William J Plut

Silicon Valley veteran. Multiple exits.

[William J Plut @WillPlut]

@_amirbar This is half correct.

Incorrect - Robot training data is specific to the hardware - motors, joint types, gearing, mass distribution, joint lengths, +10 more variables that undermine sharing data between different hardware.

Correct - OpenAI can't be trusted by robot companies

[William J Plut @WillPlut]

The most common mistake in robotics is rushing design. 99% of robots failed in design, but that won’t be apparent and inescapable until reality reveals the failures.

[William J Plut @WillPlut]

The two main ingredients in every metal robot are metal and bulljunk.

[William J Plut @WillPlut]

AI and Robotics are not synonymous.

If someone conflates the two, it really shows that they understand neither.

If someone brags about a robot’s AI is, then they are likely distracting from a crappy, unsafe, useless, and/or overpriced robot.

[William J Plut @WillPlut]

Robots-as-a-Service is a euphemism for “this robot costs too much” and “we can’t design a useful robot that is cost-effective”

[William J Plut @WillPlut]

“Personality” in a robot is what its maker boasts when the robot can’t do anything useful.

If you want personality, get a dog.

Better yet, get one from a shelter - THAT’S personality.

[William J Plut @WillPlut]

Gearing is the necessary evil of robotics.

[William J Plut @WillPlut]

ps. adding a layer of clothing on a 15-pound aluminum arm robot is akin to adding a sock onto an aluminum baseball bat.

Calling that 'robot safety' is indefensible bulljunk.

[William J Plut @WillPlut]

Robotic Perspective:

1) Swinging an aluminum baseball bat and hitting someone in the face is obviously dangerous. The bat weighs ~2 pounds.

2) An aluminum robot arm weighs 13-15 pounds. (F~1/2mv^2)

130-pound metal robots with metal arms are not safe and shouldn't be in homes.

[William J Plut @WillPlut]

@tlbtlbtlb Not 'every' - just the ones you know about ;)

I'll respectfully defer to your investment experience. The companies building today tho can't be eventual winners when their mech tech is a dead end.

Robotic design 101: its easier to start from scratch than improve a fatal design.

[William J Plut @WillPlut]

@JimRosenthal4 Overcoming filter R is different for large/small designs.

For large designs, its easy to raise A to drop R (larger surface area filters, pleating, etc) for higher Q

For small designs, A is tight and increasing Q is costly (R~Q^2). So, to increase Q, add a 2nd fan for better P/Q

[William J Plut @WillPlut]

@brosseau_lisa Healthcare orgs blaming HCW infections on community exposure is like blaming a massive spike in firefighter deaths on backyard BBQs.

Reality: Healthcare orgs gave them surgical masks, which were known to not work. Don't send firefighters into forest fires wearing hockey pants.

[William J Plut @WillPlut]

@cleanairkits @CleanAirStars @sri_srikrishna @JimRosenthal4 @robwiss Depends on the application. Total audible noise depends on how many fans a design uses. You use 6-10 fans in parallel to get through a MERV filter (high R). I use one fan and a better filter (lower R), which is quieter than 6-10 fans at 30dB. More portable, lower power, etc.

[William J Plut @WillPlut]

@kprather88 He’s 100% wrong. I was in Toronto when the Precautionary Principle was used for the first time. It’s why HCWs in 2003 Toronto had full PPE to defend all three modes of transmission.

[William J Plut @WillPlut]

@sri_srikrishna Healthcare worker PPE

The Crisis Strategies just let hospitals buy the cheapest options which we know does not protect HCWs.

https://t.co/pxUrByBDnh

[William J Plut @WillPlut]

A thread and timeline on how a cascade of CDC errors destroyed American PPE standards for Health Care Workers, and how this has contributed to a looming national crisis.

(An expansion of this earlier point.)

1/

https://t.co/UZi4GeDdFG

[William J Plut @WillPlut]

@angievassallo @AbraarKaran Well said.

ICUs though are special places that turn any virus (regardless of main mode) into all three modes.

The danger in characterizing as one mode: what if one gets the mode wrong? (cough, surgical masks for covid)

Side note: single-mode stockpiling is inherently gnarly.

[William J Plut @WillPlut]

The US puts waaaay too much faith in vaccines. It hasn't worked so far.

[Eric Topol @EricTopol]

I don't see any evidence for how an annual Covid shot will provide durable protection (current ones wane after 4-6 months) without better vaccines, such as variant-proof, improved nanoparticles, and nasal
https://t.co/uytgskKKN9