It's also an illustration of what went wrong with the "internet of things" concept. Sensors are easy today. Actuators are hard. Most "IoT" things can't do much.
Try to buy a home power window. They exist. They're an exotic luxury home item. Even controllable home HVAC dampers are rare. It makes good sense to have a system where windows, fans, dampers, heaters, and compressors are all coordinated to maximize comfort at minimum cost, automatically. Will NeXT sell you that? No. It takes too much installation.
You might see that at a well run convention hotel. They have to keep customers happy, yet many of their big rooms are empty much of the time. So they'll have CO, CO2, temperature, humidity, and motion sensors tied to a control system that senses what the room needs for the current people load.
There's real IoT, but it's under commercial building automation.
It is rare enough to find people with good electronics and software skills.
Once upon a time I worked for a manager who could only bring software instincts to any problem. He was dismissive, even, of ME problems. But not every problem can be solved in two-week sprints and pushed over LTE to the fleet. (Actually, if you want a good start-up idea, find a way to push a field update to a 5 kilo aluminum casting over LTE -- I'd invest in that.)
But here is the thing -- unless you can synchronize the release of software + electronics + mechanics, you don't have a robot. That means design, validation, component tracking, field updates, maintenance. You always need to take a step back and ask yourself what instincts to bring to a problem, and if you are weak in one/some of those areas, learn who's instincts you can trust and bring them in.
I am presumptuous enough to believe this applies to IOT mechatronics as well as robots. (Actually robots need more than SW+EE+ME, but this rant is already long enough.)
So in the good case this degree might broaden your horizon and teaches you that you don't know enough and in the bad case it might make you overconfident in even more areas while having even less knowledge than from a more focused degree.
Students should understand that what they learn at uni is as far from real world product development as possible. Uni does not have incompetent coworkers, unreasonable, conflicting or entirely missing requirements, scaling issues, vendors trying to screw you, procurement processes trying their best to prevent you from doing anything, faulty materials, etc. Knowing a bunch of mechanic formulas or software development patterns does not an engineer make.
MIT students have complained, after graduation, of finding companies that worked like soap operas.
At Uni you are taught many things including "things NOT to do"; some seeming quite obvious. I thought:
"Well, I shouldn't encounter those "things NOT to do" in the Real World TM, because professionals in the workplace have had 10 years to learn these lessons."
Ah, the naivety of youth...
Professionals are just people who convince someone to pay them for what they do. What they do doesn't have to include competence.
- They're all running at a rather leisurely pace, about 1 mask per second. That's very slow for a simple low-cost product. Compare textile machinery, progressive stamping presses, bottle making machines, and printing equipment. That's probably because demand for these things wasn't high until recently, and nobody designed a high-speed machine.
- The bottleneck seems to be attaching the elastic loops. The rest of the process is all roll to roll and could easily be 3x as fast. Is there some better way to do that?
- Can you get wider rolls of the source materials and build machines that make more masks at a time?
- At the output end, most machines just dump the finished masks into a bin. That loses order and count.
They then need a separate packaging operation. They come out of the machine in a consistent position and orientation, which can be exploited. A better machine would put 10 or 20 masks put into a printed poly bag and heat-seal the bag. Then it would drop the bag into a waiting cardboard carton. When the carton is full, switch to a new carton. The filled carton gets a shipping label and rolls down a roller conveyor headed for shipping.
Compare , an N95 mask making machine, and , a paper cup making machine. The paper cup machine is doing a tougher job at 5x the speed. It has a flying paster, like a big printing press, so you can change rolls without a shutdown. The cups are shot out through a pneumatic tube to the cup stacking and packaging machine. That's why paper cups are so cheap.
I really wish my post-college engineering experience could be anything like that. I miss making useful things.
On the other hand, her first asp.net page was demonstrated by Steve Ballmer at CES, so she did advance in the world.
Unfortunately, the power assymetry between managers and engineers frequently manifests itself as you have described:
-- managers trying to understand everything through the lens of their past experience,
-- engineers trying to conform to managers (ie. using manager's language to formulate problems and solutions)
-- managers dismissing/prioritizing down areas, ideas, problems that they are not familiar, promoting/prioritizing up where they feel they have good understanding,
-- engineers responding by doing the same (ie. focusing on where they feel they will be rewarded by manager).
-- managers hiring/promoting based on their assessment of the candidates/employee competency in the area of expertise of the manager, because it is easier to rate candidate/employee competency if it is in your area.
-- managers failing to improve in areas where they are completely new. Maybe because people in general tend to want to improve in areas they already know. Maybe because they don't want to be seen as noobs making silly mistakes with experts around, and so trying shun entire area as less important so that the discussion can focus on where they are comfortable talking to experts.
-- so on...
So where a team works on a product that requires software/electronics/mechanics, the team, regardless of its initial composition, will over time get skewed in the area of expertise of its top manager.
Same goes for startup -- duo of guys who know software and electronics will inevitably look for products in the areas they know best. Not being able to quickly rate how difficult something is, mechanically, is definitely scaring a lot of people away.
I also noticed people dealing with software and electronics rarely mingle with mechanics/material experts which I think might also explain a little bit why there aren't too many enterprises that join those areas well.
From a SWE perspective, Robotics Engineers need to know physics and math than the average SWE learns.
Interestingly almost all the successful robot projects I've worked on seem to follow a waterfall process. You keep working on it till it works. After its ready don;t change anything.
I thought waterfall was to plan it out in detail, then do all the work according to the plan despite realizing that the plan is wrong, then finally, after disbanding the development team, test it to see if it works?
That's obviously a caricature, but if you have the feedback loop to see if it works while you're working on it, you might also call it agile.
Someone who came up through software intuitively knows that while "it is only a 3 line change in each of 4 files", making the change in master is only the beginning. Also on deck are: back-porting and validating in the 4 release branches still running on the 4 hardware versions in the field, and pushing the update reliably, oh, and it changes the log format so our log monitor query needs to be updated and validated.
EE -- yes, it is just changing out one connector -- but we need the latching version and only two latching variants are available, and one requires moving a mounting hole which would require chassis rework, the other is expensive and has a 12 week lead time... and in any case we need to update the test fixtures. And we need to keep some of the old test fixtures around for the old rev, and while we could save some money by reworking some of our old test fixtures, who has time to write and validate the engineering change order for the test fixture rework?
ME -- yes it is only moving two mounting holes, but both the chassis and the skins need to move, so we either have to rework the fleet to accept new skins, or do two sets of expensive tooling for the skins, or do some kludge adapter that allows new skins on old chassis, but we need to vibration test that.
Intuition and instinct are about understanding the implications and the unarticulated risks buried with decisions.
Having worked in this space, the failures usually come from lopsided teams. Too many startups try to treat these products as technology products with a side of building materials, assuming that the technology is the hardest part.
In reality, the technology part is easy. The most difficult part is manufacturing and shipping reliable mechanisms at low cost. It can be done, but the mechanisms and mechanical parts need to be treated as the primary product, not an afterthought.
A mass market automated pill dispenser you can just dump your prescriptions in seems like an obviously viable product. But, building something that’s easy enough to use, reliable enough to be safe, and cheap enough to be affordable is a pipe dream.
However, pills are of wildly different shapes and sizes, which invalidates most of the seemingly obvious sorting solutions. Prescriptions can also call for multiple pills in the same day.
I think the main weakness is that for a few folks, medicine changes would be difficult because they often are mid-week or mid-month and for some, they would really need their medicines re-packaged. I'm not sure this is as big of a problem in nursing homes and whatnot because the nurse is dispensing the meds, while some folks at home have bad memories or the beginnings of undiagnosed dementia and just won't remember.
Though the number of old folks is getting larger, so maybe that's a market by itself.
I know for a fact that my parents and my wife's parents would all love this.
My mother was a nurse, so she handles it better than most, but she still spends an hour a week on it.
Never mind that an internet connected, custom-engineered juice press was a bad idea in the first place. It was still out-manufactured by simple tools like a garlic press, or a little metal thing you twist your fruit on top of. Or a blender.
You have to look in the manual to add or remove a code, but it remembers at least a dozen codes. It doesn't talk to a smartphone. It doesn't have a network connection. It doesn't store a log of openings, and it doesn't violate your privacy.
What it does is solve the two biggest problems with door locks: not having your key, and getting a key to a friend so that they can get into your house.
But I wonder how it’d fare if someone like the lockpicking lawyer reviewed it’s mechanical lock design? Is it actually a good lock?
I'm not convinced it matters as long as it seems, to the layperson, to be a lock of average quality. No one really wants a lock to feel insecure if it is held in the hand, for example. And sure, if tests say you can just hit it and open it, you'll pass. But as long as it doesn't have obvious defaults... It is good.
Most locks really just keep out crimes of opportunity. Few folks have locks that will keep out folks that would destroy your front door or break your windows and few are real challenges to lockpicks. And the shouldn't be challenges to lockpicks! Otherwise, how would a locksmith do their work? It doesn't matter, though. Most folks simply aren't lockpicks - and a good number of folks that can* pick locks well do not have desires to break into homes.
There's also a perception that software is more valuable than hardware, which can affect the workplace culture in a negative way.
That should be my niche (if it's small enough to be called one), but I've worked firmly in software since graduating a few years ago, and I'm increasingly worried that I'm unhirable for anything else, even if I brushed up with some hobby electronics at home, despite it being electronics that got me into CS.
Is there any truth to that? Would I have to start over in a grad role for anyone to hire me for hardware? (I'm not actually looking, I just don't like the thought that the road's closed, my next job has to be software.)
One day I opened it and it was just a cutout of the floor, directly over the washing machine in the basement below. I'm not sure if that's lazy or genius, but it doesn't get much easier than that I guess.
And to be fair, this house was built in 1918, and I think part of the purpose was to have their maids/servants wash them. Out of sight and all. This house also had doors sectioning off the kitchen and made sure the basement had its own entrance.
Really it’s just a laundry chute. If the laundry room is directly under you, why make it fancier than that?
We have kids and a cat and none have had trouble with it. It's not half as dangerous as and has a lower attractive-nuisance factor than any set of stairs.
I remember crawling up and down it all the time as a kid while playing hide and seek or whatever. In fact it had a flaw of a weak latch. Sometimes when you closed the bathroom door, The laundry chute door would fall open, blocking the bathroom door from opening. When this happened the only way to resolve it is to send a small child up through the laundry chute to close its door.
See also: the film Home Alone.
I'd never considered the child aspect, but that may explain why it had a lock on it(the turn up, push/pull, turn down kind, like on public restroom doors, unsure what they're called). Also, why it was so heavy. I wish I'd taken pics of it. From memory, it was probably 8"x8", about 2" thick. A small child could fit in it, but would probably have difficulty opening it.
If you want to something really "genius" look up razor blade wall banks.
The only good solution which works is moving it to the garage, but it's not possible in every house unfortunately.
That explains! I've wondered why people in movies/series would keep washing machines in the kitchen.
Around here in medium size houses like mine the washing machine is typically placed in the same room as one of the toilets and the room is referred to as the washing room (vaskerom).
Usually (flats/small houses aside) they'll be in the scullery (adjoining the kitchen) though, not actually in the 'cooking area'.
He said a lot of developers don't want the expense that goes into building them. And a lot of owners don't want the maintenance expense. They get clogged up more often than most people realize.
That said, when I lived in Seattle last decade, my building was new and had a trash chute. It got clogged about once a month. You'd see the maintenance guys with big long poles trying to shove the stuff through.
A little of column A and a little of column B. Collectivism is sort of self-reinforcing, but there are also legal mechanisms that keep people honest.
The problem with those little hatches is that they create a breeding ground for rats and cockroaches.
They also serve as a fantastic means for idiots to grind your building's trash disposal system to a halt - when said idiots toss oversized items into them. (Christmas trees were a popular failure mode in my apartment building. How people got them into the trash chutes was beyond me.)
CNC retrofits (even when successful) tend to be never ending projects. This is partly because the kind of person who retrofits a machine is seldom satisfied with it.
But also one tends to end up replacing the whole machine over time - the motion and way the machine gets used is different. The "wear cycle" is accelerated and new machine properties become important. Retrofits also are much more challenging to support than factory-supplied systems and tend to "frankenstein" very quickly.
IOT windows might start making real sense as an aspect of a whole building design where they can be fit for purpose, designed for expected wear and usage and work harmoniously with other building systems.
A truly native IOT window will involve the frame, the glass, the opener, maybe the shades -- and the rest of the building. We probably also need standardised "window sockets" around the frames, with standards for power and data connections to make installation, replacement and upgrade practical.
There are two primary reasons for the infinite refit:
1) No definite project for the CNC
If you have a definite project, you have a deadline and want that project done. The project also has a monetary implication so you have a budget.
I bought a Formlabs Form 2 for this reason. I didn't want to dork around with 3D printers--I needed to print parts. And it worked, we were printing parts within 3 hours of unboxing the machine.
2) Too stringent budgetary constraints
A lot of people doing CNC refits are scrounging parts from EBay, swap meets, etc. Consequently, the project doesn't have a definite plan--it proceeds in fits and starts when a critical number of pieces have been found.
I’m probably ordering a Formlabs Form 2 for work in a couple of weeks because of some of their specialty resins, but did want to point out that around $1000USD can get you a workhorse if a printer!
Another option that I’m also looking at is the Prusa Mini. We have a bunch of small parts to print, and getting an army of smaller printers to work in parallel looks like a significant speed improvement.
I've got two in my house! Of course they weren't wired to anything, and one of them had its wires hidden behind drywall (and the accessible one seems to be broken, but I'd have to disassemble the ductwork to figure out why). Thankfully, the functional one is actually pretty useful, and I was able to hack it up so call for cold/heat closes off the basement, preventing the basement from turning into a freezer. Of course, if Nest would give us an API (again), I could also modulate the cooling if the basement gets too cold.
They do basically what you describe... open and close in rooms depending on temperature in the room, automatically. You can customize preferences for each room, and have a schedule for the day.
Actuators are expensive--not hard.
Home automation is lousy because it does not scale in software.
I can't even get my phone to stop correcting "duck."
(via Daring Fireball)
When are we going to collectively unpack the fact the Media Lab/Berkman Center project has been a complete failure?
I worked at these places and mingled on the outskirts, and for a long time really believed all the bullshit they were spouting at the time - that the internet was a positive force, if we just connected enough people, surely their "stories" and "empathy" would "revolutionize" the world, etc., etc..
They were all wrong.
Digital, networked technology seemed exciting for a brief and beautiful moment, but has turned into nothing more than a global system of surveillance, advertising, propaganda and skinnerboxing, far worse than any television ever was - forget about your "global library" dreams.
And yet - they persist, despite their ever increasing irrelevance. Spineless hucksters like a Zittrain still get wheeled out occasionally for their take on some hot new trend in Wired magaize, they still get private and government grant money, and, of course, they still keep taking on hapless graduate students.
Just when will we call a spade a spade?
EDIT: This little tirade made me go back and re-read some of the Cluetrain Manifesto - I feel almost ashamed of myself to reread something like "markets are conversations" and remember thinking it was ever good, important, or even valid argument. I now see this as evil thought.
I don’t know, if you’re basically saying your stylized, hyperbolic, generalizing, imprecise, inaccurate discourse is better than theirs just because it’s negative and theirs is positive... I would rather have the positive?
The material conditions of people alive in the west today are significantly worse than they were in 1999, largely as the result of the ascendance of the digital platforms and economic arrangements that organizations like MIT and Berkman were taking money to promote.
Ordinary people today are more lonely, overwhelmed, polarized and impoverished. The "new" economics they were promoting - gig economies, crowdfunding, "conversational" marketing - have only resulted in a new form of serfdom, upward movement of capital from the poor to the rich, and a completely collapse of dignity in public life.
In fact, I'll even go a step further and suggest that what we needed was _more_ negativity. Techno-positivity and lack of critical thought is exactly how we got in this mess.
All the things you mentioned were serious problems before just in a different way. Advertising was horrible when I watched broadcast TV. I spent probably 2 hours each day looking at commercials. Propoganda was ubiquitous and there wasn't even a place you could point out it was happening. The gig economy jobs back then were just different jobs with fewer options, like pumping gas or minimum wage at McDonald's or greeting people at Walmart. And now online discussion allows people to be more open about mental health and the anonymity let's you talk about relationship or depression problems with others.
It's not perfect, but it's a hell of a lot better than what life was like before. And one thing I agree with you on, is that it isn't because of the Media Lab or Berkman. It's because of a lot of people making slow incremental progress throughout the entire tech stack.
If not for Internet, I'd be a completely different and I suspect much more limited person (I'm from EE). Internet provided me access to role models in my areas of interest who showed me how high the sky can be in our industry (if not for that, I'd be some depressed J2EE developer/Project Manager in a local telco or something). People in the US can just have in-person access to such role models (e.g. by getting a job at some cool company, doing a PhD at strong US technical school etc.), while in backwater areas of the world, if not for Internet, you would not even be aware of this whole cool universe.
That said, I deeply disagree with your conclusion. I think, barring politics or pandemics of course, the world is objectively better due to the digital revolution. It's not only a positive force, but it's an overall positive one.
Hyper Globalization, the Post Colonial re-emergence of Asia, and the forgone peace dividends of Forever Wars are not due to technology.
MIT & The Labs, for all their shortcomings laid bare by many in these comments, are contained within a larger society experiencing larger forces at play on par with the early industrial revolution where urbanized mind numbing factory exploitation existed but also eventually led to an empowered middle class with more leisure time, disposable income, and upward mobility than at any time before.
Perhaps the last 20+ years of technical progress efficiencies were unjustly captured by return on investment capital instead of by increasing everyone else's leisure time but your average western person who no longer enjoys a 1999 quality of life still has a whole lot to be thankful for even if they choose to enjoy the hard fought for luxury of taking it all for granted.
What about Guitar Hero? Are those guys, Media Lab grad students who spent a decade after graduation until they found hard earned success... did they create a “new serfdom?”
The Berkman Center houses the historic DoJ tech antitrust lawyers. Do you think antitrust won’t play a role in reversing “the movement of capital from the poor to the rich?”
Anyway, there are a lot of people who rail on the academy or whatever. I suppose your opinion, however stylized, is as good as anyone else’s.
Turns out what people in the third world need isn't donated laptops, it's liberation from an economic system which keeps them deliberately impoverished. They need food, vaccines, unions, debt jubilees and nationalization of natural resources.
The MML/Berkman ideology is literally that laptops bring wealth. In reality, wealth brings laptops.
There has been a naïve belief that the Internet would be an all-positive force, which it is not. For instance, we used to have a constant lack of information, now there is information overflow. Content consumption used to be synchronous (people watched the same broadcasts, live), now content consumption is asynchronous.
The problem is that our old tools and concepts to navigate the world around us don't work in this new era anymore, not that the development in itself is inherently bad. Real problems have been solved.
Wikipedia notes that he “...resigned from his position as director of the MIT Center for Civic Media, in protest of the Media Lab's involvement with Epstein.”
He's landed on his feet very well. He has a temporary position at Columbia this fall, and then will be starting a tenured job at UMass Amherst next year.
If the mods have an issue with this comment I can take it down
Remember when they chased Aaron Swartz to commit suicide and never gave him a civil disobedience award?
Remember their extremely close links to Jeffrey Epstein?
Sorry but MIT media Lab is absolute trash and an embarrassment to the MIT name.
And Swartz’s death was caused by JSTOR not the Media Lab FYI
-- Lawrence Lessig, "Prosecutor as Bully"
Huh? What in JSTOR's behavior  do you think caused Swartz's death?
They pushed for aggressive chasing down of an individual for stealing some articles that were largely publicly funded as I recall. This escalated to the FBI and led to a laundry list of charges that scared him into suicide. Not great for some research we should have just had available for free after already paying to fund it once
JSTOR actually does a ton of useful work scanning old journal articles. Someone has to pay for this.
Step one of not whitewashing history is to be accurate about it. You can't say that the Media Lab is an embarrassment to the MIT name when one of your pieces of evidence is that the rest of MIT - not the Media Lab - did something unconscionable. You can't hold organizations to account if your account is wrong.
And that's the danger with unrelated controversies and tangents - all of a sudden we need to have a discussion about the substance of that controversy and tangent. The angrier you are about something, the more important - and harder - it is to target your anger appropriately. Anger is a great tool. Make sure it is hitting what you intend for it to hit.
(In the interest of disclosure: I am an MIT alum who has never been particularly proud of my alma mater in general; I have no tie to the Media Lab.)
Re: the Disobedience Award. I did a google search to find the speech they gave where they discuss why they didn't give it to him. https://www.media.mit.edu/posts/reflections-on-the-disobedie....
I'm glad they recognized his principled and disobedient activism, but there's no acknowledgement about MIT's role in his death, and ultimately they chose not to give him the award.
That's fine, but that's a different argument from the one being made, that the Media Lab is "absolute trash and an embarrassment to the MIT name."
MIT arrested him for breaking into a closet to access the network. I've never heard anything about them suing him. JSTOR did some kind of settlement agreement. It was the feds prosecuting him for wire fraud etc that undoubtedly led to the suicide.
Recently I heard from some fresh MIT grads that a lot of that is now pushed to outsourcing and same cookie-cutter approach.
Some people point to Richard Stallman having a mattress in his office at MIT as some sort of sinister sign. This, I should think, proves otherwise.
That first one is going to be a challenge.
With no details on how it beeps I'm now anxious for beeping of a device many miles away
I want to put a 120mm fan in a small opening next to it that'll turn on when it's more than 3C cooler outside, and more than 24C inside.
Seems like an Arduino would do it.
I don't think it's overkill within the context of his full work. Based on my brief skimming, he's coupling the window with a personal device to modulate his environment through reinforcement learning.
Not checked on paper (especially the first) but on principle both should work and also would be extremely cheap.
Two analog temp switches will be far more reliable, easier, and adjustable.
(I get the fun with an Arduino, but it is hard to beat century old tech)
- Comfort is a function of multiple factors, not just the dry bulb temperature difference. It includes occupant metabolism, insulation from clothing, radiant heat exchange, air speed, and humidity. This is trivial to solve in the case of a single-person office (just give them control of the window), but more complex once you have multiple occupants with different needs, who are in different locations in the office. I wonder if there's value in using reinforcement learning to customize natural ventilation to optimize the comfort of multiple occupants.
- Predicting the temperature the next day, and using a fan to draw in cold night time air into your HVAC ducts when a hot morning/afternoon is expected. This way you can have a reservoir off cool air you can access the next day.
- Once you have multiple rooms, and a tall building, there's multiple effects that are worth attempting to coordinate. For example you could coordinate the ventilation of the building air with the stack effect and wind-driven air through the windows, or coordinate the cross-ventilation horizontally through multiple rooms. This would not just modulate temperatures, but also satisfy indoor air quality needs traditionally achieved through the HVAC.
Typically a lot of fine-tuning of natural ventilation isn't done because the actual energy/comfort impact for one of these ideas is small relative to complexity of building the system, and the large uncertainty associated with the dynamics of airflow. But my general hypothesis is that with the integration of cheap prediction or RL methods, we can reduce that uncertainty which would permit a more long-tail approach where we can combine multiple small improvements.
He published his thesis 2010, which aligns with when Ethan arrived at his office, and if you watch the video at that link, you can see the automated window control system at 1:47.
(Also the video is kind of hilarious and awesome in its own right).
And I'd prefer a mechanized window, myself.
(Но впрочем песня не о нем, а о любви.)