Moore’s Law and startup deflation on The Future, This Week

This transcript is the product of an artificial intelligence - human collaboration. Any mistakes are the human's fault. (Just saying. Accurately yours, AI)

Disclaimer We'd like to advise that the following program may contain real news, occasional philosophy and ideas that may offend some listeners.

Intro This is The Future, This Week on Sydney Business Insights. I'm Sandra Peter, and I'm Kai Riemer. Every week we get together and look at the news of the week. We discuss technology, the future of business, the weird and the wonderful and things that change the world. Okay, let's start. Let's start!

Sandra Today on The Future, This Week: Moore's law no more, and Silicon Valley's startup deflation. I'm Sandra Peter, I'm the Director of Sydney Business Insights.

Kai I'm Kai Riemer, professor at the Business School and leader of the Digital Disruption Research Group. What are we talking about?

Sandra Before we decide on any stories, we have to settle this once and for all. Is it JIF or GIF?

Kai Well, Wired magazine had a story about peanut butter.

Sandra Yes, Giphy and the peanut butter brand Jif reignited the argument over whether it's JIF or GIF, and they're settling it once and for all. And we want to open the podcast with that. It is...

Kai GIF.

Sandra GIF. And GIFs are animated looping images, hard G pronunciation as per one of the peanut butter jars.

Kai And Jif, J I F, as in soft G pronunciation is creamy peanut butter and we would show you the image, but it is a podcast, but they have two jars next to each other. G I F and J I F, and that should settle it, done.

Sandra Let's look at stories, though. We need to decide what we're doing today.

Kai So we could talk about satellites, but we've done this a couple of weeks ago. But there's an update here, SpaceX, who overnight became the world's largest active satellite constellation operator, right, have 240 plus satellites up in orbit and wants to put more than 40000 up there. We discussed how it interferes with optical astronomers and that it can be seen from Earth and all the rest of it. There was actually an article in ScienceAlert, which makes the point that there's yet another problem with the idea of putting that many satellites up there. And that is hackers. Because many of these satellites are built from standard IoT components, there's now a real worry that they can actually be hacked. Some of them have propulsion mechanisms.

Sandra So you could fly them into other stuff.

Kai Yes. And we talked about the Kessler Effect. So there's now a real worry that malicious entities could not only hack into whatever communication they are doing, but they could actually move these satellites and crashed them into other stuff, so.

Sandra We'll put the entire episode we did on the many satellites that are now in space, we'll put that in the shownotes.

Kai So we'll take this as an update.

Sandra We could talk about moths.

Kai Moths.

Sandra Moths. Scientists have released the first genetically engineered moths, that is on the back of CRISPR-type technology. An international team of researchers has created a variant of the diamond back moth, which is a huge, huge pest. It's estimated that it costs about five billion dollars a year in losses, and it's largely resistant to insecticides and it's also...

Kai What does it do?

Sandra It eats your brassicas.

Kai So it's the very hungry caterpillar moths.

Sandra Yes, it is. The diamondback moth is the most destructive insect pest to all the brassica crops throughout the world. And we've been dealing with this with chemicals which of course, stay in the ground, but also kill other insects other than diamondback moths.

Kai So I take it the ideas that scientists will want to eradicate this moth by introducing some deficient gene that is dominant and then will basically do away with it?

Sandra The idea is that the researchers engineered this moths so that when the males of that strain mate with the females, the female offspring would die during the caterpillar stage. However, the male offspring would survive and mate with the remaining female moths. And on and on we go, until the overall population is reduced. And of course, as with any research that involves releasing genetically engineered organisms into the wild, there are all the concerns about ongoing effects.

Kai So fingers crossed. I mean, we in Australia have a proud history of messing with the environment by releasing introduced species to fix one problem, then we create another, cane toads case in point. But we could also talk about privacy and anonymisation in data. So there's an interesting article in Vice. It's not a long article and it really only makes one point, but I thought I'd mention it because it's not a story that carries very far. But the idea being that a lot of companies say to you, 'Okay. You can give us your data because we anonymise it'. And that's fine. As far as this one, dataset goes. But the article makes the point that hackers have a long memory and usually malicious hackers, they obtain many different datasets from different sources. And even though each one of them might be anonymised, given that we anonymise all of them in different ways, by combining them, researchers have shown that you can, for the most part de-anonymise and therefore identify users quite clearly in those datasets. So there's a real worry that anonymisation doesn't really work when it comes to hackers that accumulate data.

Sandra There are actually two other stories I think we should talk about today. And one of them has to do with Moore's Law.

Kai And the other one is the one with startups in Silicon Valley?

Sandra Let's do this.

Kai So, Sandra, what happened in the future this week?

Sandra Our first story comes from the MIT Tech Review, and it's titled "We're Not Prepared for the End of Moore's Law".

Kai Yet the end is nigh.

Sandra We think this is an important story to cover because for the last couple of years, there have been stories on and off about the end of Moore's Law, then not the end of Moore's Law. But it is actually quite an important force mechanism that drove much of what happens in Silicon Valley for a very long time.

Kai I think it's pretty well known. But after reading this article, it was actually quite a few things or details that I didn't know about. So we thought we might share some of those insights with you and we got to thinking about it. And it's actually a really cool phenomenon.

Sandra First of all, let's unpack for our listeners what Moore's Law actually is and where it comes from. 'Cause it's quite important for understanding where it's going and whether or not it's coming to an end.

Kai So it's named after the person who actually first discovered the correlation that underpins this law, Gordon E. Moore. Who in 1965 in Electronics magazine published an article which was descriptively titled "Cramming More Components onto Integrated Circuits", where he predicted that in the years to come, every year, the number of components on an integrated circuit would double.

Sandra And 10 years later, in 1975, Moore looked back and saw that his observations were more or less correct. And he revised what has since been known as Moore's Law, the doubling of transistors on a chip every two years.

Kai And it's important to note that this was not a technological, but an entirely economic argument.

Sandra Back in 1965, Moore was making an economic argument, which was that the costs of the components that would go onto this chip was inversely proportional to the number of components.

Kai So basically, the more transistors you put on, the cheaper they get, which drives the number of transistors that you can put into a computer and that drives computing power. So that has widely been known as Moore's Law and it has more or less held ever since.

Sandra So just to put this into perspective, because quite often when we say, okay, exponential curves and they double and they get twice as cheap and so on, people find it quite difficult to visualise what that would mean.

Kai Yeah, we've previously had examples about, you know, envisioning exponential growth. But you have another example here for us.

Sandra A good example comes from Intel's CEO who said, let's think about what Moore's Law would look like for a car.

Kai So a 1965 beetle goes about, what, 120 kilometres an hour, tops, maybe not even that fast.

Sandra So, Megan, how fast do you think it would go today if it progressed at the same pace as Moore's Law?

Megan Ahhhh, 160.

Sandra 300 thousand miles per hour. How much do you think it would cost?

Megan $20?

Sandra Four cents. And it would also get you two million miles per gallon of gas, which I think is a pretty good deal.

Megan I'll take it!

Kai We'll take two.

Sandra Excellent.

Kai And so interestingly, that was an economic argument that was based on old fashioned transistors. Surprisingly, it held even as we move to microchips and integrated circuits and putting transistors into silicon. And over a number of waves of technological progress, which even surprised Moore himself.

Sandra Here's Gordon Moore reflecting on 50 years of Moore's Law in a clip for Intel.

Audio - Gordon Moore My real objective was to get the idea across: we have the technology that's going to make electronics cheap. But I didn't expect the thousandfold increase in complexity to be very accurate.

Kai So from 65000 transistors in 1975, we have come a long way and we're now packing almost 50 billion transistors into the most advanced chips today.

Sandra To be fair, today there are more than 100 variables involved in keeping Moore's Law alive. But nonetheless, over the past 50 plus years, these chips have become a general purpose technology that has enabled pretty much all the technological advances we're now so used to.

Kai And quite interestingly, in 1965, Gordon Moore envisioned what such a progress in computing technology would enable us to do in the world.

Audio - Gordon Moore Integrated circuits will lead to such wonders as home computers or at least terminals connected to central computers, automatic controls for automobiles and personal portable communications equipment. The electronic wristwatch needs only a display to be feasible today.

Kai So that's pretty much the Apple Watch, a computer on your wrist.

Sandra Yep, so he saw in this technology the potential for laptops, for smartphones, for mobile devices, for autopilot on cars, and even, as you said, smartwatches.

Kai But what we want to highlight here is that Moore's Law is not strictly a prediction of what would just happen. Rather, Moore's Law has become a mechanism to actually drive and coordinate activity in this industry. And again, here's Gordon Moore reflecting on what his law has become over time.

Audio - Gordon Moore The message I was trying to get across was that integrated circuits were the road to less expensive electronics. It really evolved from being a measure of what goes on in the industry to something that more or less drives the industry.

Sandra So indeed, Moore's Law became a coordination mechanism for not only the hardware industry, but the software industry as well, were working to the expectation that they could double the power of chips every two years.

Kai So not only would software companies and computer companies and producers of devices know that computing technology would evolve at that pace, Intel and chip manufacturers would also know that because the users of their chips were working to this timeline, there was actually a market to make the investments to develop technology to double computing power in that timeframe.

Sandra And indeed, many economists have actually credited the progress that has been made due to this coordination and the ability of information technology and integrated circuits to develop at this pace with all the productivity growth that we've seen since the mid-seventies until today.

Kai Didn't you have Alvy Ray Smith, one of the founders of Pixar, talk about this on the podcast?

Sandra Yeah, he was talking about the fact that they were developing business plans even before the technology existed because they knew that Moore's Law would enable them to have this five or six years down the line.

Kai Let's hear from Alvy.

Audio - Alvy Ray Smith So Ed and I had this idea, and we wrote up the business plan. And by the way, it wasn't to make movies because we knew Moore's Law really well and we knew that our computers, we needed five more years of Moore's Law development, just standard Silicon Valley development was going up by an order of magnitude, a factor of 10 every five years. And we were still a factor of 10 shy of having cost-effective computers.

Sandra And we'll put the link in the shownotes so you can hear the entire story of how Moore's Law helped shape the future of computer animation and the entire history of digital colour through one of Pixar's co-founders and lifelong innovators, Dr Alvy Ray Smith.

Kai And so because it was more than a prediction, but a governing mechanism for this industry, the industry collectively has put a lot of effort and basic research in physics into keeping Moore's Law alive. And in the past, like in 2005, there was one wave where it looked like we might no longer make this progress, there was another breakthrough in chip technology, and we were able to yet again shrink the size of the transistors that go onto a chip. And so the industry has found ways to keep this Moore's Law alive up until a few years ago, when things started slowing down,.

Sandra Things started slowing down, not really because people are not managing to have the technological breakthrough, and as we mentioned before, there is more than 100 variables going into keeping Moore's Law alive. So Intel, for instance, has eight thousand hardware engineers working at keeping this alive. So it's not for lack of trying, but because, physics. And when you start cramming more and more things into a smaller and smaller chip, obviously heat becomes a real issue.

Kai And also, as you make things smaller, you get into funny quantum territory.

Sandra Yep, so you start getting quantum effects because the things are so small now that physics work differently at that size.

Kai And so in recent years, not only has progress slowed down, the effort and the money that goes into making progress in research has gone up significantly. And creating the kind of manufacturing facilities to producing these chips has also become way, way more expensive. To the extent that where previously in 2002 there were 25 companies around the world producing chips at the high and level, there's only three today that can actually do this. And by the way, I had a bit of a nerd out, I did the maths on where we should be in terms of transistor numbers had Moore's Law held, we'd be at about 272 billion transistors and we're only at about 50. So we're slipping, but the point is that this slowing down will now progress because, physics.

Sandra So Moore's Law is not coming to some sudden abrupt end, but it is in a slow decline. And while that's been touted for the last few years, as we've mentioned, it is significant to understand what that means for the overall industry.

Kai And not just the chip industry, but anyone who depends on microchips, in producing all the kind of widgets and gadgets that make their way into every part of our lives.

Sandra So just to get this out of the way. Of course, everyone says quantum will extend Moore's Law, but quantum is nowhere near being able to practically extend Moore's Law at this point. So, yes, obviously, quantum might be part of extending Moore's Law.

Kai Yeah, but interestingly, I mean, this Moore's Law has been mythologised to the extent that, you know, we just talk about computing power now, and that with quantum, we see these charts at tech presentations where people just say 'quantum', and that means that Moore's Law will go on indefinitely. But we must not forget that quantum computing is an entirely different way of computing, it's not just an extension of what we have today.

Sandra So let's put quantum conversation aside now. So what's left in store for Moore's Law now?

Kai So on the back of the observation that it becomes more and more expensive to keep Moore's Law alive for a few more years, there's certain developments on how you can actually come up with more computing power in different ways. One is software optimisation, going to different ways of programming, putting more effort in what Apple does, for example, doing hardware and software integrated. You can optimise for that, but it is a different ballgame or specialised chips.

Sandra And this is one interesting observation with Moore's Law, that whilst Moore's Law in general as the underpinning for general purpose technology, computers in general, while that's been slowing down and it is in decline, there is one area where there are still some advances and where there is still a lot of money being invested for research, and that is in these specialised chips. So it's no longer the answer for the entire industry, but there are those with sufficient money and resources, companies like Google or Microsoft, Baidu, who are designing their own particular chips for specific applications. And artificial intelligence is probably the most common example. These things rely on graphic chips, on graphic processing units.

Kai Yeah, and NVIDIA, who used to be one of the largest graphics processing companies has now built an entire new business off the back of specialised AI chips. But these are no longer chips that can be used universally in any kind of device. So we're no longer talking what the article calls 'lifting all boats' with Moore's Law, creating almost like a democratised technology that any manufacturer can put into their devices, IoT and what have you. But more and more very specific chips that can do very specific things, but that require from companies to put their own R&D money and their own software integration into these efforts.

Sandra And let's not forget, no one really knows where the next breakthrough will come from and how much money it will cost. We also want to specify here that whilst advances in software unspecialised architectures are still being made and artificial intelligence is a good example of that, the physical limitations, heat, quantum effects still apply. So that has the risk of putting some genuine constraints down the line. And that might be in a few years, it might be in 10 years, it might be in 20 years. It might put some real constraints on how we develop and the progress that we make in artificial intelligence.

Kai So people like Alvy back in the day who could just count on Moore's Law, getting them to where they needed to be with their ideas, that might no longer hold. So that might put the brakes on some of the progress that is being envisioned in machine learning and deep learning and working with these gigantic datasets, which needs quite a bit of computing power.

Sandra This is the perfect segue actually to our next story, because our next story comes from The New York Times and it talks about the startup boom deflating and how technology is humbled. So the slowdown of Moore's Law might be part of the startup world deflating. But there are actually a number of other reasons why Silicon Valley has slowed down. So the article makes the point that for many years now, Silicon Valley has been seen as an engine for innovation and an engine for job creation and growth, as companies were growing so quickly that they couldn't find people to hire fast enough. We've increasingly seen layoffs and startups failing and companies not managing to live up to the expectations that the world had of them of delivering jobs, delivering profits.

Kai And while it's normal that startups fail, most of them fail. This is companies that have already built to a certain stage. They've been through quite a few funding rounds. They're quite well known, and yet they are shedding jobs. Such as logistics startup Flexport, or DNA testing company 23andme, Firefox maker Mozilla, and quite a few others, with some of them failing outright, such as robot pizza startup Zune, who wanted to use quite expensive robots to automate the making of pizza.

Sandra And they lost over 340-some million dollars in the process. Well, actually, the Japanese conglomerate SoftBank lost the money, so Zume's now pivoting to a delivery company. But all this came on the back of other changes in Silicon Valley. Casper Sleep, which was touted by everybody to be the new 'Nike of sleep', flopped when it went public. We've seen Uber and Lyft losing billions of dollars and having really disappointing IPOs last year.

Kai We mentioned WeWork who basically failed to go public and have since slashed about 80 percent of their valuation, realising that, no, they're not a tech company, but a real estate business.

Sandra We'll put the WeWork episode in the shownotes. And such failures have been echoed around the world. So, for instance, Oyo, the Indian hospitality startup, has undergone huge layoffs and has lost a lot of money. We've had Lime, the scooter company, withdrawing from 12 cities around the world. So there's this general background of uncertainty punctuated by mini moments of panic.

Kai And according to the article, there's spreadsheets circulating on Twitter with people who are out of a job and who are looking for work. And there's a little bit of pessimism spreading in Silicon Valley off the back of this.

Sandra So whilst many people are quite critical of the industry and becoming quite sceptical, there has been the ongoing fear that we might be facing another dot com bubble that could burst any minute. So we thought we'd have a look at this story and see exactly what's going on.

Kai And here we want to just remind people about what this dot com bubble back in 2000 was. And the bursting of this bubble is an event that is etched into the memory of anyone in the tech industry. So there is a fear that Silicon Valley might go through another long period of disillusionment and underinvestment. But we want to remind people that what happened back in the day has less in common with what's happening today than we would normally believe.

Sandra So whilst on the one hand we had a period of boom in the number of startups, we had all these inflated expectations about what these startups could deliver, and a belief in a new economy where everything will be Internet-based and where this will give companies the competitive advantage.

Kai So on the one hand, there was a misunderstanding about the role of the Internet, not as a thing that gives you a competitive advantage, but as one that actually takes away advantage because it's a general purpose technology, anyone can use it. So it fuelled a whole slew of companies that all of a sudden came into the market. Everyone wanted to be, not the Uber of something, but the eBay of something. Lots of electronic marketplaces that would bring buyers and sellers together in all kinds of different industries, which led to a high degree of competition. It was very hard for anyone to make a profit, and while that might be similar to today, where we have lots of food delivery companies, lots of companies that all do the same thing, there was one marked difference back in the day and that was that the bubble had reached the stage of rampant IPO's. So while today a lot of the money that flows into startups comes from VC companies, and that happened there back in the day. The dot com bubble was largely fuelled by lots and lots of companies making IPOs, everyday people, so-called mom and pop investors would sign up with multiple banks to participate in IPOs which became sort of like a lottery. You would get an allocation of shares and they would go up two-fold, four-fold on the first day, you would sell them, and people would just pour lots and lots of money into these IPOs of companies that were largely all the same. The bubble grew off the back of this to far larger proportions than we're seeing today.

Sandra So today, not only are we not seeing that extent of everyday people investing in these companies, but we're also seeing far fewer IPOs than last time around. We mentioned WeWork pulling off the market, Airbnb, for instance, and DoorDash, both companies that could have gone public and are expected to go public are still stalling. But there's also the fact that the startup scene actually looks quite different and a lot of the innovation is being absorbed or being created under the umbrella of the big tech companies, a phenomena that we've discussed before. We spoke about corporate inequality and the way that most of the gains in the industry, but also most of the innovation is vested within these large companies like Baidu, Alibaba, Tencent in China or Microsoft, Google, Apple in the US.

Kai So what we're seeing is quite a different phenomenon. Not only do we have big tech on the one hand, on the other hand, we're actually seeing what might be a healthy correction before we go to this IPO bubble stage that we've seen previously. And maybe this is just the market washing out some of the more bullshit ideas that have sprung up in Silicon Valley like robot pizza-making, which don't actually have a place in the real world.

Sandra So we think overall whilst there is indeed a slowdown, this might just be the healthier way in which the industry slows down. And also coming at a time where tech as an industry itself is asking quite fundamental questions about what it is and what it is for. We want to reference here the other conversations going on around the impact of technology, on democracy, on capitalism, on free speech, on privacy. Let's not forget that the tech industry started out as people making things, people building things, but they have now become social engineers. So there is a moment of reckoning in the industry, not only with governments trying to respond to what is going on, but also with markets trying to respond to a industry that has reached a certain level of maturity.

Kai And while we're not going to discuss it in detail, we will put in the show notes another article from The Atlantic, which basically makes the point that Silicon Valley has hit a sort of a midlife crisis where pessimism is spreading off the back of these negative implications of social media and that there is no quick tech fix for these problems. But also making the argument that in many ways with unionisation at Kickstarter, and we mentioned this last week, or employee walkouts at Google, companies in Silicon Valley are actually undergoing some sort of reform that over time might make them a little bit more like companies in other places of the world.

Sandra So it remains to be seen what new economic models will emerge or how Silicon Valley will deal with growth or sustainability or democracy, for that matter.

Kai And here's something that someone could research off the back of our first story, "Moore, no more: the implications of computing slowdown, and innovation in the tech sector".

Sandra And that's all we have time for today.

Kai See you soon.

Sandra On The Future...

Kai Next week.

Sandra This week?

Kai [Yes, but next week.

Sandra On The Future, This Week. Next week. Thanks for listening.

Kai Thanks for listening.

Outro This was The Future, This Week, made possible by the Sydney Business Insights team and members of the Digital Disruption Research Group. And every week right here with us, our sound editor Megan Wedge, who makes us sound good and keeps us honest. Our theme music is composed and played live from a set of garden hoses by Linsey Pollak. You can subscribe to this podcast on iTunes, Stitcher, Spotify, YouTube, SoundCloud or wherever you get your podcasts. You can follow us online on Flipboard, Twitter or sbi.sydney.edu.au. If you have any news that you want us to discuss, please send them to sbi@sydney.edu.au.

Sandra Costs of, of the components that would go on to this chip was inversely proportional to the number of, of um, was inversely proportional to the n...

Kai Proportionate.

Sandra Yes. Was invershely.

Kai To the number of.

Sandra Was inversely proportionate. Proportional.

Kai Proportionate in English, population in Germany.

Sandra I'm sure it's proportional. Inversely proportionate. Inversely proportional. And when we find it, we figure out which one. Inversely proportionate, inversely...

Kai Proportional. You were right.

Sandra I'm sorry, what?

Kai Sandra was right. Kai was wrong. See, I'm not afraid to say it.

Sandra Are you recording?

Megan Mm, hmm.

Sandra Excellent. Can I have it so I can play it back?

Kai and Megan As a ring tone.

Kai Come on.

Sandra And that's all we have time for this week. Thank you very much.