* also, an integrated touchscreen, inertial measurement sensors, a half dozen wireless protocols, two to four cameras, an array of microphones, speakers, LIDAR, OLED, a rechargeable battery....
John Ohno is a treasure. Very few people out there with well calibrated macroscopes like John.
There's some hand waving (a lot) here, but we've lived under 20 years of de-empowerment: simplifying consumerizing, cloudifying. We used to try to make computing itself better. That innovation has been quelled.
And how sterile and boring computing culture has become... it's really a shame how overpowering the business aspect is to CS teaching and employment. We're losing something important...
Title of this HN post is correct in the sense that it’s the title of the referenced article, but that title is wrong for the article’s content.
The article claims computer science hasn’t moved forward (much) since 1978, claiming that the changes in the 40 years between 1940 and 1980 were much larger than those between the next 40 years (1980 and 2020)
I think a simple model could explain that. Let’s say computing went from ‘1’ to ‘41’ in those first forty years, an improvement of about a factor of 40. Keeping the same speed, it would have gone from ‘41’ to ‘81’ in the second 40 years, an improvement of about a factor of 2. That’s an order of magnitude less than that factor of 40.
You can change those numbers to get any difference you want. Point is that, if progress is more or less linear and improvements are measured as relative size differences, improvement rate always will slow down over time.
Now, whether that model or something like it holds, I’m not sure about, but I think it might hold for many new fields (what happened after the invention of the alphabet, for example? A few thousand new glyphs, emoji, etc. were added, but there’s not much really new there)
Heading to 8 now with fewer and fewer people that have to work at producing food.
My guess is that the number of scientists and engineers alive today at least matches the cumulative number of scientists and engineers who have been alive since the beginning of humankind. We'd slow down if population halves.
Both "standing on the shoulders of giants" and number of people should apply equally well to the progress in CS. We are however slowed down by the effort put into the distraction of commercialization.
What a terribly dogmatic way of looking at things. Some things don't change because they work fine, like core network protocols. Other things have improved so dramatically that they're indistinguishable from magic to anyone who lived a century ago (real-time HD video conferencing, deep fakes, VR, speech assistants like Siri), but according to the author these are merely "incremental improvements".
Computing is the area that has seen the most innovation since the seventies. This is the point Peter Thiel often makes; computing has gone through a Cambrian explosion while other fields like medicine have been comparatively stagnant (cancer still isn't cured).
Yes, computing in general but what has SV done lately? I could likely do without most of it. A web browser and HTML for info is enough. The excess complexities keeps me gainfully employed so that's something.
I was fine with Mapquest before Maps and mostly a Wikipedia fan. Lately during lockdown though, I do find myself looking for things on YouTube. If it didn't exist I wouldn't know what I was missing and be doing something else.
It's probably different if you don't already have many sources of educational information readily available: +1 YT.
I remember that I used Mapquest and switched to Google Maps. I don't remember why but I assume that GM contained some significant improvements. I remember my enthusiasm for the maps API that let me put markers on a map embedded in my website (2005.) Maybe also satellite view. Street View later. So Google did do something substantial with their implementation of mapping.
Satellite and Street view are the main reasons I keep using GM now, plus checking addresses, some car directions and navigation and sharing links to locations with friends. I'm using OpenStreetMap based software for discovering and recording cycling tracks.
One of the things that was not at all clear in the late 70's was the direction of things towards device convergence plus consumer internet. There were plenty of commercial visions of dedicated-task machines that were quite a lot more limited than anything envisioned at PARC. Those machines did find some markets, in some cases even into the 90's, but bit by bit, they got knocked out by commodity computing into a relatively small number of form factors, with machines sufficiently open that most computing and I/O ideas are realizable, albeit with specific parts that are proprietary.
What hasn't been addressed, and I think this is really underlying the sentiment that innovation hasn't happened, is that we're still not really much better at concepts of software literacy. It's still a professionalized realm in spirit, if not in practice, and the software world has a far greater ability to reinvent wheels and avoid standardization. What we have instead are hundreds, maybe thousands of niches where there are professional developers who are using these commodity tools in surprisingly different ways from each other. That kind of experimental data and deep domain knowledge just wasn't around - there were no "VR developers" or "software engineers in test" in 1978 - and I see it increasingly informing newer generations of development tools and practices in a craft-tradition sense.
So if we narrowly define innovation in the way the article does, I would take it a step further to say that innovation isn't all its cracked up to be and we can greatly improve our technology without innovating. If innovating was so useful we would've kept going after 1978 and since we didn't I am satisfied to conclude that we knew better than to waste our time chasing this defined version of innovation.
Perhaps its because its simpler to invest in proven technology?
the pre-1978 hardware(pre-PC era) was niche computing without wide consumer base. Once there is a large consumer base it dictates the market.
27 comments
[ 2.4 ms ] story [ 65.6 ms ] thread* also, an integrated touchscreen, inertial measurement sensors, a half dozen wireless protocols, two to four cameras, an array of microphones, speakers, LIDAR, OLED, a rechargeable battery....
Invented in 1965 [1].
> inertial measurement sensor
Commercially available since 1949 [2].
> a half dozen wireless protocols
This doesn't mean anything. Wireless data transmission was invented long before 1970.
> two to four cameras, an array of microphones, speakers, LIDAR, OLED, a rechargeable battery
All of these can be done precisely because there has been a reduction in size. None of this completely new tech. It's incremental reduction in size.
[1] https://digital-library.theiet.org/content/journals/10.1049/....
[2] https://www.researchgate.net/publication/294669755_The_histo...
[1] https://www.imdb.com/title/tt0701148/characters/nm0810379
1) pre 1970:
Dynabook, AKA KiddiComp (Alan Kay, 1968)
2) 1980's PARC & Apple inspired
MagicSlate (Bill Atkinson, early 1980s)
Ubiquitous computing, AKA ubicomp (Mark Weiser, 1988 at PARC)
General Magic / Magic Cap (Andy Hertzfeld, Bill Atkinson, et al. 1990)
3) Other, screen sharing, etc
Application teleporting (mobile applications), Active Badge System (Andy Hopper et al., Olivetti/Cambridge, mid 1990s)
There's some hand waving (a lot) here, but we've lived under 20 years of de-empowerment: simplifying consumerizing, cloudifying. We used to try to make computing itself better. That innovation has been quelled.
The article claims computer science hasn’t moved forward (much) since 1978, claiming that the changes in the 40 years between 1940 and 1980 were much larger than those between the next 40 years (1980 and 2020)
I think a simple model could explain that. Let’s say computing went from ‘1’ to ‘41’ in those first forty years, an improvement of about a factor of 40. Keeping the same speed, it would have gone from ‘41’ to ‘81’ in the second 40 years, an improvement of about a factor of 2. That’s an order of magnitude less than that factor of 40.
You can change those numbers to get any difference you want. Point is that, if progress is more or less linear and improvements are measured as relative size differences, improvement rate always will slow down over time.
Now, whether that model or something like it holds, I’m not sure about, but I think it might hold for many new fields (what happened after the invention of the alphabet, for example? A few thousand new glyphs, emoji, etc. were added, but there’s not much really new there)
According to https://ourworldindata.org/world-population-growth
1700 660 million
1800 990 million
1928 2 billion
1999 6 billion
Heading to 8 now with fewer and fewer people that have to work at producing food.
My guess is that the number of scientists and engineers alive today at least matches the cumulative number of scientists and engineers who have been alive since the beginning of humankind. We'd slow down if population halves.
Computing is the area that has seen the most innovation since the seventies. This is the point Peter Thiel often makes; computing has gone through a Cambrian explosion while other fields like medicine have been comparatively stagnant (cancer still isn't cured).
Web Browser and HTML is useless without the videos and photos which have been aggregated and precomputed for distribution from the datacenter.
It's probably different if you don't already have many sources of educational information readily available: +1 YT.
Satellite and Street view are the main reasons I keep using GM now, plus checking addresses, some car directions and navigation and sharing links to locations with friends. I'm using OpenStreetMap based software for discovering and recording cycling tracks.
What hasn't been addressed, and I think this is really underlying the sentiment that innovation hasn't happened, is that we're still not really much better at concepts of software literacy. It's still a professionalized realm in spirit, if not in practice, and the software world has a far greater ability to reinvent wheels and avoid standardization. What we have instead are hundreds, maybe thousands of niches where there are professional developers who are using these commodity tools in surprisingly different ways from each other. That kind of experimental data and deep domain knowledge just wasn't around - there were no "VR developers" or "software engineers in test" in 1978 - and I see it increasingly informing newer generations of development tools and practices in a craft-tradition sense.
But we will have flying cars (working for over 100 years), so worth it.
Software: GPT-3, voice assistants, blockchain, compression, H.264/H.265/AV1, image processing, search, databases, BitTorrent/DHT, open source, MMORPG, autonomous driving, Wikipedia, Linux, map/reduce, deep learning, deep fakes, NEAT
It was really interesting to learn how long this tech has been in use.