[HN Gopher] The End of the Beginning
___________________________________________________________________
The End of the Beginning
Author : nikbackm
Score : 85 points
Date : 2020-01-07 16:40 UTC (6 hours ago)
(HTM) web link (stratechery.com)
(TXT) w3m dump (stratechery.com)
| christiansakai wrote:
| I've been thinking along the same line, albeit on a more personal
| take as a software engineer.
|
| Basically, starting around 15 years ago, there's the
| proliferation of bootcamps teaching fullstack development,
| because software startups were the new hot thing, and they
| desperately need generalist engineers that were capable of
| spinning up web app quickly. Rails was the hot thing those days
| because of this as well. Hence we saw many new grads or even
| people who change careers to fullstack development and bootcamps
| churning out these workers at an incredible pace (regardless of
| quality) but the job market took it because the job market was
| desperate for fullstack engineers.
|
| During that time, the best career move you can do was to join the
| startups movement as fullstack engineers and get some equity as
| compensation. These equities, if you are lucky, can really be
| life changing.
|
| Fast forward now, the low hanging CRUD apps (i.e., Facebook,
| Twitter, Instagram, etc) search space has been exhausted, and
| even new unicorns (i.e., Uber) don't make that much money, if
| they do for that matter. Now those companies have become big,
| they are the winners in this winner take all filed that is the
| cloud tech software. Now these companies these days have no use
| for fullstack engineers anymore, but more specialists that do few
| things albeit on a deeper level.
|
| Today, even the startup equity math has changed a lot. Even with
| a good equity package, a lot of the search space has been
| exhausted. So being fullstack engineers these days that join
| startups don't pay as much anymore. Instead, a better move would
| be to try to get into one of these companies because their pay
| just dwarfed any startups or even medium / big size companies.
|
| Just my 2c as someone who is very green (5 yrs) doing software
| engineering. Happy to hear criticism.
| onlyrealcuzzo wrote:
| My understanding is that startups never paid well. Sure, if you
| got lucky and were employee #7 at Facebook, it paid off great!
| But even during that time frame, working at startups instead of
| MS or Google was not a good proposition. And even during DotCom
| and the rise of Microsoft, it paid a lot better to work at IBM
| than these classes of startups.
| timClicks wrote:
| Isn't the perception of a saturated market persistent though? I
| mean, there were many social media apps when Facebook started.
| Twitter was created when microblogging had already become a
| trend.
| jonny_eh wrote:
| Anecdotally, it seems like the rate of disruption in this
| space has been steadily slowing down.
| zozbot234 wrote:
| Software is still eating the world, and there will be plenty to
| eat for a long time. Cars (the foremost example in the OP) had
| basically eaten the world by the 1950s (sometimes even in a
| fairly literal sense).
| mooreds wrote:
| The essay reminds me of The Deployment Age:
| http://reactionwheel.net/2015/10/the-deployment-age.html
| tlarkworthy wrote:
| Thats a very bold claim, that goes against Ray Kurzweil's
| hypothesis tech is accelerating. Maybe (unlikely) that
| cloud/mobiles is the end game for silicon. But what about
| quantum? What about biological? What about Nano? What about AI?
| Literally there are a ton of potential generational changes in
| the making that could turn everything on its head _again_
| the_af wrote:
| Why is Ray Kurzweil's hypothesis particularly important to
| contrast other hypotheses against? What sets it apart in
| relevance and/or authority?
| throw_14JAS wrote:
| Because it's evidence is pretty straightforward: you can take
| wikipedia's list of important inventions and plot their
| frequency on a chart.
|
| Of course, there are debates around which inventions count as
| significant. And there is recency bias.
|
| Never underestimate the power in something easy to
| communicate.
| beat wrote:
| I think population growth also factors in. Population is
| leveling off. In the past century, the global population
| has quadrupled, so there are four times as many people to
| invent things in raw numbers alone. But global population
| will increase no more than 50% in the next century, which
| means we aren't creating a lot more inventors than we are
| now.
| davnicwil wrote:
| This is a good point, but also consider the proportion of
| the global population who have the opportunity to become
| inventors is hopefully going to grow over the next
| century. As a result, the absolute number of inventors
| may grow faster than by just growing the overall
| population size.
| tlarkworthy wrote:
| It's very well known and makes a compelling argument that
| tech progress has been accelerating since the Stone age.
| AnimalMuppet wrote:
| Well known is irrelevant. "Has been accelerating" is also
| irrelevant. _Will continue to accelerate to something very
| close to infinity_ is the relevant part. There are plenty
| of us who do not find Kurzweil 's argument on that topic to
| be at all compelling.
| the_af wrote:
| I agree Kurzweil's hypothesis is well known... within
| techie/Silicon Valley circles, somewhat like the
| Singularity -- a related concept -- is common in those
| circles as well. Regardless, there's no particular weight
| to Kurzweil's hypothesis, tantalizing as it may seem, and
| it's not reasonable in my opinion to use it as a measuring
| stick of other hypotheses as if Kurzweil was a _proven_ and
| _acknowledged_ authority on this topic.
|
| Likewise, if someone said "human aging and death are
| unavoidable" this wouldn't be bold just because Kurzweil
| has written a lot about immortality.
| gfodor wrote:
| I think his hypothesis deserves more credit than that --
| many of his predictions have come to pass, and many of
| those that have, at the time they were predicted, seemed
| somewhat far fetched since they were firmly in the realm
| of science fiction.
| gfodor wrote:
| I think it will, at best, be a semantic argument in retrospect.
| The companies highlighted are all clearly defined as being
| bolstered by computing technology. But what about next
| generation, huge companies that are bolstered by computing and
| _other_ technologies fused together? For example, if a company
| manages to create a brain-computer interface that gains global
| adoption and equivalent valuations to the existing tech giants,
| but the software layer is a mashup of, by that time,
| commoditized services from the existing tech giants who fail to
| enter this industry, does it count?
| davidivadavid wrote:
| This myopia really puzzles me.
|
| It seems like the whole analysis is predicated on the idea
| that technology = software made in Silicon Valley, with
| unimportant secondary factors. That 3M and ExxonMobil are not
| "tech" companies because they don't make iPhone apps.
|
| _Every_ company is a tech company, not because we 've had
| computers for a while, but because technology is what we
| build to get what we want.
|
| These kinds of narrow, myopic, siloed takes miss the forest
| for the trees.
|
| If you think the epitome of human evolution is going to be
| people looking at bright rectangles for eternity, you haven't
| been paying attention to what technologists are doing.
| dsalzman wrote:
| I don't think the claim is technology in general, but non
| quantum based computing.
| jdmichal wrote:
| I don't think that's accurate. The article goes through pains
| to discuss that sometimes castles are simply routed around.
| Quantum computing would potentially be one of those.
| bduerst wrote:
| It's a substitute computing product though - kind of like
| electric cars for combustion cars.
| legitster wrote:
| I've been reading Zero to One, and one of the ideas the book
| pitches is that monopoly and innovation are two sides of the same
| coin. Only monopoly-like companies have time and money to dump
| into innovative products (Bell, GE, IBM, Google). And people only
| invest in an idea if they think they can profit from it (look at
| how crucial a patent system was for the industrial revolution).
|
| Competition is important, but to drive efficiency - weed out bad
| ideas and bring down costs of already created innovations. But
| the thing that usually drives monoliths out of business is... new
| monoliths.
|
| The somewhat contrarian takeaway is that some (keyword) amount of
| consolidation is good.
| gz5 wrote:
| >And, to the extent there are evolutions, it really does seem
| like the incumbents have insurmountable advantages...
|
| By definition, doesn't it _always_ seem like this?
|
| Jim Barksdale (Netscape) said there are 2 ways to make money -
| bundling and unbundling. What can be unbundled from the incumbent
| bundles, in order to be offered in a more fit-for-purpose way, or
| with a better experience?
|
| How might that answer change if the world's political structure
| changes? How might that answer change if processing, storage and
| networking continue their march towards ubiquitous availability?
| tudorw wrote:
| Don't agree, comparing histories is not a reliable way to predict
| the future, I think we'll see the growth of governance level
| disruption, a pushback that will encourage home grown solutions
| for countries that are not necessarily aligned with US interests.
| That field is wide open and growing!
| camillomiller wrote:
| Policy driven disruption is the only option I see to break the
| cycle. Let's see.
| whatitdobooboo wrote:
| I think if you abstract away the specific companies mentioned and
| stuck to the technology, the point about people building on top
| of already "accepted" paradigms is a good one, in my opinion.
|
| The rest doesn't really seem to have enough evidence for such a
| bold claim.
| mirimir wrote:
| > What is notable is that the current environment appears to be
| the logical endpoint of all of these changes: from batch-
| processing to continuous computing, from a terminal in a
| different room to a phone in your pocket, from a tape drive to
| data centers all over the globe. In this view the personal
| computer/on-premises server era was simply a stepping stone
| between two ends of a clearly defined range.
|
| Sure, that's what happened.
|
| But what jumps out for me is that, at both ends of that range,
| users are relying on remote stuff for processing and data
| storage. Whether it's mainframes or smartphones, you're still
| using basically a dumb terminal.
|
| In the middle, there were _personal_ computers. As in under our
| control. That 's often not the case now. People's accounts get
| nuked, and they lose years of work. And there's typically no
| recourse.
|
| As I see it, the next step is P2P.
| oflannabhra wrote:
| I'm not exactly sure where I fall on this. Ben is a really smart
| guy (way smarter than me), but I feel like this could be a
| classic case of hindsight.
|
| Now, looking back, it makes sense that the next logical step
| after PCs was the Internet. But from each era looking forward,
| it's not as easy to see the next "horizon".
|
| So, if each next "horizon" is hard to see, and the paradigm it
| subsequently unlocks is also difficult to discern, why should we
| assume that there is no other horizon for us?
|
| I also don't know if I agree that we are at a "logical endpoint
| of all of these changes". Is computing _truly_ continuous?
|
| However, I think Ben's main point here is about incumbents, and I
| agree that it seems it is getting harder and harder to disrupt
| the Big Four. But I don't know if disruption for those 4 is as
| important as he thinks: Netflix carved out a $150B business that
| none of the four cared about by leveraging continuous computing
| to disrupt cable & content companies. I sure wasn't able to call
| that back in 2002 when I was getting discs in the mail. I think
| there are still plenty of industries ripe for that disruption.
| pthomas551 wrote:
| Was it really that hard to predict the Internet? SF authors
| picked up on it almost immediately.
| oflannabhra wrote:
| I'd say networking was not incredibly difficult to predict,
| but the businesses and products it allowed for (and how we
| use them) was very difficult.
| marcosdumay wrote:
| > it seems it is getting harder and harder to disrupt the Big
| Four
|
| Microsoft, IBM, Oracle... What is the other one?
|
| Or, right, wrong decade.
|
| (My point is, it completely not obvious if it is getting harder
| to disrupt the incumbents.)
| umeshunni wrote:
| > Microsoft, IBM, Oracle... What is the other one?
|
| Cisco, of course.
| ghaff wrote:
| It was actually Oracle, Sun, Cisco, and EMC who were the
| four "horsemen of the Internet" in the run up the dot-com
| bubble.
| oflannabhra wrote:
| The conclusion of the article is that it _is_ getting harder
| to disrupt the incumbents. I 'm saying that regardless of
| whether it is or isn't, there are still lots of new companies
| to come that can take advantage of technology to disrupt
| other, old-guard incumbents.
|
| That, I think is where the metaphor Ben uses breaks down. The
| automobile is a single idea (move people around with an ICE).
| Tech is more like the ICE than the car. So, there might not
| be much disruption to consumer hardware (Apple) companies, or
| search (Google) companies, or cloud computing (Amazon,
| Microsoft) companies. But there will still be lots of
| disruption to come as tech (just like the ICE) gets applied
| to new features.
| myblake wrote:
| Isn't that kind of his conclusion too though? It matters in as
| much as we're less likely to see new general purpose public
| clouds come into play, but he didn't seem to predict there was
| no more room for change in the industry, just that were
| unlikely to see those incumbents toppled from certain
| foundational positions in the ecosystem.
| oflannabhra wrote:
| Much of Ben's writing recently has been on the topic of
| regulation and anti-trust, specifically in relation to tech
| companies. If I had to summarize his thesis, I'd say it's
| something along the lines of: "Previous antitrust regulation
| prioritized price. Tech allows for better products by nature
| of aggregation and network effects, and to promote
| competition, we need a new prioritization in our regulation".
|
| So, I see this article as being a part of that thread. The
| conclusion is that the Big Four are _not_ going to get
| disrupted, which is bad, and _drawing some conclusions_ we
| need a new framework of antitrust to allow for it. I might be
| putting words in his mouth, but I don 't think it is really
| that much of a jump if you read his body of work, especially
| recently.
| jdmichal wrote:
| Until I have something resembling Iron Man's Jarvis with at
| least a subvocal interface, I think there's still a long way to
| go for "continuous" computing. I currently still have to pull
| out a discrete device and remove myself from other interactions
| to deal with it. If I'm not on that device all the time, then I
| don't have continuous computing. Maybe continuously _available_
| computing is more accurate?
| repsilat wrote:
| Right -- today you need to remember to charge your phone, you
| don't take it everywhere (and don't have signal everywhere,
| especially internationally), and you need to take it out of
| your pocket to use it, and type into it with your thumbs
| (though voice "assistants" are here, and some people get use
| out of them.)
|
| The end-goal is being able to talk to anyone at any time,
| remember anything you've seen before, and know the answer to
| any question you can phrase that someone has already
| answered.
|
| (Now, you might say that parts of it sound less than ideal,
| but I think we'll get there by gradient descent, though may
| be with some legal or counter-cultural hiccups.)
| solidasparagus wrote:
| Bah. He took three datapoints, built a continuum out of it and
| says that since the third datapoint is at the end of his
| continuum, we must be at the end.
|
| But this doesn't fit any of the upcoming trends. The biggest
| current trend is edge computing where cloud-based services
| introduce issues around latency, reliability and privacy. These
| are big money problems - see smart speakers and self-driving
| cars. The cloud players are aware of this trend - see AWS
| Outposts that brings the cloud to the needed location and AWS
| Wavelength where they partnered with Verizon to bring compute
| closer to people.
|
| But privacy in a world full of data-driven technology is still
| very much an unsolved problem. And most of the major technology
| players have public trust issues of one sort or another that
| present openings for competitors in a world where trust is
| increasingly important.
| Animats wrote:
| His graph conveniently stops in the 1980s. Since then, there have
| been many new US car companies, mostly in the electric of self-
| driving spaces. Lots of little city cars, new imports from China,
| too.
| the_watcher wrote:
| He specifically mentions excluding imports. Outside of Tesla,
| what are the new American car companies that made any kind of
| mark?
| kevin_thibedeau wrote:
| Most of those are NEVs with a 25MPH max speed to bypass safety
| regulations. $20K golf carts.
| ropiwqefjnpoa wrote:
| The dealership model really helps manufacturers keep a tight
| reign on the market, look at all the trouble Tesla had.
|
| In a similar vein, Apple, Google and Microsoft control the medium
| and have grown so powerful, I can't imagine there ever being a
| new "Google" that comes about the old grass roots method.
|
| Someday Apple will be bought though, probably by Facebook.
| streetcat1 wrote:
| This is wrong on merit, and I am not sure why it is presented
| this way.
|
| The difference between a car company and a software company is
| economy of scale. I.e. economy of scale dominate the physical
| world but does not exist in the software world since I can
| replicate software at zero cost.
|
| In addition, new tools and new processes for software has
| increased the productivity times fold, which means that you need
| fewer developers for new software.
|
| I predict two shifts in the tech world:
|
| 1) Move to the edge. Specially for AI, there is really no need
| for a central public cloud due to latency, privacy, and dedicated
| hardware chips. I.e. most of AI traffic is inference traffic
| which should be done on the edge.
|
| 2) Kubernetes operators for replacing cloud services. The value
| add of the public cloud is managing complexity.
| zebrafish wrote:
| Your point contradicts itself, does it not? Movement to the
| edge necessitates edge hardware and edge personnel to manage
| the hardware.
|
| Network effects are THE factor in software because the marginal
| cost tends towards zero with each incremental user in the
| network. The edge adds cost per node.
|
| Up until the point that users are paid to connect to the
| network and/or the network is directly linked to the user with
| the I/O line completely obviated, the economics of hardware and
| management underlying the network will tend towards economies
| of scale... which is the point Ben is trying to make.
| mooted1 wrote:
| If you read more of Ben's writing, he talks extensively about
| how software companies dominate market share through network
| effects and vertical integration.
|
| You don't hear him talk about economies of scale because
| marginal costs are negligible for software companies. Besides,
| network effects and vertical integration are sufficiently
| powerful to control the market.
|
| > In addition, new tools and new processes for software has
| increased the productivity times fold, which means that you
| need fewer developers for new software.
|
| There are other barriers to entry besides the cost of writing
| software, like product, sales, operations, and most
| importantly, network.
| streetcat1 wrote:
| However, the network effect in tech can be leapfrogged due to
| the zero marginal cost (as shown in this post). I.e. what
| network effect do you get from doing ML inference in the
| cloud?
|
| The case for big tech today is still the economy of scale and
| not network effects (maybe facebook have those, but it exists
| only if the interface to facebook does not change).
|
| The big tech players have economy of scale, due to their
| ability to use automation and offload the risk of managing
| complexity (I.e. one AWS engineer can manager 1000's of
| machines with AWS software).
|
| No wonder, that the software that manages the public cloud is
| still closed source.
|
| However, with Kubernetes operators, there is a way to move
| those capabilities into any Kubernetes cluser.
| mooted1 wrote:
| Did you actually read the post?
|
| > The case for big tech today is still the economy of scale
| and not network effects (maybe facebook have those, but it
| exists only if the interface to facebook does not change).
|
| This is only true if you believe that the greatest cost of
| developing software is running hardware. The greatest cost
| of developing software is developing software. Not only are
| economies of scale in compute management negligible except
| at massive scale, the cost of compute has declined
| dramatically as the companies you've described have made
| their datacenters available for rent through the cloud. Yet
| the tech giants persist.
|
| Facebook, Google, Netflix, Amazon all have considerable
| network effects that you're not considering. For each of
| these companies, having so many customers provides benefits
| that accrue without diminishing returns, giving them a firm
| hold on market share. See
| https://stratechery.com/2015/aggregation-theory/
|
| Ben is saying that the only way to topple the giants is by
| working around them and leveraging new computing
| technologies better than them. He makes the (admittedly
| speculative) case that this is no longer possible because
| we can't bring compute any closer to the user than the
| mobile devices.
|
| > However, with Kubernetes operators, there is a way to
| move those capabilities into any Kubernetes cluser.
|
| Kubernetes, at the scale of technologies we're discussing,
| is a minor optimization. Introducing k8s costs more than it
| helps far until far into a company's infra maturity. Even
| if most companies deployed k8s in a manner that
| significantly reduced costs, it's not enough to overcome
| the massive advantages existing tech companies have
| accrued. Not to mention all of the big tech companies have
| internal cluster managers of their own.
| mr__y wrote:
| >this is no longer possible because we can't bring
| compute any closer to the user than the mobile devices.
|
| this is based on a very dubious assumption that bringing
| compute closer is the only path for innovation.
|
| and even that is not true, you could imagine compute
| being even closer with a direct brain interface (actually
| you could consider google glasses to be an attempt at
| bringing compute closer)
| streetcat1 wrote:
| I don't think that the amount of current customers is any
| indication of network effects or any other kind of moat.
|
| See: Walmart -> Amazon, Nokia->Apple, MSFT -> Andriod.
|
| I mean, what more of network effect did MSFT had in the
| 90's. It was dominating both the OS layer AND the app
| layer (office). And yet, it does not have ANY share in
| mobile.
|
| Kubernetes is not minor optimization if you think about
| what it is. Yes, if you see it as mere container
| orchestration. But it is the first time that a widely
| deployed, permissionless, open API platform exists.
| yodon wrote:
| Your comment would work equally well without the initial
| inflammatory "Did you actually read the post?" opening
| line.
| foobiekr wrote:
| It really doesn't make a lot of sense to do AI at the edge (in
| terms of the various edge providers).
|
| But then a lot of edge cases don't make a lot of sense. The
| best edge use cases are fan-in (aggregation and data
| reduction), fan-out (replication and amplification -
| broadcasting, conferencing, video streaming, etc.) and caching
| (which is just a variant of fan-out).
|
| The rest of the cases are IMHO largely fictional - magical
| latency improvements talked about in the same context as
| applications that are grossly un-optimized in every way
| imaginable, AR/VR, etc. Especially the AR/VR thing.
|
| Beyond that the only thing left is cost arbitrage - selling
| bandwidth (mostly) cheaper than AWS.
|
| What's the use case for moving inference to the edge? Most of
| the inference will in fact be at the edge - in the device,
| which has plenty of capacity - but that's not the case you're
| describing.
| streetcat1 wrote:
| Why would you run AI in the cloud? It is a closed, expensive,
| high latency, etc. You might want to train in the cloud,
| maybe.
|
| For inference, I See 90% on the edge (I.e. outside of the
| clouds).
| xapata wrote:
| > 1) Move to the edge. Specially for AI, there is really no
| need for a central public cloud due to latency, privacy, and
| dedicated hardware chips. I.e. most of AI traffic is inference
| traffic which should be done on the edge.
|
| Inferencing is done at the edge, but training must be done
| centrally.
| Slartie wrote:
| Right now, the only market participant I see doing some
| inferencing at the edge is Apple with its photo analysis
| stuff that runs on the phone itself.
|
| Anyone else is busy building little dumb cubes with
| microphones and speakers that send sound bites into clouds
| and receive sound bites to play back (heck, even Apple does
| it this way with Siri). Or other dumb cubes that get plugged
| into a wall socket and that can switch lights that you plug
| into them by receiving commands from a cloud (even if the
| origin of the command is in the same room). Or dumb bulbs
| that get RGB values from a cloud server which inferred
| somehow that the owner must have come home recently and which
| then set the brightness of their RGB LEDs accordingly. Or
| software that lets you record sounds bites, send them into
| the cloud and receive transcripts back. Or software that
| sends all your photos to a cloud library where it is scanned
| and tagged so you can search for "bikes" or whatever in your
| photos.
|
| No matter what you look at in all that stuff that makes up
| what consumers currently consider to be "AI", it does
| inference (if it even does anything like that at all) on some
| cloud server. I don't like that development myself, but
| unfortunately that's how it is.
___________________________________________________________________
(page generated 2020-01-07 23:00 UTC)