Building AI systems that work is still hard


Even with the support of AI frameworks like TensorFlow or OpenAI, artificial intelligence still requires deep knowledge and understanding compared to a mainstream web developer. If you have built a working prototype, you are probably the smartest guy in the room. Congratulations, you are a member of a very exclusive club.

With Kaggle you can even earn decent money by solving real world projects. All in all it is an excellent position to be in, but is it enough to build a business? You can not change market mechanics after all. From a business perspective, AI is just another implementation for existing problems. Customers do not care about implementations, they care about results. That means you are not settled just by using AI. When the honeymoon is over, you have to deliver value. Long term, only customers count.

And while your customers might not care about AI, VCs do. The press does. A lot. That difference in attention can create a dangerous reality distortion field for startups. But don’t be fooled: Unless you create universal multipurpose AI there is no free lunch: Even if you are the VC’s darling, you have to go the last mile for your customers. So let’s get into the driver’s seat and look how we can prepare for future scenarios.

The mainstream AI train

AI seems to be different from other mega trends like blockchain, IoT, FinTech etc. Sure, its future is highly unpredictable. But that’s true for almost any technology. The difference is that our own value proposition as a human being seems in danger — not only other businesses. Our value as deciders and creatives is on review. That evokes an emotional response. We don’t know how to position ourselves.

There are a very limited number of basic technologies, most of which can be categorized under the umbrella term ‘deep learning’, that form the basis of almost every application out there: convolutional and recurrent neural networks, LSTM, auto-encoders, random forests, gradient boosting and a very few others.

AI offers many other approaches but these core mechanisms have shown to be overwhelmingly successful lately. A majority of researchers believe that progress in AI will come from improvements of these technologies (and not from some fundamentally different approaches). Lets call this “mainstream AI research’ for that reason.

Any real world solution consists of these core algorithms and a non-AI shell to prepare and process data (e.g. data preparation, feature engineering, world modelling). Improvements of the AI part tend to make the non-AI part unnecessary. That’s in the very nature of AI and almost its definition — making problem-specific efforts obsolete. But exactly this non-AI part is often times the real value proposition of AI driven companies. It’s their secret sauce.

Every improvement in AI makes it more likely that this competitive advantage is open-sourced and available to everyone. With disastrous consequences. Like Frederick Jelinek once said : “Every time I fire a linguist, the performance of the speech recognizer goes up”.

Machine learning basically has introduced the next phase of redundancy reduction: Code is reduced to data. Almost all model-based, probability based and rule-based recognition technologies were washed out by the Deep Learning algorithms in the 2010s.

Domain expertise, feature modeling, and hundreds of thousands lines of code now can be beaten with a few hundred lines of scripting (plus a decent amount of data).  As mentioned above: That means that proprietary code is no longer a defensible asset when it’s in the path of the mainstream AI train.

Significant contributions are very rare. Real breakthroughs or new developments, even a new combination of the basic components, is only possible for a very limited number of researchers. This inner circle is much smaller as you might think (it’s certainly less than 100 developers).

Why is that? Maybe it’s rooted in its core algorithm: backpropagation. Nearly every neural network is trained by this method. The simplest form of backpropagation can be formulated in first semester calculus — nothing sophisticated at all (- but no grade school stuff either). In spite of this simplicity — or maybe for that very reason — in more than 50 years of an interesting and colorful history only a few people looked behind the curtain and questioned its main architecture.

If backpropagation would have had the visibility as it has today, we might be 10 years ahead now (computation power aside).

The steps from plain vanilla neural networks of the 70s, to recurrent networks, to LSTM of today were earthquakes for the AI space. And yet it only needs a few dozen lines of code! Generations of students and researchers went through its math, calculated gradient descents, proved its correctness. But finally most of them nodded and by saying “just a form of optimization” they moved on. Analytical understanding is not enough. You need some form of “inventors intuition” to make a difference.

Since it is very rare be on top of research, for 99.9% of all companies a passenger’s seat is all they can get. The core technology is provided by the industry’s major players in open-source toolsets and frameworks. To be on the latest level, proprietary approaches vanish over time. In this sense, the overwhelming majority of all AI companies are consumers of these core products and technologies.

Where are we heading?

AI (and the required data) has been compared to many things: electricity, coal, gold. It shows how eager the tech world is to find patterns or trends. That’s because this knowledge is absolutely essential for hedging your business or your investments against one simple fact. If you build your business in the path of the AI mainstream train, nothing can save you.

Because of the engine that’s already hurtling down the tracks toward business, there are a few scenarios that are important to consider.

In the first, the mainstream AI research train will get significantly slower or has already stopped. This means no more problem classes can be addressed. That means we get out of the train and have to walk the “last mile” for our customers. This would be a big chance for startups because they have the opportunity to build proprietary technology with the chance of creating a sustainable business.

The second scenario has the mainstream train rolling along at at its current clip. Then it is all the more difficult to get out of the way or get off the train. At high speed, domain knowledge of individual approaches are in great danger of being ‘open-sourced’ by the big guys. All the efforts of the past may be worthless. At present, systems like AlphaGo LINK require a very high percentage of proprietary technology apart from standard (“vanilla”) functionality offered by open-source frameworks. I would not be surprised if we see basic scripts with the same capabilities in the very near future. But the “unknown unknown” is the kind of problem class can be solved with the next wave. Autoencodersand attention based systems are promising candidates. No one can image which verticals can be solved by this. Probability: Possible.

The mainstream AI research train will get significantly slower or has already stopped.

In the fourth scenario, the train gains even more speed. Then, finally: “The singularity is near”. Books have been written about it. Billionaires have fought about it. And I will probably write another article about it. The end game here is Artificial General Intelligence. If we achieve this, all bets are off.

Finally, there’s the  black swan scenario. Someone in a garage discovers the next generation of algorithms away from the mainstream. If this lone rider can use it for themselves we might see the first self-made-trillionaire. But where would this come from? I doubt that this could be done out of the blue. It may be a combination of mainstream techniques and abandoned model based algorithms. In the 2010’s the rise of neural networks some once promising approaches (symbolic approaches etc.) lost parts of their research base. The current run on A.I. also revives other, related research fields. It’s becoming difficult to find an ‘unpopular’ technique or algorithm that isn’t already swarming with researchers. Nevertheless, there might be an outsider who finds or revives an approach which changes the game.

Who is winning?

Let’s put all of this together and finally ask the million dollar question. The answer depends not only on the scenarios above, but foremost on who you are. A business’ starting position is a crucial factor in this equation as its resources and existing assets are key to the strategies they’re deploying.

In the AI champions league are a few companies that have deep pockets and can attract critical talent. Since this is a rather ‘endothermic’ process right now you need other sources of income. That limits the players to the well-known Google, Facebook, Microsoft, IBM club. They built huge proprietary systems apart from the status quo, open-source stacks to arrive at new problem classes. A certain amount of time later you will then put this into the next generation of open-source frameworks to build a vivid community.

These players also have existing platforms that lend themselves to train better algorithms. AI might be a megatrend but its application for and by companies in the daily businesses they’ve built is also critical to their success. These platforms: Amazon, Facebook, Google Apps, Netflix, and even Quora use AI to defend and strengthen their core business model. They find ways to better serve their customers by AI but they are aware to keep their core business distinct from the work they’re doing with artificial intelligence (at least publicly).

Some emerging platforms have found ways to adopt this strategy for their own toolsets. These companies found a claim which AI only made possible or monetizable in the first place. One example is the grammar-checker Grammarly.

At first glance you could think of it as a nice add-on that existing vendors can easily build themselves. But there is more. They are building two assets here: a community generated dataset for further quality improvements and more sustainably, an incredibly personalized marketplace for advertising partners.

Then there are the tool-makers. As Mark Twain suggested — Let others dig the gold and stand at the sideline to sell them the shovels. That worked in the past it might work here as well. Providing data, hosting contests, trading talents, educating people. The blueprint for that gas been to find something that every AI aspirant needs (or wants), then charge for it.

Udemy teaches AI courses, and Kaggle initiates AI competitions to help other companies out and let data scientists build their skills. Neither need to build a core competency in AI. Companies also need petabytes of data to be successful. Most of them use supervised learning, so there has to be someone who supervises this.

Finally there  are the companies that have found their niche in AI consulting. Because even on the shoulders of the giants’ open-source frameworks there is still a lot of work to do to.

Companies like Element AI were able to put parts of that extra work into a product and make it into a service. Indeed the recent investment of $102 million makes sure that they have the deep pockets needed to succeed.

There are other companies that are waiting in the wings, those companies that have a targeted artificial intelligence solution that they’re touting to replace an existing business process. However, these companies face challenges on two fronts. Open source projects could be developed to solve the same problem and the existing vendors are investing heavily in more automated solutions to solve the same problems.

The most important factor in the industry is the speed of the mainstream AI research, which happens amongst a very small group of researchers. With little delay, their results are open-sourced in frameworks developed by the AI champion players. The rest of us are either passengers on the artificial intelligence train or obstacles in its path. Ultimately, positioning is everything and the companies that determine their place with the above context in mind, can still reach their desired destination. 

Featured Image: MF3d/iStock

Researchers can now make neighborhood voting predictions from Google Street View images


In a sign that computers will be able to perform image analysis as fluently as text analysis, a group of Stanford-based researchers were able to make accurate predictions about neighborhood voting patterns based on millions of pictures collected from Google Street View, reports The New York Times. While other academic projects have used artificial intelligence to mine Google Street View for socioconomic insights (such as Streetchange), this project is notable because of the vast quantity of images that its AI software processed.

Led by Stanford computer vision scientist Timnit Gebru, the team of researchers used software to analyze 50 million images of street scenes and location data. Their goal was to find data that could be used to predict demographic statistics at the zip code and precinct (which usually contain about 1,000 people) level.

From those images, they were able to glean information, including make and model, about 22 million cars, or 8% of all cars in the country, in 3,000 zip codes and 39,000 voting districts. After cross-referencing that data with information from other sources, including the Census Bureau’s American Community Survey and presidential election voting records, the researchers found that they were able to make accurate predictions about a neighborhood’s income, race, education and voting patterns.

In order to get their AI algorithms to classify cars accurately, the researchers trained it by recruiting hundreds of people from places like Mechanical Turk, as well as car experts, to identify vehicles in a sample of millions of pictures. In the end, their software was able to classify cars in 50 million images in just two weeks, a task the Times said would have taken a human expert 15 years to finish.

In an article published in the Proceedings of the National Academy of Sciences, the team wrote that their technology can supplement the American Community Survey, which costs more than $250 million each year to perform. Since the survey is also labor-intensive, with workers going door to door, that means smaller areas with populations of less than 65,000 are often overlooked. As technology improves, demographic statistics may eventually be updated in real time, though the researchers noted that policymakers will need to be careful to make sure data is collected only at the community level to safeguard individual privacy.

Featured Image: Bloomberg/Getty Images

A conversation with Dean Kamen on the myth of “Eureka!”

In the final episode of Flux podcast for 2017 I sat down with Dean Kamen, engineer, businessman and inventor extraordinaire.

As an undergraduate, Kamen developed the first portable drug delivery device, a wearable infusion pump. Today he spends much of his time running DEKA, his New Hampshire based company, which has been focused on R&D since 1982. DEKA has helped develop the portable dialysis machine, the vascular stent, the iBOT a stair-climbing wheelchair, the Slingshot water purifier and the Sterling generator. Kamen is also behind FIRST, the non-profit behind the successful robotics competitions which last year saw half a million students participating. Most recently he was awarded $300 million and is building up ARMI to focus on regenerative medicine.

These days Kamen may be best known as the inventor of the Segway, that two-wheeled personal transport wonder that launched 1,000 internet memes.

“I think invention is maybe like love,” says Dean. “Everybody wants to have it. Nobody knows what it is.”

He explains why interdisciplinary exposure and freedom to tinker at the edge are so critical to human progress and why the business community, government and technologists need to move from focusing on short-term wins to creating long-term value and solving real global needs. He also shares why he thinks the world is headed for catastrophe and what we can do to ensure that in the race between education and catastrophe, education wins.

AMLG: Lets begin with a day in your life. You have a million day jobs and projects going on. How do you organize your time?

DK: I could try to act like it’s organized and planned and I actually control my environment and my schedule. But none of that would be true. I’ve never had a business plan in my life. I’m unorganized. I don’t schedule much of anything. I react to the crisis of the day when things are going wrong. Or I seize the opportunity of the day when things are going right. I try to connect disparate resources and people that I realize might be powerful and successful in some way at doing what they’re doing, but are unaware of how much impact they can have by short-cutting the development by some other industry or some other company to do it, if they just knew about each other.

DEKA partnered with Coca-Cola to distribute the Slingshot water purifier in the developing world.

For instance you talked about our water project. We spent years building a vapor compression distiller a small portable device that could make clean water out of anything. That was the easy part. It was just physics and thermodynamics. But how are you going to get it all over the world? It’s arguably a medical device because the number one killer of kids worldwide is simply no access to clean water. Medical companies aren’t going to start doing that. They don’t have the resources to go to most of the developing world countries with expensive medical products. But the Coca-Cola Company has the largest global footprint of any company in the history of civilization. So let’s go work with the Coca-Cola company to deliver a critical healthcare need.

AMLG: Tell us about DEKA and your latest research there?

DK: I was recently summoned to the White House where they announced that my little company DEKA was going to receive an $80 million dollar federal grant to be matched by well over $200 million dollars in funding to create a whole new industry in regenerative medicine. Over the last few months we’ve been scrambling to put together an organization that will over the next five years deploy this $300 million dollars such that by the end of it we believe we’ll be making at industrial scale cells, tissues, skin and bone, and whole organs — kidneys, livers, lungs — to meet the desperate needs of people that need organ replacements, for which there will be no reasonable supply of donor organs. All the complications of taking a donor organ typically leaves you needing immunosuppressives for the rest of your life. What if we could literally print pieces of an organ and a scaffold for it, then reintroduce to it induced pluripotent stem cells from the donor, who then becomes the recipient? It’s sort of like I can replace your old worn out part with a new one from the original equipment manufacturer — your mom and dad — because it will have your DNA in it.

AMLG: Most of what DEKA has done so far has been in the mechanical engineering sphere. How did you get into the bio side of things?

DK: When we were asked to bid on it we told the Department of Defense, who’s sponsoring, that we were flattered that they felt we could be a credible partner. But we said, you realize that all the magic goo in these roller bottles and petri dishes by which the world of biology and medicine is creating these miracles is not our expertise. I have some of the best engineers in the world, but we said to them — we don’t do the basic biology. To which their answer was, “we know that. There’s lots of guys that are creating those miracles but they’re going to stay in those petri dishes and roller bottles for another 20 years as all of these researchers win Nobel Prizes and get faculty positions and write white papers. We need somebody that’s on the engineering side that knows how to scale these things up and bring them to where you can meet the needs of 300,000 people right now waiting for organs.”

So we did a search of our own around the country to see what was the state of these miracles and petri dishes. And we realized that if we can bring some engineering discipline to the whole community, we probably can dramatically accelerate the rate at which these miracles leave the world of science and become part of the world of industry to meet the needs of the American public.

AMLG: If you go back in evolutionary history, amphibians were able to regenerate. They then evolved into mammals who lost that ability. Science is finding that some creatures have been able to regain that ability and people are thinking it will be possible for humans. Our livers are already able to grow themselves back. Do you think it will be possible for humans to regain that lizard-like capability to regenerate, and is this any part of what your regenerative medicine group is looking at?

DK: What I can tell you is we now have, in this regenerative medicine cohort that we’ve put together, 26 different world class university medical school affiliated research centers. They’re all on our team in one way or another, and they all have examples of being able to regenerate cells, tissues, entire organs. They’re each screaming ahead towards understanding the mechanism of action by which they can make that more realistic. In fact it now is relatively straightforward to take adult cells from your skin, and through a process turn it into what’s called induced pluripotent stem cells, we can take your adult cell that decided to be an eyelash or a fingernail or hair or liver and turn it back into a cell that when put in the right environment will still be your cell with your DNA. It’s you. But it will start growing itself to become what you happen to need, a new kidney or a new lung or more skin or you name it. The understanding of how to do that and how to control that is leaving the realm of science fiction and becoming the realm of science. We want to turn it into the realm of high-volume, high-scale access for people that need it.

AMLG: With so many disparate projects going on at DEKA, I’m curious how you structure the workplace. Is it an open floor plan? Do hardware and software sit together? Is it top down or bottom up? Do teams set their own direction? How do you organize it all to inspire creativity?

DK: Personally I don’t think whether they’re all in one room or separated makes a big difference. We’re in a bunch of magnificent historic old buildings along the Merrimack River. The Millyard was once the largest single operating industrial complex in the United States. When textiles became a commodity the Millyard suffered and the buildings ended up nearly empty. I moved up there from New York and saw it as a golden opportunity to revive them. I thought if we turned them into a neat space to do R&D it would be a great attraction to young smart people. My office is halfway between Dartmouth and MIT and Harvard. It’s just a good place.

But maybe the real answer to your question, how do you organize people to be creative? I don’t know. I don’t know how to make people creative. But I know what can take that out of them — a big top-down infrastructure that’s well organized. I don’t say that to be pejorative. Society depends on big, well-structured, well-organized things. You like to know when you flip that switch the light will come on. You like to know when you turn that tap that the water coming out is safe. You don’t want to give it another thought. To have organized societies we depend mostly on big systems that do things consistently and reliably. In fact we measure quality by how consistent things are. The trouble is an unintended byproduct of well-organized large organizations is that while they snuff out anything that might be a change, which might be bad, they also snuff out anything that might be a positive surprise. Which is why big organizations always struggle with trying to be efficient, well-organized, scaled up, but also innovative. I don’t have an answer to that, so my solution is don’t ever become big.

I said, I will keep populating my little company with the kind of people that like to live on the edge. That don’t mind failing, picking themselves up and trying again. I said, if I can keep an environment where people can work hard, fail fast, learn from it, get get up and keep moving until we have something that we know is ready for scale, then — instead of building out a whole global organization to meet the needs of something at scale, which would take everybody’s time — take that thing and give it to these big organizations. Who may or may not have a tolerance for the risk and wouldn’t have liked the 10 times we failed and wouldn’t have been able to have people in that environment with the confidence that they can fail and keep going. But they are big. They have a global reach. They are disciplined. They understand scale. So give it to them. Get a little royalty or share in the upside with them. Then use what they give us back as the the fuel for our next project. DEKA has worked for a lot of giant companies. We’ve made a lot of products that have reached scale. But we have nothing in the world that says DEKA on it.

In 1947 Bardeen, Brattain and Shockley invent the transistor, the building block enabling many digital technologies including portable radios, touchtone phones, computer microchips and color television. Also invented at Bell Labs: data networking, the first binary digital computer, solar battery cells, the laser, communications satellites, Unix operating system.

AMLG: So you let them handle the scaling part and focus on the exciting bit, invention and cracking problems. I want to dig into that more. There’s a few institutions I’m fascinated by and wanted your take. I spent a bit of time working with researchers at Lawrence Berkeley National Labs when I did research on the electricity grid a couple of years ago. I would walk around the grounds and there were scientists hula-hooping and wild turkeys roaming and an environment of people doing interesting things. One place I’m fascinated by is Bell Labs. If I had a time machine I would love to vaporize myself back there and appear in about 1947 — 

DK: Shockley.

Claude Shannon demonstrating “Theseus,” a magnetic mouse controlled by an electromechanical relay circuit. The mouse was programmed to search through any new configuration of the maze until it found its target. It was one of the first artificial learning devices of its kind. [Source]\

AMLG: Yes Shockley. For listeners who don’t know, Bell Labs was behind many innovations like lasers, transistors, the silicon solar cell, the computer operating system Unix, feedback. And Claude Shannon the father of Information Theory. One of my favorite stories about him is he would ride his unicycle up and down the hallways, he juggled, and he had a lot of fun side projects like the electrical mouse that navigated through a maze. I guess my point is that all these people were thrown into one building. Chemists, mathematicians, engineers, physicists. They were encouraged to exchange ideas and were given time to pursue their own investigations without concrete goals or deadlines. I’m interested what you take away looking at that? Do you think that can be reproduced or was it a one-time thing, and when you look at corporations who are trying to do this now, the modern day Googles and Google Xs, do you think they’re doing a good job?

DK: The only thing I have in common Shannon is, I ride a unicycle.

AMLG: You do??

DK: But you mentioned a bunch of interesting organizations that I have involvement with. I was on a two-hour board call with the folks at Berkeley night before last. Because Paul Jacobs, the CEO now of Qualcomm, helped them build a whole new piece of their engineering school the Jacobs Institute. The goal of which is to figure out how to make engineering innovation more relevant and exciting, at the level of teaching it. There is so much evidence that schools have been efficient at making isolated courses — electrical engineering, mechanical engineering, and even within electrical engineering you’ve got hardware, software, analog, digital. But most innovation comes at the intersection of technologies and ideas. Schools are good at giving you the basics. I’m not against schools. I love schools, I love education. But schools tend to be about analysis — here’s the situation, here’s F=MA. They’re good at explaining what we found out, what we learned what became a law. That’s analysis. Break everything down so they understand it. Analysis is great. But the opposite of analysis is synthesis. Let’s build something new. Schools aren’t there to teach you how to do synthesis. Some people do it naturally. The inventive type — why can’t I try this way? But that’s not generally taught. And I’m not sure if it is easy to teach it well.

To your point, I also would like to have been a fly on the wall when people like Shockley were walking around. When people were looking at background radiation and wondering what is this stuff? And how big is the universe? By the way Galileo worried about that 400 years ago. Curiosity-based research, to people who are looking at the near term, always seems like a distraction or a waste of time if you’re a business type. But in the end, with hindsight, everything that seemed like it was just somebody off doing something crazy in one generation turns out to be the core of an entire new industry or an entire new field of science and understanding that we have today.

There’s a famous story about a British physicist, Faraday. He put together some of the understanding back then around current carrying wires causing magnetic fields, and conductors moving in magnetic fields causing currents. People like Gauss and Ampere and those guys in the mid-1800s were starting to understand electricity and magnetism and how they related. The story goes, Faraday demonstrated that closing a circuit and putting a current from a battery through a coil of wire with a nail in it — what you and I would call electromagnetism — he was able to pull the needle of a compass off its axis. That was a big deal. The scientific community got excited. The Royal Society back then was excited and got him an audience with the Queen. He showed the Queen. He pulled the switch closed and the magnet moved with nothing in between. This is action at a distance, the magnetic field is pulling this thing. And it moved and supposedly the Queen said, “of what use is this?”

AMLG: Classic.

DK: To which his answer was, “your Highness, of what use is a baby?” Obviously all of modern electricity, magnetism, motors, generators, communication is based on that. In any generation people that are tinkering with something that’s not yet at scale and isn’t yet part of life, in a product or service, people easily say it’s a distraction. But all progress depends on people pushing the edge of what we know, doing things that may not have an obvious application to life as we know it today. But they will literally determine what life will be like in the future.

AMLG: It sounds like tinkering and interdisciplinary exposure is critical to get these crazy outcomes one wouldn’t expect. I wanted to ask about Intellectual Ventures, started in 1999 by Nathan Myhrvold, who ran Microsoft’s research division. He wanted to see whether the kinds of insights that lead to invention could be engineered. So he formed IV, with all these people coming together in a room for “invention sessions.” Do you think if you cram a ton of PhDs in a room to brainstorm you can get a multiplier effect?

DK: You should ask Nathan about this yourself. But he sat in my living room during that year after I spent a little time with him and a bunch of other incredibly creative interesting guys in the early days of that process. Some of his original cohort in fact. They said, why don’t we all do this together, why don’t we use the DEKA model. You should ask him. I’m not trying to flatter myself but —

AMLG: So he copied you?

DK: Well I wouldn’t say copied us. But we were all out having a fun weekend, and they realized DEKA really does have a business model — as students at business school say today — which was to do the inventing and leave the scale to other people. He was fascinated by that. He’s a brilliant guy and surrounded himself with other brilliant people and they said, how can we take that model to the extreme? Let’s get all sorts of smart people in a room and figure out if we can turn invention into a well-defined process. They’ve had some successes with that. In general though invention is a messy proposition. You said, well what if we just put a bunch of PhDs in a room? Being a PhD certainly doesn’t hurt in having deep, well-defined knowledge of a particular subject. That can be useful or at least getting access to people that have that knowledge is useful. But I’m not sure that it’s the PhDs that are likely to come up with the bold new ideas. Because again, education, which is what gets you a PhD, is analysis. It’s a deep, thorough, breaking down of everything we know. Whereas invention is synthesis. In many ways I think it’s the opposite. Some of the best inventors I know didn’t have a deep knowledge of any particular technology. They had a deep understanding of what the need is and what we would today call intuition — because we don’t know at the granular level what intuition is — they’d have an intuition about how to solve the problem and then they might go to PhDs for help in implementation.

I think invention is maybe like love. Everybody wants to have it. Nobody knows what it is. It’s an amorphic process. The public has an overly simplistic view of inventors. They suddenly have this brilliant vision and they go running down the street saying,“eureka I’ve got it!” Invention is an iterative, frustrating process in which you keep finding all the wrong ways to get to where you wanted to go. You back up, try a new route, hit another stumbling block, fall down. Eventually you integrate enough of the ideas that might have should have could have would have worked into something that actually does work. Then the world sees it and think it was a straight line from your idea to that solution. That there was instant clarity. As opposed to this iterative, long struggle.

AMLG: Right — a lot of failures and a lot of squiggly lines on the way to the eureka moment.

DK: Except for me. Everything always happens on time, ahead of schedule, under budget and works on the first shot. Ha.

On February 14th, 1876 Elisha Gray (left) and Alexander Graham Bell (right) both filed telegraphy patents.

AMLG: Haha. One thing I’m curious to ask is about the idea that inventions come in multiples, that people tend to have the same idea all at once. Which points at this foundational, iterative, long process. “The Calculus Wars” describes how Newton and Leibniz both discovered calculus at the same time. Color photography was invented by two Frenchmen at the same time. Five people came up with the steamboat. Nine people came up with the telescope at the same time. Alexander Graham Bell had a simultaneous discovery — Elisha Gray also worked on the telephone and they filed patents on the same day in 1876 for the telephone. Are some inventions inevitable, are they waiting to happen? Are they the product of an intellectual climate of a specific time and place and build up? Are discoveries, rather than being in our heads which is the general perception, rather are they in the air? 

DK: On that subject I happen to have a strong opinion and I think you’re already biased by pointing this out. I unequivocally believe that most invention is the intersection of basic science or basic capabilities that have been smoldering along in development for a long time, maybe for a different application. The world sees them, enough people see them and then it’s only a relatively small increment to put them together in a way that makes this “new invention.” You mentioned all the examples except the airplane. People were starting to understand that heavier than air machines could work — birds work, if their wings aren’t flapping they fall down. Right. We act like it was a shocking thing. Well there are things up there that are just floating like bubbles because they’re lighter than air. Birds fly. But when you calculate how much energy it would take to keep something the size of a human being up there, we’re kind of big and heavy, you’d need gigantic wings. How are you going to power them? But at the same time people were understanding what we now call aerodynamics. The gas engine was being developed.

You talked about Elisha Gray and Graham Bell. Well the telegraph had been around for decades. People had finally figured out, you close a switch and an electric current moves through a wire at the speed of light and if you let it go a mile or two it still goes essentially instantly at the speed of light. Because there’s a current in that wire, it pulls down a switch and you hear a click and we can do that multiple times, and we call that a telegraph. If you just replace the mechanical switches that go click click click — you have to decode them very slowly — if you put a membrane at each end of that set of wires and you let it vibrate in a magnetic field or an electric field, that electrical information will also travel down those wires to the other end. You replace the click click click of a telegraph with “Watson come here I need you.” It was because of the Amperes and the Faradays, all the people who made it clear how electric fields and magnetic fields and electric currents and batteries work, that first made it into a telegraph and eventually made it operate at human voice scale and now of course at Internet speed. All of these things that are “inventions” evolve when the background technologies become well enough understood and well enough available to the inventor community that they become practical tools. Once they’re one of your tools you can build with them. These inventions are the result of capabilities created by core technologies that evolve over time.

AMLG: So it’s the fluency of what’s around us, of our tools as you as you put it. If you’re a teenager these days there’s a lot of resources available to build an on-demand app. A disappearing photo app or a social media app, which is what people are calling innovation. Software is easier to scale and people also see Mark Zuckerberg as an example of success. By contrast, as Ben Horowitz puts it, there’s a hard thing about hard things. You’ve said you think technology is being squandered on quick buck applications and that people are defaulting to the short term. How do we get kids who are glued to their cellphones more interested in science and technology? Perhaps this links to the work you’re doing with FIRST. 

DK: FIRST is a not-for-profit organization. The letters stand for For Inspiration and Recognition of Science and Technology. A lot of people in this country believe we have an education crisis. My mom’s a teacher. You and I both know there’s great teachers. We don’t have an education crisis. We spend more on education than the rest of the world. We have a culture crisis. We have kids that are obsessed and distracted by other things. I said, if we can get kids as passionate about science, technology, engineering and inventing as they are about kicking a ball, bouncing a ball or being on stage. If we can compete with the excitement of sports and entertainment but have the content be tech, we would change the perspective of kids, particularly girls. So I started FIRST as a not-for-profit. It’s now grown to be 50,000 schools, more than a million kids, 140,000 volunteer mentors working with kids, building robots to have fun competing in an after-school activity that rivals any other sport they could participate in. Except every kid on a FIRST team is developing the skills to become a pro.

Founded in 1989 and based in Manchester NH, FIRST is a not-for-profit public charity designed to inspire young people’s interest and participation in science and technology, and to motivate them to pursue education and career opportunities in STEM fields.

One of the reasons I started FIRST was I wanted kids to see how exciting and accessible science, technology, engineering, mathematics, inventing are. Yes they’re frustrating. But the wins you get by finally understanding that idea, or the win that comes when you realize I can build that, I can do that, I can understand that — is so powerful. It hooks people to devote their time and attention to things that will give them career opportunities in a world that will soon have few career opportunities for people that might 100 years ago have been able to do just fine because they had a strong back. Or women that could have done just fine if they wanted to do things by rote and sit at a desk and do what we now call mindless tasks that are better done by a simple machine or a computer. It’s mutually beneficial to society and to kids to keep raising the bar for kids’ passion and understanding of technology.

To your other point — sadly the attention span, not just of little kids that are multiplexing between their phone and their TV and whatever, but the attention span that has been compressed in the business community, who wants their quarterly numbers, or the attention span of government, that doesn’t seem to be ready to make long-term commitments, whether it’s infrastructure, education or basic research. It’s hurting us that people now have a short attention span. They focus on doing projects that by definition are only incremental because all they’re doing is reassessing and reapplying things they know how to do, but slightly modified. This app versus that app. I don’t want to be judgmental about that industry or another industry. But I have to tell you, getting up in the morning to work hard at making the Nth new video game isn’t exciting to me. Whether I could succeed at it or not, whether I’d make money at it or not, isn’t as important to me as working to help a person that has diabetes or needs a liver or a kidney. Or a kid in some far off place whose biggest problem today is they don’t have access to clean water. Frankly, too many kids in this country that do get these powerful tools and are able to develop systems, they take the easy road of developing products that might have a short-term win in it for them but don’t have a long-term value to society and to the world.

AMLG: Speaking of short time versus long term, what do you think of folks like Elon Musk who are becoming heroes to some kids and really thinking of the long term, Mars type expeditions?

DK: Elon’s another guy I’ve known a long time. One thing that distinguishes him is that most inventors that stay in control of their technology to create a whole business, since it takes a long time typically, do it in one industry. Or they do it once. You create a company, that takes a long time. Elon went from things like PayPal, which is primarily a software to, Let’s go make electric cars. Let’s go do Solar City. Let’s go build rocket ships. He has to be given credit for the fact that he has gone from industry to industry. He’s bet the farm with a lot of his own resources, which a lot of people aren’t willing to do. Elon has demonstrated the willingness to put his money where his mouth is. Over and over again. You’ve got to give him credit for that.

AMLG: I want to go back to you. Since we’re talking about what inspires kids I’d love to hear what were you like as a kid?

DK: I’ll probably disappoint you with that. I was not exceptional at anything. Other than I hated school. I keep saying I love education. I loved education back then. I’ve always loved learning things. It’s what makes us uniquely human. We can benefit by standing on the shoulders of the giants before us. We can learn Archimedes Principle and he’s been dead for 2200 years! So I’ve always loved education. But school seemed like a firehose. They throw everything at you out of context — this morning spelling. Then phonics. Then arithmetic. Then science. Then math. I would hear something that would be amazing and I’d be thinking about it, then the school and the class is off to the next subject and I’m stuck thinking about this amazing thing I just heard which I didn’t really understand. Then I’d be accused of not paying attention. Well I’m sorry, it’s not that I’m not paying attention, I should have said. But I wasn’t wise enough back then to say I just don’t really understand this. Your threshold for thinking I understand it is that I can regurgitate to you what you told me. But that’s being familiar with something not understanding it. As I got older I realized school wasn’t there to give you an understanding. It was there to give you this hodgepodge of stuff. I’m a very slow reader. I’m a very slow learner. I’m a pretty slow thinker, but I stay focused on something until I think about it in enough different ways that I would look at you and say, I understand it now. I do think I understand Newton’s laws and Maxwell’s equations and the Second Law of Thermodynamics. But I don’t just know them —

AMLG: You’ve internalized them.

DK: Right. I have an intuition about them. It takes years and years of thinking about them and rethinking about them to get there, and that doesn’t make you a good student. So I didn’t like school. I always felt dumb because I couldn’t just regurgitate things. I always felt like it was passing me by and I couldn’t keep up.

AMLG: Did you invent stuff as a kid? What was your first invention?

DK: My first invention never made it to the big time. When I was young I shared my bedroom with my older brother. I was probably five, he was eight. It turned out my mom still had this peculiar idea that every morning we should make our beds. I remember saying to her, well listen it’s my bed and I don’t mind that it’s ruffled up. In fact when I get back into it tonight I’m just going to ruffle it up again and while I’m not in it I might as well leave it pre-ruffled. It’ll be better. At the end of most of those discussions what a mom says is, well you are still going to make your bed. And you’d say why and she’d say because I’m your mother.

I realized that when I made my bed with my brother it was easy. He would stand at one side, I’d stand at the other and we’d say pull and the covers would go straight. Then we’d run to the bottom edge of the bed, do the same thing and be done. But when you have to make your own bed, as we all know, you can pull a string but you can’t push it. I was small, I was barely the height of the bed and running all the way around the bed from one side to the other, and if you pull too much you have to go back. I remember seeing my mom put clothes on a clothes line from the second story of the house. Because it had a pulley, 20 feet away on a pole in the back yard, she could get it all the way across the back yard from one position. I went out, took the pulley down, attached it to the corner of the bed, tied knots in the covers, put the strings from the pulleys so I could stand at one corner of the bed, pull on the cover, pull on the ropes to the pulleys and… puff I could make my bed as easily myself as if I had my brother there to help me.

AMLG: So it was an automatic bed maker?

DK: It was an automatic bed-maker. My mom was not impressed, a) She wanted her clothesline back and b) I had knotted it up and damaged it—

AMLG: And you had cheated on the bed making.

DK: She wasn’t happy. So that product never made it to the big scale.

I realized that when I made my bed with my brother it was easy. He would stand at one side, I’d stand at the other and we’d say pull and the covers would go straight. Then we’d run to the bottom edge of the bed, do the same thing and be done. But when you have to make your own bed, as we all know, you can pull a string but you can’t push it. I was small, I was barely the height of the bed and running all the way around the bed from one side to the other, and if you pull too much you have to go back. I remember seeing my mom put clothes on a clothes line from the second story of the house. Because it had a pulley, 20 feet away on a pole in the back yard, she could get it all the way across the back yard from one position. I went out, took the pulley down, attached it to the corner of the bed, tied knots in the covers, put the strings from the pulleys so I could stand at one corner of the bed, pull on the cover, pull on the ropes to the pulleys and… puff I could make my bed as easily myself as if I had my brother there to help me.

AMLG: So it was an automatic bed maker?

DK: It was an automatic bed-maker. My mom was not impressed, a) She wanted her clothesline back and b) I had knotted it up and damaged it—

A look back at Uber’s hellish year

Waymo drops most patent claims in Uber lawsuit

In July, there were a few big breaks in the case between Waymo and Uber over self-driving car technology. As a result, the scope of the case started to come into focus as both companies began preparing for a trial set to begin in October.

Waymo, the self-driving technology arm of Google parent Alphabet, filed the lawsuit in February, alleging theft of trade secrets that Uber planned to use in its autonomous vehicles. The on-going case centers around engineer Anthony Levandowski, who Waymo claims stole 14,000 documents before leaving the company and founding Otto, a self-driving trucking company, which Uber later acquired.

Waymo decided to drop its claims on U.S. Patent Nos. 8,836,922, 9,285,464 and 9,086,273, noting that they were related to an earlier version of Uber’s autonomous lidar design nicknamed “Spider” that the company was no longer using. The remaining patent claim targets a newer version of lidar technology called Fiji, which is still in use by Uber.

In addition to the patent news, U.S. District Judge William Alsup asked Waymo to narrow its theft of trade secret claims from more than 100 down to 10 that could be put in front of a jury.

Over the course of the last several months, the judge had urged both parties to simplify the scope of the case so that each could be adequately prepared to argue the merits of the strongest claims post-discovery.

Waymo, for its part, continues to argue that Uber was aware of the confidential information Levandowski took before leaving Google.

Smart lock maker Otto suspends operations


Otto showed the world its digital lock in August. Four months later, the company has suspended operations. Hardware is hard. It’s a cliche for a reason.

The company made the decision just ahead of the holidays, a fact that founder and CEO Sam Jadallah recently made public with a lengthy Medium post now pinned to the top of the startup’s site. The extended survey of the Bay Area company’s short life is punctuated with the pithy title, “So Close,” a nod to the spitting distance the startup came to actually bringing a product to market.

In a conversation over the weekend, Jadallah told TechCrunch that the company’s lock made it as far as the manufacturing process, and is currently sitting in a warehouse, unable to be sold by a hardware startup that is effectively no longer operating. How does a company get so close to the finish line without being able to take that final step?

The executive lays much of that out in his own explainer — a post he considers a sort of cautionary tale for the volatility of the Valley. The long and short of it is that the company was about to be acquired by someone with a lot more resources and experience in bringing a product to market, only to have the rug apparently pulled out at the last minute.

“You’re not in charge of your own destiny, and the margin for error is a lot smaller,” Jadallah told TechCrunch. “Building a really exciting hardware product needs a ton of resources, and is probably best inside of a bigger company. Frankly, that’s part of the reason I was excited about the acquisition. I knew it would take us out of the cyclical venture capital market and put us inside a company that knew how to make and ship products.”

The executive wouldn’t name the interested party during the call, but Otto was almost certainly made hopeful by the recent acquisition of August Home by Assa Abloy, the world’s largest lock manufacturer. The big players have no doubt that there’s plenty of room to grow in the space, and the connected home category shows no apparent signs of slowing. NPD reported a 43 percent growth in smart home sales in 2017. Security is a big piece of that puzzle, but there’s still plenty to unlock on that front.

Otto thought it had found the key, though the company’s product garnered a fair amount of pushback at launch. Sure,it followed in Nest’s footsteps and brought some former Apple employees on board for the creation of what is, by all accounts one nice looking door lock. But even in the age of the $1,000 iPhone, a $699 smart lock is a tough pill to swallow. If the smart lock is still searching for its mainstream moment, was a flagship-phone-priced device really going to be the product to put it over the edge?

Jadallah certainly believed so, as apparently, did the unnamed company that came within days of acquiring Otto. And while the  buyers apparently never gave a reason for their decision to pull out, the executive says that the product’s price was never a concern.

“They knew about the price before the first meeting, and they are very smart people,” he says. “This isn’t the story of an ambitious product that didn’t have a market. I was convinced that we had priced it the right way for the product, and we knew that the technology that we had innovated was something that we could use in different ways at other price point.”

In fact, he adds in a followup email, the acquiring company was apparently convinced that it could sell the product for even more in certain markets. Of course, that’s all a bit of a moot point now. While what remains of the company is attempting to figure out what to do with all of those smart locks currently populating a warehouse somewhere, there’s currently no one around to actually sell them.

The seemingly imminent acquisition meant the company had no plan B.  “The life of the startup is a binary thing,” Jadallah says. “To go from what could be an incredible high to crushing low in a matter of hours is what we do.”

Earlier this month, the company’s Facebook page was still promoting the product with winking reference to the new Star Wars film in a video that asked users to “Unlock the dark side.” Two weeks later, another startup has, for most intents and purposes, gone dark.

Germany starts enforcing hate speech law

Man walking in front of a mural of Facebook usersImage copyright Getty Images
Image caption Facebook is one of the social media companies affected by NetzDG

Germany is set to start enforcing a law that demands social media sites move quickly to remove hate speech, fake news and illegal material.

Sites that do not remove "obviously illegal" posts could face fines of up to 50m euro (£44.3m).

The law gives the networks 24 hours to act after they have been told about law-breaking material.

Social networks and media sites with more than two million members will fall under the law's provisions.

Facebook, Twitter and YouTube will be the law's main focus but it is also likely to be applied to Reddit, Tumblr and Russian social network VK. Other sites such as Vimeo and Flickr could also be caught up in its provisions.

Act faster

The Netzwerkdurchsetzungsgesetz (NetzDG) law was passed at the end of June 2017 and came into force in early October.

The social networks were given until the end of 2017 to prepare themselves for the arrival of NetzDG.

The call to police social media sites more effectively arose after several high-profile cases in which fake news and racist material was being spread via the German arms of prominent social media firms.

Germany's justice ministry said it would make forms available on its site, which concerned citizens could use to report content that violates NetzDG or has not been taken down in time.

As well as forcing social media firms to act quickly, NetzDG requires them to put in place a comprehensive complaints structure so that posts can quickly be reported to staff.

Image copyright Getty Images
Image caption Twitter recently updated the guidelines it follows when tackling hate speech

Most material will have to be removed within 24 hours but networks will have a week to act on "complex cases".

Facebook has reportedly recruited several hundred staff in Germany to deal with reports about content that breaks the NetzDG and to do a better job of monitoring what people post.

The law has been controversial in Germany with some saying it could lead to inadvertent censorship or curtail free speech.

The German law is the most extreme example of efforts by governments and regulators to rein in social media firms. Many of them have come under much greater scrutiny this year as information about how they are used to spread propaganda and other sensitive material has come to light.

In the UK, politicians have been sharply critical of social sites, calling them a "disgrace" and saying they were "shamefully far" from doing a good job of policing hate speech and other offensive content.

The European Commission also published guidelines calling on social media sites to act faster to spot and remove hateful content.