[HN Gopher] The Fifth Generation Project in Japan (1992)
___________________________________________________________________
The Fifth Generation Project in Japan (1992)
Author : tosh
Score : 119 points
Date : 2024-10-17 22:03 UTC (1 days ago)
(HTM) web link (www.sjsu.edu)
(TXT) w3m dump (www.sjsu.edu)
| varispeed wrote:
| It's like binary system was "invented" like 300 years before it
| actually has been used, most likely the Japanese idea will
| actually happen at some point, just that when they were trying to
| develop it, the technology was not advanced enough for it.
| bruce511 wrote:
| While it's certainly interesting to read about this sort of thing
| - and I'm interesting in all the computer history - it's worth
| pointing out that some very large percentage of all computer
| systems, from every country and every era failed.
|
| So it would be a mistake to read too much into this project being
| Japanese, or indeed criticize their goals and failures. Like all
| of their contemporaries they boldly went where no-one had gone
| before, and like the vast majority of their contemporaries (and
| most everyone since) discovered limitations in both the
| technology of the day, and their own abilities.
|
| The important thing is not that they failed. The important thing
| is that they tried.
| duxup wrote:
| Yeah until we got a sort of core dominant systems, late 90s?
| There were a lot of creative attempts at new systems and ideas.
| They were either going to surpass the existing systems or fall
| on their face entirely.
| Dalewyn wrote:
| >So it would be a mistake to read too much into this project
| being Japanese,
|
| As a Japanese(-American) I can tell you why this project failed
| and more broadly why Japan's computer industry has failed
| outside of vidja gaemz (and that too is on its way out).
|
| So why did the Japanese fail? It's very simple: Lack of
| cooperation. The article describes the project "was a
| collaborative effort of the Japanese computer industry" but the
| results could not put any more mud on that thought.
|
| Japanese companies _HATE_ cooperating, they will not create nor
| utilize common standards nor share knowledge with each other.
| Japanese companies will strive at every turn to create a
| captive market, hoping that once you 've set foot in their
| little quagmire you will stay stuck with them.
|
| If you ever wondered why Canon camera lenses do not work on
| Nikon cameras and vice versa, why Kenwood microphones do not
| work with ICOM radios and vice versa, why Sony insisted on
| Memory Stick over SD card, why Toshiba threw out NAND flash
| technology, and so on then this is why. Try naming even a
| single international industrial standard to come out of Japan,
| you probably can't and I don't blame you. Sharing is not in
| Japan's DNA.
|
| Contrast the US computer industry where our companies besides
| Apple can and will cooperate to create and utilize common
| standards and push the market wider and forwards. PCI is a
| foremost example, but the very foundation of the industry is
| built on IBM PC-compatibles.
|
| All this means Japan will always be stuck with a bunch of
| small, individual pies nobody will care about, losing to the
| rest of the west who cooperate (and nowadays China who force
| entire industries at gunpoint) for a much bigger pie. You can't
| win when you're all trying to drag each other down in a zero
| sum game.
| AlbertCory wrote:
| You might well be 100% right. I'd like to see the Japanese
| succeed, actually.
|
| However, it was simply not possible for _anyone_ to succeed
| at AI in the early 80 's. It took some Nobel-prize-winning
| software, a change of approach, and a massive increase in
| compute power to finally break through.
| llm_trw wrote:
| The only thing we needed was the compute. Everything else
| had been discovered by the 80s.
| philipkglass wrote:
| I think that we also needed a much larger training corpus
| than was available in the 1980s. Back then the largest
| textual data sets were orders of magnitude smaller.
| adastra22 wrote:
| Using neural nets. The approach taken towards AI in the
| 80's was much less compute intensive.
| numpad0 wrote:
| Neural nets with hidden layers is literally older than
| the Apollo 11 landing. That doesn't make sense but it is.
| adastra22 wrote:
| Read TFA. The Fifth Generation Project involved expert
| systems & prolog, not neural nets.
| anon-3988 wrote:
| What do you attribute this to?
| jdietrich wrote:
| Japanese corporate culture has historically been defined by
| extreme loyalty - salarymen were recruited directly from
| university and were promised a job for life. Promotion was
| overwhelmingly from within and based substantially on
| seniority, with a steep power gradient.
|
| If you were _trying_ to design a system to produce
| proprietary silos, you 'd struggle to beat large Japanese
| corporations circa 1980.
| numpad0 wrote:
| The lore I've heard is that hire-for-life system was
| supposed to create psychological safety as a response to
| excessive siloing habits exhibited by WWII veterans,
| after surviving in reverse order of their individual bus
| factors - e.g. skilled maintenance crews, medics, were
| more likely to be spared in last-ditch assaults, which
| happened a lot in the war.
|
| IMO, that after series of economic downturns and
| corporate restructurings during 80s-00s in turn created
| excessively risk-averse Japanese corporate culture; _now
| that they can be fired_ , psychological safety was gone,
| and now maintaining hire-for-life eligibility was seen as
| crucial. It may have been implicitly reinterpreted as
| eliminating others is a must for survival.
|
| That's clearly one of major contributing factors to
| counter-productive behaviors causing stagnation of
| traditional Japanese tech companies; outright refusing
| sales, exports, opportunities, that may affect internal
| politics and justify job terminations.
|
| It sucks to watch a lot of cutting edge technologies
| slowly evaporating from such fear of terminations.
| Hopefully someone figures a way out.
| Dalewyn wrote:
| >Hopefully someone figures a way out.
|
| The only realistic solution I've seen so far is placing
| Japanese companies under foreign control.
|
| Kioxia is an absurd example. It went from being Toshiba's
| rejected-and-dejected NAND flash division to a world
| leading company literally overnight once it was bought
| out by an American investment firm (Bain Capital).
|
| On that note, Toshiba proper is in the process of getting
| bought out and chopped up by Japan Industrial Partners,
| which is really just a front for an American investment
| firm (Bain & Company) among others. Olympus also suffered
| the same fate several years ago.
|
| Another is Sharp, which managed to claw back from
| bankruptcy once China/Taiwan (Foxconn) bought them out.
|
| Yet another is Nissan, which became relevant on the world
| stage after getting effectively bought out by France
| (Renault, Carlos Ghosn). Nissan probably would have gone
| even bigger if Ghosn and Renault had succeeded in totally
| taking them over.
|
| Even 7/11 (Seven & i Holdings) which was a rare _reverse_
| success story of Japanese management bailing out a
| failing American business is now in dire straits and
| risking a hostile takeover from a Canadian business
| (Couche-Tard).
|
| Japan has the technology, engineers, and designers; what
| Japan doesn't have is an appropriate culture.
|
| It's frustrating watching as a Japanese-American.
| ahartmetz wrote:
| There is one thing that I've seen and heard Japanese and
| Korean companies having over Western ones: a lack of
| infighting and turf wars. Everybody pulls in the same
| direction for the most part. That part of their culture
| is a really strong suit - it's not only the capable
| individuals.
| Dalewyn wrote:
| >a lack of infighting and turf wars.
|
| If I didn't know better I would think you were trolling.
| Japanese companies _absolutely_ have infightings and turf
| wars the likes of which make western ones look like child
| 's play.
|
| Toshiba literally _hated_ their NAND flash division
| (former Toshiba Memory, now Kioxia) despite being one of
| their only consistently valuable assets.
|
| Sony Interactive Entertainment (SIE) despises Sony Music
| Entertainment Japan (SMEJ) because SIE perceives SMEJ
| with its subsidary Aniplex to be encroaching upon their
| core business of video games.
|
| Subaru for the longest time brushed away their team
| working on EyeSight, now a standard feature and widely
| heralded, because they perceived the guys working on it
| as infatuated dead weights.
|
| ANA and JAL loathe each other (also with a healthy mix of
| commercial vs. governmental rivalry due to their history)
| and will drag each other down at every turn.
|
| I can go on but I think you can get the idea: Japanese
| companies really do not like playing together.
| adastra22 wrote:
| > Japan's computer industry has failed outside of vidja gaemz
|
| Worth noting that the big video game companies in Japan--
| Nintendo, Sony, Square Soft, etc. are atypical Japanese
| companies too, run a bit differently than the others. Which
| is no doubt a big reason for their success.
| dcrazy wrote:
| Sorry, but I am skeptical of your argument that Japanese
| companies are any less cooperative or more avaricious than
| companies from other countries.
|
| I am aware of several examples that directly contradict your
| argument:
|
| - JIS screwdrivers.
|
| - Really, JIS anything.
|
| - MSX computers.
|
| - FourThirds/MicroFourThirds.
|
| And there's a practical reason for camera brands to have
| different mounts: different camera systems have different
| flange distances! You'll even see multiple mounts within the
| same brand as the line evolves, as seen with the transition
| from Minolta/Sony A mount to E mount.
| numpad0 wrote:
| OT but flange focal distance is overmystified. It's just a
| mechanical interfacing spec.
|
| You can't drill a hole on the sensor and mount lens body on
| it, and lenses must be placed at exactly focal distance
| away for parallel beams to converge on film or farther for
| diverging beam(from non-infinitely-far objects), so body
| designers pass a distance measurement from film to a known
| feature(mounting ring top surface) to be able to design a
| lens barrel. That's the flange-focal distance, sort of a
| "design length of body" to subtract from your overall lens
| barrel length.
|
| It's actually possible to drill a hole on the PCB and
| woodscrewing lens directly on in cases with webcams and
| CCTV cameras, and in those cases the distance is 0.
| avidiax wrote:
| For the case of 35mm SLRs, Nikon/Canon/Pentax etc. all
| have 45mm flange distances, give or take a couple of
| millimeters. That's basically the minimum to have room to
| swing the mirror and for a couple of shutter curtains.
|
| So all of them having completely incompatible mounts is
| because they wanted to compete based on lock-in to a
| complete system, rather than compete for bodies and
| lenses separately.
|
| Even worse, the fact that their flange distances are all
| within a couple of mm makes passive adapters impossible,
| since just the bayonet part of the mount is longer than
| that.
| hcarvalhoalves wrote:
| The Pentax followed the distance of the M42, which was
| pretty much an open standard by then, and many
| manufacturers at the time released lenses in K bayonet.
| The other camera makers diverged from the standard.
| Dalewyn wrote:
| >several examples
|
| JIS isn't relevant outside of Japan, though. Everyone cares
| about ISO (USA) and standards like RoHS (Europe), nobody
| even knows what a JIS is.
|
| >there's a practical reason for camera brands to have
| different mounts
|
| The reason is irrelevant, it's a camera and a lens and the
| fact they are _incompatible_ is inexcusable. Japanese ham
| radio takes this further; when I said Kenwood microphones
| don 't work with ICOM radios I mean they _could_ use the
| same number of wires and perhaps even the same connector
| but they keep the wiring positions proprietary.
|
| So you want to use your Kenwood gear with ICOM? Or Yaesu?
| Well too fucking bad bud, either get out some wire cutters
| and a soldering iron or buy into one of the ecosystems like
| a good Nippon-jin.
|
| Imagine if Logitech, Corsair, Steelseries, and Razor all
| used USB for their peripherals but the wiring is unique to
| each of them, and imagine if mobos brands like ASUS and MSI
| partnered with specific peripheral brands and only featured
| USB ports compatible with them alone.
|
| That's the Japanese ham radio scene and yes it fucking
| sucks; and yes, they are getting eaten alive by cheap
| Chinese ham radio equipment that doesn't do that
| proprietary shit.
|
| If ham radio is too niche for you, don't worry: It expands
| to stuff like household appliances too. If you buy a
| Panasonic microwave oven or fridge and it breaks down you
| bet your ass you will need to buy replacement parts and
| perhaps even the repair labor from Panasonic.
|
| We all love to complain about walled gardens and planned
| obsolescence here in the Anglosphere, but Japan is several
| orders of magnitudes worse. They make Apple look like a
| paragon of standards compliance and interoperability.
| new299 wrote:
| Essentially all medium to high-end ham radio brands are
| Japanese. Cheap Chinese radios only dominate at the low
| end, and often have emission issues. I doubt the Japanese
| manufacturers are interested in competing here...
|
| Chinese radios do not dominate in Japan, I suspect almost
| nobody in Japan uses a non-domestic radio. Possibly the
| primary reason being that they generally have not passed
| the local emission certifications (which is annoyingly
| required for all ham radios).
|
| I don't see the incompatible microphone thing as an issue
| unique to Japan. Pretty much every 1980s computer had
| weird propriety interfaces. I suspect it was competition
| that forced them to standardize. You can find similar
| issues more recently (Apple/Lightening port, Dell/IBM
| laptop chargers, docking stations etc.)
|
| If you want to point to issue with ham radio in Japan,
| I'd go with the certification issue, which helps lock
| non-domestic players out of the market...
| Barrin92 wrote:
| >and nowadays China who force entire industries at gunpoint
|
| China's software sector was arguably at its most innovative
| when it was in what can only be described a state of
| perpetual warfare. During the so called "QQ-360 wars"
| companies didn't just not cooperate, they pulled stuff like
| this:
|
| _" Qihoo, in turn, updated its software to strip away access
| rights for key components of QQ, while essentially labeling
| the program spyware. (The jury's still out on whether Qihoo
| is right.) In response to Qihoo partially disabling QQ,
| Tencent then forced Chinese users to choose between using QQ
| or using 360--you can't use both. This is akin to Apple
| saying you can't view Google websites on your Mac because
| some Android phones have multitouch. To add PR-based insult
| to software-based injury, users of the older web-based QQ
| found their interface replaced with a letter about Qihoo's
| alleged misdeeds, creating the impression that Tencent had
| blocked Web QQ on all machines just to spite 360 users."_
|
| The Chinese tech sector was full of mutual sabotage like this
| and while it was at times annoying for users, the absurd
| level of competition arguably benefited the ecosystem
| overall. I'd argue Japan if anything doesn't compete enough,
| there's often not strong enough of a focus on growth and
| clawing market share back.
|
| https://www.forbes.com/sites/china/2010/11/10/the-secret-
| win...
| Dalewyn wrote:
| This reminds me of electronic payment systems like SUICA
| and ICOCA.
|
| Once upon a time: The Kanto region had SUICA which was
| introduced and pushed by JR East. The Kansai region had
| ICOCA which was introduced and pushed by JR West.
|
| Yes, the two JR companies are affiliated and work
| together...ish.
|
| No, the two payment systems are _NOT_ compatible with each
| other. It 's only since the last 10 years or so that SUICA
| and ICOCA coverage has overlapped enough that holding just
| one or the other has become practical; you needed both if
| you did any amount of travelling between the regions.
|
| Japan seriously hates cooperation.
| numpad0 wrote:
| Or monopoly. Suica is a JR East product, so baked in
| interop across Japan would mean near total domination for
| small electronic payments by JRE. That's bad.
|
| Although, in hindsight, that wouldn't have been as
| terrible as it sounds. VISA payWave would have been even
| more DOA than it is today in Japan if it was done that
| way.
| Dalewyn wrote:
| What ended up happening is that actually universal
| payment systems like credit cards (Visa, MasterCard,
| AMEX, and JCB) and digital wallets (PayPay, Alipay, Apple
| Pay, Google Pay) were ultimately successful in carving up
| the formerly cash-dominated pie, forcing SUICA and ICOCA
| to open themselves up to broader interoperability
| including beyond just JR offerings.
|
| Also, note that JCB is the only Japanese one of those
| universal payment systems. All the others are either
| American or Asian ex-Japan (PayPay which is South Korean,
| Alipay which is Chinese).
| numpad0 wrote:
| That's a strange timeline. Suica transit and payment
| started in 2001, flip phone integration in 2004, Android
| in 2010. PASMO-Suica interop started in 2007, expanding
| to 10 cards across regions in 2013. Apple Pay with NFC
| launched in 2016, iPhone 7 with FeliCa launched in 2016,
| PayPay was 2018. Those all came after success of Suica
| and its slowly expanded mutual interoperations.
|
| JCB is also a card brand, not a payment system. System is
| more like payment networks like CARDNET.
| dcrazy wrote:
| JCB has its own payment network in Japan. In other
| countries they piggy back off AmEx or Discover/Novus.
| fomine3 wrote:
| VHS, CD, DVD, Blu-ray, MIDI, JAMMA, SD card,
| new299 wrote:
| You mention SD Cards.
|
| The SD Card was developed by two Japanese companies
| (Panasonic and Toshiba) and Sandisk.
|
| VHS is another Japanese standard, adopted by multiple
| manufacturers.
|
| Then there's the MSX standard...
|
| Japanese industry is no doubt somewhat different and may have
| its issues. But I don't think it's as simple a picture as you
| paint... some external factor is often required to force
| large companies to cooperate. Those factors may be more
| common outside Japan, but I don't see that the fundamental
| issues are very different.
| Dalewyn wrote:
| >The SD Card was developed by two Japanese companies
| (Panasonic and Toshiba) and Sandisk.
|
| Yes, but that is after Toshiba threw away NAND flash
| technology saying Intel invented it (yes, really...) and
| Sandisk went and picked up free real estate. Not to mention
| Panasonic and Toshiba basically got roped into marketing
| what was Euro-American tech: MMC, SD card's predecessor,
| was invented by Sandisk and Siemens.
|
| Sony meanwhile never got on board until the bitter end with
| MemoryStick.
|
| >VHS is another Japanese standard, adopted by multiple
| manufacturers.
|
| VHS is actually an exception and exemplifies what happens
| _IF_ (and that 's a big if) Japanese companies can be
| convinced to work together on something bigger than
| anything they could individually achieve.
|
| There are other examples like Nintendo and the wider
| Japanese video game industry sharing and protecting patents
| for each other, or Nissin (Cup Noodle!) giving away their
| trade secret to kickstart the instant noodle market and
| improve food supply in the immediate post-war era.
|
| Sadly, these things don't happen that often.
|
| >Then there's the MSX standard...
|
| That's actually a Microsoft invention which was co-marketed
| with ASCII Corp.
|
| If it wasn't obvious by now, what seemingly "Japanese"
| international standards we do get are usually not Japanese
| at all having leadership or core involvement from the rest
| of the west.
|
| eg: Blu-Ray involved Phillips and HD-DVD involved Warner
| Brothers, among others.
|
| This actually goes to a sibling comment[1] I made here,
| which is that stagnant or failing Japanese companies are
| very likely to explode into success if unshackled from
| Japanese culture.
|
| Japanese are amazing inventors and innovators, but Japanese
| are also horrible pioneers and trailblazers.
|
| [1]: https://news.ycombinator.com/item?id=41876563
| tkgally wrote:
| Not to disagree with you or the commenters pointing out other
| Japan-specific factors, but I think there is one more possible
| explanation for the project's failure: its top-down,
| government-bureaucrat-led nature. I've lived in Japan for forty
| years and have worked at a national university for the past
| twenty. The government regularly launches initiatives, sets
| goals, and promotes catch phrases that are supposed to lead to
| greater progress and prosperity. But, because the initiators
| and funders are often only tenuously connected to the complex
| and dynamic reality, the projects become rigid and wasteful,
| and eventually wither and die, more often than not.
| komali2 wrote:
| I've lived in Taiwan for a while, not nearly as long as you
| have in Japan, but long enough that I have a couple
| observations I'd be interested in your take in, about "old
| hats" such as yourself.
|
| The first is that the folks that immigrated decades ago tend
| to hold on to the perception they had upon arrival to the
| country, and the various impressions, for the rest of their
| lives there, and these impressions inform all future
| observations they have about the country. For example in
| Taiwan a lot of older American immigrants will talk about
| things like crime or the police as if it was still the KMT
| era when Taiwan was basically a different country, and their
| thoughts are really strange to hear when considering modern
| taiwanese police (who are basically teddy bears) or crime
| rates (extremely low).
|
| The second is that, having spend decades in the country to
| which they immigrated, which is probably longer than they
| lived in their origin country, they seem to view everything
| that strikes them as odd, inefficient, or bad, as unique to
| their new home country. In this case such as your perception
| of Japan as being lead top-down by bureaucrats, when that
| description to me applies to basically every liberal
| democracy on earth. Thus:
|
| > The government regularly launches initiatives, sets goals,
| and promotes catch phrases that are supposed to lead to
| greater progress and prosperity. But, because the initiators
| and funders are often only tenuously connected to the complex
| and dynamic reality, the projects become rigid and wasteful,
| and eventually wither and die, more often than not.
|
| imo, this is true for most government-led initiatives in
| nearly every country on earth. Not that I'm arguing against
| government-lead initiatives because they do result in
| sometimes incredible things like train infrastructure,
| spaceships, the internet, medical breakthroughs etc, but for
| the most part it seems politicians make big pitches to get
| elected and then just maintain status quo.
|
| Obviously Japan has some uniquely bureaucratic things about
| the culture as a whole (lol faxes), but on the other hand
| this doesn't seem to be an obstacle considering they have
| some of the safest pedestrian-friendly streets in the world,
| some of the best public transit, an excellent healthcare
| system, low crime rates, low homelessness, low unemployment,
| etc. On the ground they're doing much better than many
| countries.
|
| What do you think? You probably know much more about such
| things given your time in Japan as well as the fact that you
| work at a university.
| tkgally wrote:
| Thanks for the thought-provoking comments and questions!
| I'm away from my computer right now--in a coffee shop at
| Ueno Station in Tokyo, in fact. I'll reply again at length
| tonight or tomorrow.
| komali2 wrote:
| Looking forward to it! - posted from a coffee shop next
| to Kodemmacho Station near Akihabara, ironically, just
| popping by after Japan burn ;)
| watwut wrote:
| I cant help but say: Americans tend to see high crime rates
| even when sitting in the safest city in the world. It seems
| like Americans almost culturally see it as a duty to be
| afraid.
| Isamu wrote:
| >Americans tend to see high crime rates
|
| A small rise in crime is _reported_ as a crime wave,
| because that grabs the attention of the older people that
| still consume news.
|
| Or political hacks complain about high crime when it is
| down. Refuting this isn't effective.
| 71bw wrote:
| Topic change, but how manageable is living in TW nowadays
| as somebody who knows 0% of any Chinese (Simp./Trad.)?
| BlueTemplar wrote:
| > The second is that, having spend decades in the country
| to which they immigrated, which is probably longer than
| they lived in their origin country, they seem to view
| everything that strikes them as odd, inefficient, or bad,
| as unique to their new home country.
|
| I would bet that people that never emigrated are even worse
| about that.
| forgotoldacc wrote:
| > The government regularly launches initiatives, sets goals,
| and promotes catch phrases that are supposed to lead to
| greater progress and prosperity.
|
| Some stuff has worked pretty well. I registered for a
| MyNumber card after years of dragging my feet. Now I can
| check how much tax I've paid, how much I have in my
| retirement fund, health insurance things, and a bunch of
| other stuff from my phone in like 10 seconds all from a
| single app. Previously I'd need to take a half day off, bring
| my wooden stamp to the appropriate government office, and
| wait for 10-30 minutes in order to make sure my stuff is in
| order.
|
| I'm actually amazed how well it works.
|
| Japan also builds infrastructure really fast.
|
| It's not perfect, and some things lag behind other countries,
| but some things blow some other countries out of the water.
| If anything, I'd say the government moves pretty quickly with
| a lot of decisions (one consequence of being what's
| effectively a one party state, I suppose). Companies,
| particularly big ones, are more risk-averse.
| authorfly wrote:
| Japan has stagnated and had debt problems over those 40
| years. May I ask you if you saw these as they began to accrue
| and if you think there is a long term strategy to address
| them?
|
| Seems to me as an alien to the culture, that Japan is
| uniquely happy to continue on conservatively (culturally)
| without giving in to medium term fixes(e.g. immigration) but
| with no long term plan for the economics or population. Is
| this because of a different cultural view on those things and
| whether they are successes, or failures?
| RaftPeople wrote:
| I remember when this project was announced and I rolled my
| eyes. It has nothing to do with which country or who announced
| it, it was a typical grandiose announcement around things that
| are difficult to accomplish, probability of success was low.
| anigbrowl wrote:
| tbh the Wikipedia article is a lot better
| https://en.wikipedia.org/wiki/Fifth_Generation_Computer_Syst...
| airstrike wrote:
| Related: https://news.ycombinator.com/item?id=41872982
| sabareesh wrote:
| Recently came across this video which goes into history of
| Japanese computers by Asianometry
| https://www.youtube.com/watch?v=CEtgzO-Im8w
| coreload wrote:
| This project was attempting to spin up relatively exotic hardware
| and software concurrently. The Lisp folks had less exotic
| software already more or less established but attempted to spin
| up corresponding hardware. The 5th generation project failed
| faster due to its greater ambition. Both failed ultimately due to
| commodity hardware greatly outpacing the ability of exotic
| hardware to keep up.
| AlbertCory wrote:
| It's somewhat appalling that they don't mention that there's a
| book with this title:
|
| https://www.amazon.com/Fifth-Generation-Artificial-Intellige...
|
| As for "The important thing is that they tried" : would you give
| them credit for "trying" to invent real-time, neuron-by-neuron
| brain imaging, in 1982? And giving billions of taxpayer Yen to
| the effort? How about quantum computers?
|
| Or would you have been justified in saying that there are
| projects with a better prospect of success right now, and we
| shouldn't waste the taxpayers' money on things guaranteed to
| fail?
|
| So what ARE your limits? Or are there any?
| mindcrime wrote:
| _As for "The important thing is that they tried" : would you
| give them credit for "trying" to invent real-time, neuron-by-
| neuron brain imaging, in 1982? And giving billions of taxpayer
| Yen to the effort? How about quantum computers?_
|
| For me, I'd say:
|
| 1. Yes
|
| 2. Taxation is theft
|
| 3. Yes
|
| _Or would you have been justified in saying that there are
| projects with a better prospect of success right now, and we
| shouldn 't waste the taxpayers' money on things guaranteed to
| fail?_
|
| See again: taxation is theft. But regardless of how you fund
| research, it remains the case that "you don't know what you
| don't know" and "hindsight is 20/20". I'm certainly not going
| to slag the Japanese for the effort they put into the 5th gen
| project just because things didn't work out as favorably as
| they might have.
|
| And also, while the project might not have met their immediate
| goals and is deemed a "failure" that doesn't _necessarily_ mean
| that the research they did is value-less. Speaking for myself,
| I actually recently went on a bit of a buying binge, buying a
| bunch of books on the 5th gen project (read that Feigenbaum &
| McCorduck book ages ago) and related ideas (eg, concurrent
| prolog). Because I have a hunch there's a kernel of something
| useful there still waiting to emerge. Now my research may lead
| me nowhere, but that's OK. And at least in my case I'm spending
| only my own money.
| AlbertCory wrote:
| Let me just repeat: So what ARE your limits? Or are there
| any?
|
| So I guess you're saying there are no limits? Every project
| should be funded to the max?
|
| and since you're against taxation, does that mean ordinary
| people are to be compelled to "voluntarily" fund them? Or
| where is the money supposed to come from?
| mindcrime wrote:
| _and since you 're against taxation, does that mean
| ordinary people are to be compelled to "voluntarily" fund
| them? Or where is the money supposed to come from?_
|
| Honestly, not interested in going any deeper on this topic
| here on HN. It won't lead to anything productive and will
| just start a flamewar. In the off chance that you are
| actually especially interested in hearing my opinion(s)
| here, drop me an email at prhodes@fogbeam.com and we can
| continue there. If not, then I'm satisfied to drop this
| here.
| AlbertCory wrote:
| You made some nonsensical statements on HN, and now
| you're dropping out rather than trying to defend them.
| Got it.
| mindcrime wrote:
| I offered to continue the dialog. You decided to go for
| the "zinger" approach. I think that tells us both all we
| need to know about the merits of me continuing this
| little dialog with you. Have a nice day.
| SSJPython wrote:
| > The Japanese Fifth Generation project was a collaborative
| effort of the Japanese computer industry coordinated by the
| Japanese Government
|
| > In a sense, Japan's ability to stay the course in pursuit of a
| long-term payoff-- usually considered one of the country's
| strongest assets-- turned into a liability.
|
| Japan is an expert at public-private partnerships. Their entire
| economic development story was based on the government and
| private enterprises working together. They didn't take an
| ideological approach to development. They ignored both the
| neoliberal route and the socialist route.
| bobthepanda wrote:
| Well, they're certainly very good at favoritism, which is
| basically what the keiretsu-led model ended up being. And
| favoritism and picking winners only really works as long as
| you're able to pick the right winners.
|
| This model works well to supercharge initial growth if done
| correctly, but eventually you can't really become richer
| through exports alone and have to pivot towards more
| consumption, which can be difficult to pull off.
| 082349872349872 wrote:
| come to think of it, early ARPA was pretty much about picking
| the right winners
| BubbleRings wrote:
| If you happen to click the link at the bottom of the page "HOME
| PAGE OF applet-magic" like I did, you get one of the fake virus
| web sites. Just do Ctrl-Alt-Del (on Windows), select Task
| Manager, sort on Name if you have to, right click on Google
| Chrome (or whatever your web browser is), and choose End Task. Or
| just cold start your computer.
|
| But hey lets keep web sites with links like that off Hacker News,
| what do you say?
| foco_tubi wrote:
| It's a long expired domain that originally belonged to that
| SJSU professor, I doubt he will ever update the links on these
| articles. That doesn't negate the value of the article itself.
| BubbleRings wrote:
| A hundred dollar bill crawling with Ebola virus is still
| worth a hundred dollars, yes. But I would really prefer if
| nobody tried to hand it to me.
| bee_rider wrote:
| Ebola doesn't live very long outside the host, so I think
| you could probably have them put the hundred dollar bill
| aside for a day or so and then come back and grab it.
| userbinator wrote:
| That page did absolutely nothing with JS off; just shows a
| blank page. Now I know it's not worth turning JS on for either,
| like the majority of the stuff I come across.
| numpad0 wrote:
| The title is somewhat intriguing but the content is extremely
| light, and on top contains this: <!-- ZoneLabs
| Privacy Insertion --> <script language='javascript'
| src='http://127.0.0.1:1026/js.cgi?pcaw&r=14945'></script>
| cen4 wrote:
| Its not a coincidence that the Japanese Economy imploded around
| the same time. If your country has 40% of global corporate market
| cap value and then it drops to 8% after an asset bubble bursts,
| how many ambitious long term projects you can sustain drops too.
| The US came really close to following the same trajectory with
| the 2008 bubble bursting. If the banks hadn't been kept afloat
| and there were alternatives to Wall St a large chunk of the tech
| sector would have tanked or moved out. There is no global tech
| leadership possible, if there is no global Finance domination.
| shiroiushi wrote:
| >a large chunk of the tech sector would have tanked or moved
| out
|
| Where exactly would the tech sector have moved to? There's no
| other country that's really able to absorb high-skill
| immigrants the way American can and put them to work in that
| field.
| alephnerd wrote:
| > Where exactly would the tech sector have moved to
|
| It would have been much more decentralized.
|
| Entire generations of Indian, Korean, Chinese, Canadian,
| Israeli, and European origin founders and executives helped
| build the US tech industry.
|
| Heck, a lot of prominent Chinese and Indian startups got
| their start due to the GFC era layoffs and hiring freezes in
| the tech industry.
|
| Look at Flipkart and Ola in India, ByteDance in China, and
| Coupang in SK/US for example
|
| COVID and the recent layoffs had a similar impact on the
| startup scene as well.
| shiroiushi wrote:
| >Entire generations of Indian, Korean, Chinese, Canadian,
| and European origin founders and executives helped build
| the US tech industry.
|
| Yes, but they would not have built such companies if they
| were stuck in their own countries. There's a reason they
| left those countries. Europe, for instance, is notorious
| for being _extremely_ unfriendly to startups.
|
| >Look at Flipkart and Ola in India and ByteDance in China
| for example
|
| I've never even heard of the Indian companies here. As for
| ByteDance, TikTok is likely to be banned in the US and
| maybe elsewhere because it seems to be a tool for the CCP.
| It would be difficult to impossible for any Chinese tech
| company to achieve the status of Google or Microsoft for
| this reason. Companies in these countries may be able to do
| well inside their own countries, but they can never achieve
| real global dominance.
| alephnerd wrote:
| > There's a reason they left those countries
|
| Pay and upskilling. But domestic tech salaries have risen
| across Asia now, and the US has gotten much harder to
| immigrate to.
|
| Most people who joined the tech industry in the US from
| China and India in the 2000s felt they were getting a bum
| deal at the time in those countries because of a lack of
| domestic options, but that's changed since 2008 -
| especially after the 2008-10 layoffs (I remember those -
| they were scary, lots of my middle school friends had to
| move back to China and India because their parents lost
| their H1B or EB1/2 sponsor).
|
| The next generation of Chinese and Indian startups are
| now being founded by former diaspora members after the
| hiring freezes during COVID.
|
| > I've never even heard of the Indian companies here. for
| ByteDance
|
| That's on you tbh. Not everyone is expected to follow
| every industry trend per country, but these are all
| unicorns or exited unicorns that were founded by cohorts
| in the late 2000s who had the option to move abroad but
| chose to remain in their own countries.
|
| ByteDance for example had PMF in China well before it
| entered America as "TikTok", and Flipkart generated $7B a
| revenue a year - which isn't shabby by Indian or even
| American standards.
|
| > Companies in these countries may be able to do well
| inside their own countries, but they can never achieve
| real global dominance
|
| First, Big Tech (FAANG) is not representative of the tech
| industry as a whole, and does not have strong PMF in
| plenty of markets and sectors abroad.
|
| Secondly, this also underestimates the fact that most
| tech companies have strategy and R&D making parts of
| their organizations abroad now.
|
| Plenty of NLP work that became LLMs came out of research
| done in Microsoft China and India, Google Pay's
| development org is largely in India, Apple's R&D labs in
| China and India played a major role in developing the
| manufacturing process and chip design respectively for
| the iPhone, etc.
| edm0nd wrote:
| >First, Big Tech (FAANG) is not representative of the
| tech industry as a whole, and does not have strong PMF in
| plenty of markets and sectors abroad.
|
| It's not? Android is the most popular OS in the world
| with Apple and Microsoft making up the bulk of the rest.
|
| >As of September 2024, Android, a mobile OS that uses the
| Linux kernel, has 45.38% of the global market and is the
| world's most widely used operating system. This is
| followed by Windows with 25.61%, iOS with 18.39%, macOS
| with 5.53%, and Desktop Linux with 1.64%.
|
| https://en.wikipedia.org/wiki/Usage_share_of_operating_sy
| ste...
| alephnerd wrote:
| Depends segment to segment and sector to sector.
|
| Take instant messaging as an example.
|
| In the US, iMessage and FB Messenger are fairly prominent
| messaging services.
|
| Yet in Japan+Taiwan+Thailand it's Line, in Korea it's
| KakaoTalk, in Vietnam it's Zalo, in China it's WeChat,
| etc.
|
| Furthermore, the entire publicly listed tech industry
| globally generates around $6T in revenue, but FAANG
| represents only $1.4T in revenue. The other $4.6T are
| spread across multiple decently large companies globally.
|
| In addition, the stereotypical big tech companies are NOT
| market leaders in plenty of very large market tech
| sectors. For example, Cybersecurity, ERP, Payment
| Processing, CRMs, etc.
|
| Big Tech is big, but the world is big as well. There are
| plenty of markets where a regional competitor or a
| specialized player can outcompeting big tech. For
| example, look at Coupang and Flipkart decimating Amazon
| in SK and India.
|
| Big Tech is big, but it is not an end-all.
| shiroiushi wrote:
| >In the US, iMessage and FB Messenger are fairly
| prominent messaging services. >Yet in
| Japan+Taiwan+Thailand it's Line, in Korea it's KakaoTalk,
| in Vietnam it's Zalo, in China it's WeChat, etc.
|
| Notice how none of those companies in the second line
| have anything close to a globally dominant position?
| They're only popular locally. LINE is probably the best
| here since it's dominant in more than 1 country, unlike
| the others. Meanwhile, people all around the world use
| Facebook (though this is falling off lately admittedly).
| If you want to chat with friends in another country, FB
| Messenger is probably the go-to chat app for this; your
| friends in Europe or the US aren't going to be using LINE
| or KakaoTalk if you live in Japan or Korea, and pretty
| much no one outside the US/Canada uses SMS messaging (and
| thus iMessage) for talking to friends/family.
|
| >Furthermore, the entire publicly listed tech industry
| globally generates around $6T in revenue, but FAANG
| represents only $1.4T in revenue.
|
| MAMAA (because of Microsoft) builds the infrastructure
| most of the tech industry is built on. This
| infrastructure isn't going to come from China; no one in
| the western-aligned nations would trust infrastructure
| tech from there.
|
| >look at Coupang and Flipkart decimating Amazon in SK and
| India.
|
| Inside their own countries, sure, just like KakaoTalk
| dominating Korea. But just like KT, they have no exposure
| outside their countries. Amazon is huge not just in the
| US, but many other countries: Canada, various European
| countries, Japan, etc.
| numpad0 wrote:
| I think GP's point is that LINE or KakaoTalk or whatever
| aren't the only ones that are failing to reach global
| audience, but FAANG is too and "people are stupid they
| don't understand" or something.
|
| I kind of agree with him, there definitely are region-
| dependent lunch combo of dominant players rather than
| there's the single universal set. You seem to have an
| impression that FB Messenger is globally dominant, and
| frankly a lot of people do feel that way, my hunch is
| that's where disagreement is.
| shiroiushi wrote:
| I wouldn't say FB Messenger is globally dominant _at
| all_. However, for a messaging platform that has global
| reach, I think it 's easily #1 (WhatsApp (also owned by
| FB) is probably #2). This doesn't mean most people use
| it, by a long shot; most people mostly only communicate
| with people inside their own countries, which is why
| WeChat, KakaoTalk, etc. are dominant inside their own
| countries.
|
| My broader point is that US big tech companies have
| _global reach_ , and also tend to set standards. I think
| the chat stuff is really a red herring; there's far more
| to tech than chat apps, and those tend to be very
| regional/national as we've discussed.
| alephnerd wrote:
| And my point is that the people within BigTech who lead
| product lines that have eaten the world would have not
| been in the US if there was an extended tech recession in
| the late 2000s.
|
| The product leaders for Google Docs/Spreadsheet/Slides,
| Google Chrome, Google Pay, Azure, etc were all on EB1/2s
| or still in the process of naturalizing in the late
| 2000s, and if there was an extended tech recession in the
| US, they absolutely would have returned to India or
| China.
|
| Same with startups - Zoom, Rubrik, Cohesity, Nutanix,
| Databricks, Datadog, Spotify, etc would have not moved to
| or get founded in the US if there was an extended tech
| recession in the 2008-12 period as the founders were all
| in the process of naturalizing at the time.
|
| As I've already shown you, multiple startups in Asia have
| already been founded by alumni from American BigTech
| during the 2008-12 layoffs and hiring freezes, and
| already a new generation of startups are getting funded
| due to the COVID era layoffs and hiring freezes.
|
| > However, for a messaging platform that has global reach
|
| And that's the issue I'm getting at. WhatsApp has strong
| PMF in Europe and South Asia because the founding team
| and later leadership at Meta were immigrants from Ukraine
| and India respectively who were able to execute fairly
| successful growth strategies in Europe, South Asia,
| ASEAN, and parts of South America due to domain
| experience.
|
| Yet they were unable to crack other large markets.
|
| PMF requires domain experience, knowledge, and
| localization, and a lot of that comes from immigration.
| alephnerd wrote:
| Basically, except it's not that "people are stupid", it's
| more that a lot of people on HN are myopic and think the
| tech industry is just 5 companies.
| adastra22 wrote:
| Israel is a definite exception. It is very startup
| friendly. Just a small market compared to Europe, let
| alone the US.
| alephnerd wrote:
| I wouldn't call Israel Europe - it's a night and day
| difference from Europe, plus that undercuts the
| Sephardim, Mizrahi, Druze, and Israeli Arab contributions
| to Israel.
|
| Also, buses actually works in Europe :p
|
| But yes, the startup scene is strong in Israel, and
| that's thanks to the Israeli diaspora in SV and Boston
| funding startups in Israel in the 2000s - especially Nir
| Zuk of Palo Alto Networks and much of the team at Akamai.
| adastra22 wrote:
| I was explicitly comparing against Europe. I wasn't
| calling Israel part of Europe.
| adastra22 wrote:
| Forgot to add: you add valid points to the conversation
| though, thank you!
| alephnerd wrote:
| No worries!
| AlbertCory wrote:
| No, the Japanese implosion came at the end of the 80's and
| early 90's. That book is from 1982.
| binary132 wrote:
| GP is saying the USian tech industry nearly shared the same
| fate after our bubble burst in 2008
| ww520 wrote:
| That's unlikely for 2008. Tech companies and large corps in
| general had massed a huge amount of cash after the dotcom
| meltdown.
| Yeul wrote:
| The ability to run a trillion dollar deficit cannot be
| downplayed.
|
| When the Netherlands has too much debt politicians and civil
| servants alike go into panic mode.
| 082349872349872 wrote:
| anyone know of discussion on whether the Plaza Accord had had
| any effect on the japanese bubble?
| readyplayernull wrote:
| What would have been the expected 5th gen computer? An OS with
| natural language prompt instead of command prompt? So, we finally
| made it with LLMs.
| twoodfin wrote:
| There's a somewhat heterodox theory that rather than some kind
| of AI moonshot, the Fifth Generation project was an attempt to
| build an alternative interaction model to Latin alphabet
| keyboards feeding teletypes. The idea (supposedly) was that
| ideographically centered Japan could not compete in a software
| market dominated by alphanumeric text I/O; "smarter" systems
| that could leverage arbitrary natural language (written or
| spoken) would level the playing field.
| mindcrime wrote:
| I gotta side with anigbrowl - the Wikipedia page[1] on the Fifth
| Generation Project is a much better read than this article. I'd
| encourage anyone interested in this topic to give it a look. For
| even _more_ detail, consult the Feigenbaum & McCorduck book[2]
| mentioned by AlbertCory. It's a very interesting story.
|
| [1]:
| https://en.wikipedia.org/wiki/Fifth_Generation_Computer_Syst...
|
| [2]: https://www.amazon.com/Fifth-Generation-Artificial-
| Intellige...
| contextfree wrote:
| One of the researchers involved (I think) also maintains a
| "Fifth Generation Computer Systems Museum":
| https://www.ueda.info.waseda.ac.jp/AITEC_ICOT_ARCHIVES/ICOT/...
|
| As the OP article mentions much (most? all?) of the software
| developed in the project was freely released (under an MIT-like
| license?
| https://www.ueda.info.waseda.ac.jp/AITEC_ICOT_ARCHIVES/ICOT/...
| ) and it seems to be available for download (in source form)
| from here:
| https://www.ueda.info.waseda.ac.jp/AITEC_ICOT_ARCHIVES/ICOT/...
| most of it would only run on the prototype computer they wrote,
| but some of it was ported to Unix.
| mindcrime wrote:
| Wow. This is amazing. I can't say how happy I am that you
| posted this. I've never seen this site before... and I
| explicitly went looking for something like this a while back,
| and somehow did not find this. I've been fascinated by the
| FGCS stuff for a long time, and had always thought there must
| be an archive of that body of knowledge somewhere, but other
| than a few books on Concurrent Prolog I had not managed to
| scare up much of it.
|
| Now, I find that more or less all of it is online and
| available? All the Technical Reports, all the Technical
| Memoranda, and even the software? Amazing. Now if only I read
| Japanese...
| lenerdenator wrote:
| If only the Japanese had been willing to make some of our
| companies' devices for cheap, then they would have found out that
| we'd be perfectly willing to ship them _all_ of our IP for the
| promise of a good quarterly earnings call, regardless of what it
| 'd do to our nation's economic and political future.
| numpad0 wrote:
| IIUC, the Fifth Generation Computer project was more of James
| Bond style hallucination than something worthwhile. The project
| was supposed to produce a completely novel tech stack from
| hardware to user interface, one such that leapfrog and obsolete
| all conventional computer systems: it would run on custom AI-
| optimized parallel CPU designed for a futuristic post-C parallel
| programming language for all-custom AI inference OS and
| applications. It was of course all proudly national made with all
| kinds of scientific theories and inputs from scholars worldwide
| incorporated into it.
|
| It's like trying to make Star Wars technologies happen by sheer
| cash flow and technological optimism. All fever dreams.
| csomar wrote:
| Kinda similar to the current AI bubble, don't you think?
| geysersam wrote:
| Nah the AI hype doesn't seem similar
|
| 1. It hasn't really required or designed new computer
| architecture. Instead it repurposed hardware that was
| designed and tuned over years by the game industry.
|
| 2. It's more empirical. Little to no theory building and more
| "throwing stuff at the wall to see what fits"
|
| 3. Not really "proudly national made" right? I don't think
| Washington minds being world leading on AI, but from where
| I'm standing it doesn't seem to be something that's been
| championed nationally.
|
| That's not to say it's not a bubble
| morninglight wrote:
| I agree with your assessment. Through the 1980's, corporate Japan
| seemed to be on a more cooperative path. However, there were
| major changes during that decade. It's difficult to pinpoint any
| particular cause, but I think it can be linked to the decline of
| MITI's influence. MITI was a key to Japan's dominant role leading
| up to the 1990's.
|
| I believe that MITI's success made it a political target, and
| politics can kill anything. My friends from MITI tended to share
| a similar opinion.
| fnord77 wrote:
| hmmm
|
| > MITI bureaucrats attempted to deny fledgling Sony the $25,000
| the company needed to license transistor technology from
| Western Electric.
| dang wrote:
| Related. Others?
|
| _Fifth-Generation Programming Language_ -
| https://news.ycombinator.com/item?id=36080365 - May 2023 (1
| comment)
|
| _The Fifth Generation Computing Project_ -
| https://news.ycombinator.com/item?id=20569098 - July 2019 (49
| comments)
|
| _' Fifth Generation' Became Japan's Lost Generation (1992)_ -
| https://news.ycombinator.com/item?id=16630174 - March 2018 (47
| comments)
|
| _The Fifth Generation Computer Project in Japan_ -
| https://news.ycombinator.com/item?id=8256961 - Sept 2014 (3
| comments)
| PeterStuer wrote:
| The fifth generation story was hyped to the gills in European
| (and I assume US) AI research circles at the time to increase
| public funding. It was a bit funny as we were all knees deep into
| 'nouvelle AI' as a reaction to the crafted reasoning/logic from
| before, so a thin line had to be walked between 'we need to fund
| this emergence/embodied/subsymbolic research because the
| logic/symbolic reasoning stuff is a deadend' and 'those Japanese
| with their logic/symbolic reasoning approach are going to eat our
| lunch if you do not give us more money'. Good times were had.
| openrisk wrote:
| Very interesting story (more context in the related wikipedia
| entry). Makes you wonder where we are today in terms of computing
| "generations".
|
| > It seemed that logic programming was a key missing connection
| between knowledge engineering and parallel computer architectures
|
| This is still missing. And both ingredients, "logic programming"
| and "parallel computing" are rather arcane and underdeveloped, so
| many decades later.
|
| The current AI hyperventilation illustrates the impact quite
| nicely (algorithms and models have no connection to "ground
| truth", they need expensive and hard to use HPC hardware etc.)
|
| On the other hand we had in the meantime the internet and mobile
| revolutions. While those are not so fundamental from a pure
| computing perspective, they have increased the utility of the
| digital domain tremendously.
|
| It is conceivable that a fifth generation architecture is lurking
| somewhere in the chaotic mess that is the current technical
| landscape. E.g. the internet does enable a certain form of
| massive parallel processing on commodity hardware. And overtures
| toward "logic programming" are even long-time W3C standards...
|
| But in the end we know by now painfully well that digital
| technology is not progressing monotonically. By its nature it
| gets embedded deeply into societies and thus its state reflects
| progress (or regression) of society as much as technical
| breakthroughs.
| zozbot234 wrote:
| The modern take on "logic programming" is just database query,
| and that can definitely be parallelized. The issue there is
| that non-trivial logical inference over a knowledge base (the
| kind of thing that was popular wrt. expert systems and "fifth
| generation" proposals) turns out to be computationally
| challenging unless you heavily restrict the kinds of
| "inferences" that can be done. I'm not sure that this was
| clearly understood back in the 1980s or so.
| fedeb95 wrote:
| don't aim at some far goal, aim at being prepared for whatever
| comes next.
| ggm wrote:
| Some of my first work (as a very junior programmer and
| researcher) in the UK was in part funded and instantiated in the
| reaction to the 5th generation initiatives. The research
| community in the UK reacted to what was called "the Alvey
| report". It led to an explosion of inter uni collaborative
| projects, the newer GEC 400 (?) Series computer was often
| deployed in this work. I think it wound up almost exclusively in
| defence deployment like AWACS roles in the nimrod project. It had
| a bit of an overheating problem and led to many fine "halt and
| catch fire instruction" jokes.
|
| The idea that both the 5th generation projects in Japan would
| work and that without the alvey projects we'd somehow "lose the
| race" was compelling pressure to seek funds and do things. It
| didn't turn out that way.
| rjsw wrote:
| The GEC Series 63 [1] was used in the Alvey programme. I was a
| founder of a startup that used a 63/40 to provide a (pre-
| internet) online service over X.25 links. It was a bit too
| different to a VAX to be able to easily port standard AI
| software to it.
|
| The GEC 4000 series was what was used for the Nimrod AEW and
| for other real-time applications as well as for X.25 networking
| around the UK.
|
| [1] https://en.wikipedia.org/wiki/GEC_Series_63
| sub7 wrote:
| The Japanese made Topre switches and they have already greatly
| increased my happiness.
___________________________________________________________________
(page generated 2024-10-18 23:02 UTC)