Ask HN: Are paradigm shifting breakthroughs rarer or is it a bias in my eyes?

The creation of mRNA vaccines for covid was pretty impressive.

I'm particularly excited about VR computing (plug: https://www.simulavr.com) fundamentally changing the way we work and get things done (as opposed to just changing the way we game and entertain ourselves).

Your larger point that non-incremental progress seems to be slowing down was at one point popularized by Peter Thiel and Tyler Cowen in the early 2010s. I not only agree, this was the fundamental thing that made me want to get into technology startups and build something. I think malaise over slow tech process is a great motivation to dive into tech and make the world a better place.

Commercially viable quantum computation immediately comes to mind.

I am by no means an expert, although if I was younger and still in school it would be my area of focus.

Speaking from my personal fields of work, AI (computer vision & generative), AR, and VR, I'd say we've made tremendous progress since 2011.

I'd be really curious to hear someone in those fields justify thinking otherwise.

I'd dare to say the progress in AI from 1981-2011 (Watson, CNN's, Chess) was less than the progress from 2011-2021 (StyleGAN, GPT-3, AlphaGO).

As for VR, if you had told someone in 2011 that there would even be a mainstream company making VR in 2021 they might not believe you. Instead, many major companies are building VR and AR glasses, and the leading market one has inside-out tracking, hand tracking, done from an onboard processor, for only the inflation adjusted cost of a Nintendo 3DS at that time.

> Any specific tech or breakthrough you are personally waiting for coming to fruition ?

Honestly, no. When I look back 40 years when there was little tech, and think about the future with more tech... I'm quite comfortable where we are now. We are in a decent middle ground where we have vast improvements vs. my childhood. But we also are a bit overwhelmed - society is still navigating the current tech and its impacts to our selves, our governments, our culture, and figuring out how to integrate tech with the whole of humanity. I think slowing down for a while to get all that sorted out is just fine before we make any new large moves.

I am most excited by the conjunction of speech recognition, natural language understanding, language modeling all coming together to create solid conversational AI interfaces.

There's a lot of things I could get done a lot faster if I could converse with a slightly dumb ML algo the way I would talk to a person, even just for getting tasks done, etc.

Easy looking back across history at all the various paradigm shifts simultaneously to wonder why there isn’t one happening now, but the collapsed time of history isn’t the same as living in real time.

I find that technology’s exponential change tapers off and looks more like an S over time, but after the flattening another S starts with a new exponential curve. Zoom out enough and it looks like steps to a staircase.

All of these are overlapping and have their own time scales.

Graphene, for example, is manufactured and used today—but not for the things it was originally hyped for.

mRNA tech has been around for decades, but clearly in its next step of exponential growth—not just in the vaccines but the “programmable” versions many companies are exploring currently.

> However exponential tech that takes us to something like 'singularity' does not seem to come up any more. Most growth is incremental and we hype every single Lab output with rarely seeing commercial breakthroughs. I was personally always excited about material science breakthroughs using graphene etc. but seems like it is difficult to manufacture at scale even today.

I think at least a decent portion of this feeling can be explained by selection bias and connectivity of the internet.

When you look backwards, all you see are successes. Nobody remembers the innumerable failures that just didn't work out. Now, with a vastly connected internet, you'll hear about more potential technologies than ever before. Research organizations have stepped up their game to publicize and promote their nascent technologies, because it helps obtain funding. And funding is needed to figure out if and how tech can move from a lab into commercial viability.

So I don't think it's all doom and gloom, just that we have unprecedented access to information, and the present always feels slower than the past.

I think WWII and the Apollo project were catalysts for a lot of innovations that produced big shifts. Those are a distant memory, now.

On top of that, currently we have our best and brightest minds trying to figure out how get people to click on ads.

I'm guessing it's a combination of bias, realism, and a change in how things come to market.

On Realism. Looking back on my thinking about "breakthroughs", a lot of the things that were presented with a lot of hype turned out to be duds. The media played up a lot of things as amazing that just never made much of an impact. Now I have a more realistic view of things because I have the experience of seeing "breakthroughs" that were just not breakthroughs. I have more of an "I'll believe it when it's real" mentality which means that the number of breakthroughs I see is lower since I only see the real ones and not the stuff that gets hyped up and never really happens.

On coming to market. I think that today, something goes from idea to market very quickly. The iPhone's 6-months from introduction to market is positively slow by today's standards. By the time something can get hype, you're already using it so it doesn't feel like a breakthrough. It's just a part of life.

> exponential tech that takes us to something like 'singularity' does not seem to come up any more

Just as flying cars doesn't come up so much anymore. Sometimes we have ideas that simply remain as ideas (for the most part). Sure, flying cars do come up, but it's rare. Likewise, we also have plenty of movies about The Singularity.

> Any specific tech or breakthrough you are personally waiting for coming to fruition?

mRNA tech. Seriously, in the past two years we've seen a huge revolution in what we're going to be able to do. The vaccines are wonderful, but they're the tip of the iceberg. But if mRNA tech comes out with an HIV vaccine, treatments for lots of diseases, etc. over the next 5-10 years, we'll probably react with "yea, but that's old-hat. I had their Covid vaccine 5-10 years ago."

A really great Apple Watch. They're very good now, but I would love a watch that would make me ditch my phone. The problem is that it'll never feel like a breakthrough because it'll just be a decade or two of refining the idea as more and more people buy smartwatches. It won't feel like a breakthrough because there are too many people working on it already and because it's already a profitable idea.

The iPhone felt like a breakthrough because all the big money had ignored the space. They were content to just keep offering the same old stuff with minor tweaks.

I think one of the big shifts is in how money is approaching tech investment.

No one wants to get scooped on the Next Big Thing and so lots of money is chasing it, even before the tech is ready (and often finding ways to monetize things along the way). I think that decades ago, companies would ignore things that they thought would have lower margins or eat into established businesses. Xerox decided against computers, IBM thought that software was a crappy business so they let Microsoft be their OS, every car company decided that LiIon batteries were too expensive for cars, etc. Today, companies will keep dumping products on the market in the hope that they won't get scooped - and VCs are willing to fund small companies.

For example, Amazon never wants to be scooped. When Apple launched the iTunes Music Store, it really hurt their CD business. So they brought the Kindle to market fast so that they could become the e-book store. They brought the Echo/Alexa to market fast and have pushed it in all sorts of places - from car add-ons to glasses - even if it isn't profitable yet because they want to own the space. But it also means that by the time something becomes ubiquitous, it seems old. 15 years from now, you could be walking down the street and say "Alexa, take a picture" and some barely-noticeable device on you takes a picture of what you're seeing. But that isn't transformative. You've had Alexa for 20 years.

Breakthrough is a feeling and I think that the way things are coming to market...

I think there’s a lot of biotech development underway that will make our world look fundamentally different in the future. We’ve seen this a bit with the mRNA covid vaccine. I’m excited about the prospect of novel cancer treatments, crispr and its potential to treat genetic defects, protein folding, etc. Maybe these types of techniques can continue to benefit from even incremental progress in computer technology. I feel similarly about other fields — that progress there is just waiting on today’s computing frontier to be applied to them. This all may be an incorrect view, however. My only experience with the biotech I mentioned has been through press, which can overhype.

Innovations and breakthroughs are best recognized in hindsight.

It is very hard to predict which combination of advancements are greater than the sum of their parts while simultaneously making sense in their surrounding economic and social contexts.

I have the feeling that more and more exponential tech advancements are taking place right in front of our eyes. But that the public is somehow actively looking away.

Take a look at self driving. Watch some videos from 5 years ago and from last month.

Take a look at GPT-3. I am pretty sure a lot of articles are already written by NNs these days. Put into the name of any stock into Google News and you will see a flood of articles "Why XYZ has risen/tanked today" and "Should you invest $100k into XYZ?" which probably are written by AI with some human curation. Nothing like this existed a few years ago.

Look at how "money of the people" is now a thing via crypto. Money that can be digitally transferred and not be printed by a government. This was just a dream 10 years ago.

And have you looked into DAOs? These are virtul autonomous organizations, straight out of a science fiction book. But they are provably real now.

I think that these topics are always downvoted on HN is an example of "actively looking away". Change is scary.

I think big technical leaps have slowed down in "tech," meaning anything with computers.

However, giant leaps are happening and will happen biologically.

I think 2020-2050 will produce radical breakthroughs in biological—everything.

> However exponential tech that takes us to something like 'singularity' does not seem to come up any more.

I don’t know who else was saying this, but the story of exponential growth leading to a singularity is/was the schtick of Ray Kurzweil, and it’s a narrative that was never true.

Kurzweil was one of the people trying to sell the idea that we will eventually unlock human immortality through technical progress. Along the way he made many dubious and false claims. One of them, for example, a key component of his story, is that human longevity has increased by many years in the last 200 years of human history. He does this by showing a graph of life expectancy that starts at a low point when disease and war was rampant, and by omitting data points immediately after world war 1, then “fitting” a completely misleading curve to the cherry picked data. The life expectancy plot at the very bottom of this page is complete bunk: https://www.kurzweilai.net/the-law-of-accelerating-returns

This feels like spotting people at the beach. The crowds and breakthroughs always look closer to each other from a distance.

people are the bottleneck

I think that's the typical curve of advancement. You expect it to move with noticeable linear speed, but actually what one usually perceives is that short term changes are surprisingly small, while long term changes are surprisingly large.

I mean - I took 25 years from first non-exploding V2 flight to the first moon landing. Given it took them 7 years just to build the turbo pump in the engine 25 years is quite a short time to get from a munitions deployment platform to a human space program.

I think there is stagnation in some areas, probably because everyone was been interested in finding out what you can do with the internet, but I think we've now reached the stage where it's part of daily plumbing and not exciting and new anymore, and perhaps more brainpower will be available to figure out what fancy things we can design with this ephemeral super computing capacity.

I like to drop this link when this topic comes up and also recommend reading anything by Vernor Vinge. My favorite is Across Realtime.

https://longnow.org/seminars/02007/feb/15/what-if-the-singul...

Paradigm shifts, at least in the Khunian sense, tend to happen over a generational scale, which means, firstly, that they take a considerable time to become recognised as the new paradigm, and secondly, that you tend not to see too many in a lifetime.

If by 'singularity' you mean the one defined by Kurzweil/Verner Vinge, it is worth keeping in mind that he thinks that exponential advancement has a huge component to it which appears linear, where things appears to be hardly advancing or only marginally advancing.

I think e-bikes are the next thing.

Exponential tech bringing us to a singularity was never happening.

We had at least two tremendous breakthroughs in technologies that act as enablers and multipliers for other technologies since the last industrial revolution. The first was figuring out how to store, transport, and use energy. We lucked out a bit on the very existence of fossil fuels and then made an all-time great physics breakthrough discovering nuclear fission and fusion. All of these are limited by storage capacity and transmission bandwidth, and once the road and wiring infrastructure was in place, there isn't a whole lot more than incremental improvements to be had. The other paradigm shift we seemed tantalizing close to was Tesla trying to develop wireless energy transmission, but it never happened.

The second was electronic computing. When we figured out we could use transistors to print computers onto increasingly smaller circuit boards, that unlocked decades upon decades of rapid scaling potential that we have mostly done a great job of continually realizing. Computational signal transmission, unlike energy transmission, actually works fine without wires, which enabled mobile and grid computing and build out of massive richly connected networks.

These aren't going to unlock singularities, though. If we're only looking for advances in energy and computing, then yeah, of course, those were going to slow down at some point compared to when we found major paradigm shifts like transistors and fossil fuels. Wireless energy transmission and cold fusion are probably the two holy grails that could make a lot of other stuff happen, but we have no idea if they're really possible, and the past few decades of "our greatest minds are focused on how to price a CDO" shifting to "our greatest minds are figuring out how to get people to click on ads" probably isn't helping. Without the commercial incentive, research to enable larger breakthroughs probably has to come from governments, but making better governments isn't in the purview of technology. As others have mentioned, we don't have the world wars, space race, and cold war to motivate us any more. Maybe a greater US/China rivalry can actually rekindle governments that fund important basic research?

There are other obvious enablers we need beyond energy and computing, too. Transportation is one. I don't think we've made any major breakthroughs here since the automobile and airplane were invented. Widespread work from home and remote collaboration could probably go a long way toward solving this problem by doing an end-around and making people not need to move as much rather than figuring out better ways to move them.

Food production is another. Healthcare is another. Our greatest sci-fi imagines teleportation, splicing apart air molecules and putting together protein and carbohydrates, and immunity from all diseases and possible curing aging. Are those really possible? Who knows, but again, it'll take a lot of commercially unviable, largely failed, basic research to even figure this out. Thousands of people will need to spend many decades of career time, possibly entire lifetimes, achieving nothing so that one person can achieve a breakthrough.

And, at some point, we need to figure out how to keep an exponentially growing population housed and productively occupied. That isn't a technological problem. We have the tech right now to give everyone their own cubbyhole and coloring book, but humans want more than that.

You're not looking very hard. There are three major technologies that are on the horizon: solar, AI and cryptocurrency.

Solar is probably the safest bet among the three. A combination of Swanson's law [0] and Wright's law [1] have solar panels dropping exponentially, halving every 6-7 years. Battery technology needs to go hand in hand with the increased solar capacity in order to actually use it and I'm a little weaker on evidence of how quickly that's dropping but I think there's some expectation that battery technology is also experiencing it's own exponential price decrease [2]. The current US average household uses about 30kWh per day and has an average yearly energy bill of around $1200. My guess is that when solar and battery technology start to be priced in the $5k range for a 30kWh system, that's when we'll see a dramatic shift to solar. I would expect this to happen in the next 10-15 years. With solar and battery technology, you have the potential to get other unforseen events, like decentralized energy grids, ubiquitous and cheap energy etc. which all could lead to new markets and innovation.

My guess is that AI is coming in the next 10-15 years. The human brain is estimated at 2.5 petabytes [3] of information and using cost of hard disk space as a proxy for the compute and other infrastructure to create general level AI, with current trends in exponential decreases in hard disk cost (and exponential cost decreases in other compute) a $1000 petabyte hard drive will be here within the next 10-15 years. We can already see spectacular results from speech synthesis, deep fakes, language synthesis etc. Two minute papers is a train of incremental spectacular results [4]. Kurtzweil has been claiming the singularity would happen around 2030 [5] and I see no reason to suspect this to be substantially wrong. There are many different interpretations of what "singularity" means and when we can expect human level AI but most of the reasonable predictions, even ones based on "back of the envelope" calculations (like the one I just did) all lead to the 2030 range. Many predictions about AI haven't come true so people get soured on it but the ones that should be believed are based on evidence, give reasons why and when the singularity will happen and, of importantly, are falsifiable.

Cryptocurrency give the promise providing a ubiquitous payment system, capable of allowing for near instant microtransactions outside of a single nations control. HN is notoriously vitriolic towards cryptocurrency for some reason, which I've always found disconcerting since cryptocurrency is much in the same spirit as the early internet's decentralized philosophy. Cryptocurrency also gives, in my opinion, the only viable option for digital micropayments, which is something I haven't heard since the early 2000s when it was abandoned in favor of Amazon's "1-click shopping" or Ebay's payment system. Proof of work is known to be energy intensive but the response is that there's a cost to the current banking system we're paying, sometimes in the form of externalities, so it's not clear we're doing worse with PoW. I've heard, though I still don't feel like I have enough evidence to decide one way or the other, that cryptocurrency might actually speed the adoption of solar as cryptocurrency mining is an ideal energy source (sporadic, excess energy that can be used to give a return on investment, especially in areas that claim to not be able to support excess solar energy because of legacy infrastructure). I'm focusing just on the "developed nations" consumer level benefits without touching on some of the other more fundamental issues that cryptocurrency might enable through accessible banking to portions of the population that are unbanked, providing banking infrastructure to developing nations that have troubles with the current banking infrastructure and providing a unifying currenc...