers. They have too many financial incentives to advocate for specific messaging to ever have public takes that are at all based in truth. In some sense, I think that even critiquing their “opinions” is not very helpful because of how shallow and fake those opinions obviously are to any informed person.
Even if the piece does feel like preaching to the choir, Matt does a great job breaking down how American capitalists will choose to compromise American interests if they will get these capitalists greater returns. While I’m not sure that the national security concerns are legitimate, the consequences are still clearly a result of Wall Street favoring profits over stability and expecting government support.
I paraphrase the argument because I’ve heard some concerns from people in the AI space about the extraordinary investment in companies causing perverse incentives; investors expect enormous returns on their enormous investment, and so companies may be pushed to make otherwise bad decisions in pursuit of these profits. I agree with the take, but these incentives really aren’t anything new, and would not disappear absent the AI investment numbers. This piece is a good reminder that those incentives will always exist in a financialized capitalist economy.
Great point, it's true that treating them as actual opinions rather than cynical strategies is a contradiction in terms. I thought the author ultimately did a fine job of making that clear... and of reminding us that, despite sounding like a dream come true in the current political reality, the Obama administration was nearly as co-opted as the Bush admin.
I think the national security concerns are legitimate, it's safe to assume that any data that goes into China, goes to the government, and Roombas are collecting a lot of sensitive data. Not just mapping homes physically, also network information and physical cameras. It's a mobile surveillance device in 10's of millions of households. At the very least such a sale deserves significant scrutiny.
A million years ago (maybe 5-10), I was briefly involved in a safety assessment for a robot vacuum (not this company). When assessing, we found they had no mechanism to detect negative obstacles. They said they didn't think they needed one. I said, "What about stairs and balconies with horizontal railings? If you drop this on someone, it would be pretty severe." .... Crickets .... Also no answer when we asked about eye safety rating for the laser emitters they has chosen.
The main objective seemed to be to complete the software in time for it to be loaded in the factory before they needed to be shipped from Asia in order to get units in stores for the holiday season. I think they were expecting the assessment to be largely a formality, but when it wasn't, we were dropped from the project.
This is bit of sideways to the article.
I really do have to wonder what goes through the mind of someone willingly buying an internet required device they don't urgently need. It really goes to show the absolute unwillingness to consider the broader impact. Not to say average person should have to do that but that is the world we live in.
It also highlights the glorification of computer illiteracy.
A corporate mesh network for hire in a variety of products.
Another excellent argument against just don't connect it to the internet stance on smart things. It really is only a question of cost effectivity, public perception and legislation if forced connections are added to more classes of products(for now I am only aware of cars).