Can the Internet of Everything bring back the High-Growth Economy?

That is the new paper by Michael Mandel, interesting throughout, here is one excerpt:

…we estimate that the Internet of Everything could raise the level of U.S. gross domestic product by 2%-5% by 2025. This gain from the IoE, if realized, would boost the annual U.S. GDP growth rate by 0.2-0.4 percentage points over this period, bringing growth closer to 3% per year. This would go a long way toward regaining the output—and jobs—lost in the Great Recession.

And what is the Internet of Everything?:

The Internet of Everything is about building up a new infrastructure that combines ubiquitous sensors and wireless connectivity in order to greatly expand the data collected about physical and economic activities; expanding ‘big data’ processing capabilities to make sense of all that new data; providing better ways for people to access that data in real-time; and creating new frameworks for real-time collaboration both within and across organizations.

Read the whole thing.


I question whether it would actually translate to jobs. In fact, the internet of everything would make it easier to automate even more jobs. I do think it will increase gdp, but it will also exacerbate the discrepancy between those who produce for the IoE and those who are simply its consumers.

Agreed, I see the IOE compounding growing unemployment and income inequalities. IOE will make everything more efficient and it will reduce waste, but it will only create computer science employment and maybe some one time sensor installers.

You fail to appreciate the wealth created by replacing trained electricians who have learned how to find problems by driving around and walking around and crawling around to find problems which they fix in 30 minutes by replacing a part. The highly skilled electrician who can find the problem in an hour gets paid $50 an hour because he is more productive than the $25 an hour guy who takes three hours.

By the IoE, the problem and fix can be isolated and the electrician will be so much more productive, able to fix the problem in 30 minutes. Given the cost of finding the problem is $50, the cost is cut $50, but the customer value is the same, so the price of finding the problem will remain at $100. As the electrician is much more productive in fixing the problem, the $25 a hour electrician will be richer because he will replace the $50 an hour electrician for all the part replacement. And the $50 an hour electrician will retire because he was smart enough to find problems fast so he was smart enough to have bought stock the triples in price because profits go from $50 to $100 by eliminating his job. And the smart customer smart enough to buy IoE products was smart enough to buy IoE stocks, so they are richer after they pay $100 for their IoE investment finding the problem faster boosts their stock prices.

The only people who suffer are the people who nature deems must suffer, the stupid who failed to buy IoE stock.

Of course, the Steve Jobs of IoE will ensure that the customers of IoE will pay $100 a month to have the IoE finding the problems and fixing them for $100 without any human labor. The higher profit means IoE is higher status so you should be happy paying more for high status IoE which eliminates labor costs. Look at how much wealth that creates when IoE stock prices increase ten fold.

What we need is small government which eliminates poverty by erasing it by not paying for it, If you stock paying for food and housing for the hungry and homeless, nature will eliminate the hungry and homeless and thus reduce unemployment.

Exactly. Ignoring the fact that this is more a solution searching for a problem, increases in efficiency do not lead to more wealth when the increases are coming at the cost of income. Labor is inherently inefficient but its also basically your entire market.

I explain it like this, when automated factories use robots to load self driving trucks to stores with self checkou who exactly is going to buy the goods you make? The 10 to 20% (at most) of people still needed cannot support that many businesses. There is despite what the unlimited demand crowd says actually only a fixed need for goods and services.

And worse if you don't get a revolution by th ehave notes you might get a Japan style revolt with massive population shortfalls and that will be well not good at all.

Why would you question that? This GDP thing is awesome! 95% of it has gone to 1% of the population since the "recovery" started! What number could be more important?

P != NP. That means we can't automate everything and baring the singularity many humans will need to overseer the algorithms and machines that work on NP problems. And given IoE, there should be an incredible amount of job opportunity here.

Nope. Most people don't have the mental horsepower or the right personality index to oversee algorithms or NP problems. You cannot train the 90-110 crowd who make up most of the population to do this kind of work any more than Matt Rollof can play in the NBA.

The study glosses over a large problem: existing "big data" architectures and computer science like Hadoop are fundamentally inadequate for the purposes of building this envisioned Internet of Everything. It focuses on the necessity of investing in sensor infrastructure but overlooks large gaps in basic software technology required to implement this. It is a hardware-centric view of the topic, which is not surprising given the contributing companies.

The study simply assumes that a radically novel big data platform, from completely new computer science foundations than all existing such platforms, will be invented any day now. In fact, it is not clear that the authors even recognize that this issue exists. Those kinds of quantum leaps are relatively rare in computer science so it seems imprudent to assume that one will occur on cue.

I've been heavily involved in the technology around the nascent Internet of Everything for 7-8 years now. A large part of the reason it is still "nascent" is that the software technology available to support it has remained stagnant over all of these years despite existing latent demand.

The data demand depends a lot on what people want to do with their "things." I can't think of a lot of data intensive things I'd want (predictive thermostats?) but to be honest, I'm not likely to spring for more connected things at all. Even as a programmer, I don't think I need it.

We need the data to be there first, so demand can actually push us over the CompSci problems. I work on a Quantify Me device's platform, and the big data applications are surface level. We aren't collecting near the right amount of data to actually help people. We can make broad generalizations, but Hadoop is all we need. I'd be curious to know what field has the requisite amount of data that actually needs a better algorithm than MapReduce.

Any that actually needs to react to the data quickly: MapReduce is a lumbering giant that lets you use tons of processing power, but won't get you anywhere if you want to look at a lot of data fast. Most solutions built around removing MapReduce latency are all about doing partial calculations using MapReduce, with lag, and then using something else on top to get fast queries. But if you need to consume your data fast, not just get data returned fast, hadoop won't do what you want.

And yeah, NDAs and all that.

Oh, is that all!


I'm moving to Antarctica.

I can see it costing that much to build and maintain. Then what? The only sector that would benefit from the ability to know everything are the regulatory agencies.

If regulatory agencies benefit disproportionately from the Internet of Everything then that needs to be taken into account when estimating its impact on U.S. GDP growth rates. A Panopticon of ubiquitous networked sensors might lead to slower economic growth.

Expanding on Derek's post, the only ones that would benefit from the purported ability to know everything would be central planners under a command and control economy. I've read the article from the Progressive Institute, and one thing I really miss from their analysis is a proper thinking-through of what it could mean, rather than what they would like it to be.

First and foremost, this reduces consumers to their data, rather than their needs and desires: not only are there privacy concerns (and don't think for a second that Chrome's incognito window will protect you from the government knowing what sort of porn you may prefer), but your behavioral patterns are then available for analysis by whoever is willing to pay the user fee. The ability to match up the anonymous statistics with the real world is, for me, largely a given, considering that at the end of the day it will only be a question of how much someone is willing to pay for that information (and generally speaking, the ability to know what a potential customer actually consumes is a goldmine of information just waiting to be exploited in order to manipulate consumers into buying from x instead of y.

Further, and this is the real crux of the problem for me, quis custodiet ipsos custodes? Who is watching the watchers? With such a wealth of information about individual behaviors at their disposal, those whose desires to do good overweigh the realities of civil rights will be given, effectively, carte blanche to target you with their behavior-changing legislation and propaganda. Will you really be looking forward to hearing from a future Bloomberg who has reached a power fulcrum where they can make what you like doing - say, reading Marginal Revolution in your spare time - either illegal, or harass you for doing so? Perhaps couple your internet speed with your browsing patterns and change your bandwidth to push you away from sites like Marginal Revolution to sites like MoveOn? Perhaps push you up on the list of IRS audits? Or simply let your children's school teachers know that little Johnny comes from a family that dares to question and doesn't toe the line? Change what is deducted from your salary in taxes to account for what someone else believes will do you harm down the road, based on what passes for science these days? Take a Sunday drive and see your gasoline price increase just for you the next time you tank up because someone doesn't want you to behave the way you want to? The possibilities of the perfectly transparent consumer in a fully interconnected world are not all on the upside: political control - and given the article is from a "progressive" organization, you can be sure that is the motive behind this all - sold as an increase in GDP and jobs.

Seriously: the Internet of Everything, especially a government-run one, means that the playing field will be vastly tilted against you as an independent consumer and private person. Don't think for a second, either, that the data generated would be anonymized and publicly available, or you could opt out: this is far too useful a tool for targeted marketing, political parties, governments, people with a grudge, and everyone inbetween.

Example: in John Brunner's novel Shockwave Rider, which basically has an internet of everything, the protaganist turns away a private investigator seeking to recover a runaway so that she can be brainwashed into doing what her parents want her to. In retaliation, the investigator uses his contacts to literally erase the electronic identity of the protaginist (one of his many) and cancels his utilities, collapsing the inflatable church where he lives (read the book). That's the downside of an internet of everything.

The Internet of Everything sounds like a wonderful data idea. In reality, the dangers to civil rights (of us all, not merely the left or the right) is enormous. Even worse, it might bring back the idea that now that we know the true interconnectedness of the economy, exactly what each and everyone of us consumes and produces (a ginormous input-output table, as it were), that now is the time we can try True Socialism and plan the economy in the name of social justice, in the interests of a progressive society, each according to their abilities, each according to their needs.

We all know where that ends.

If we want to worry about privacy we don't have to go further than our current, existing, Internet. That good old Internet has your shopping history, which is actually more important then the number of times you open the fridge.

I think the hardware hacker's Internet of Things is being projected into an unlikely Internet of Everything. I mean, there are Nest thermostats, and I suppose that if Nest has house traffic patterns it could be subpoenaed, but certainly hey and every "of things" player is going to also see their data trove as their proprietary advantage. It will be fragmented, sharded, across owners. Similarly some turbine could produce some gigabytes of data, but there will be people who feel very strongly that they own that data.

Maybe I've been around long enough to hear these sales pitches already. Sensors in everything exist already in certain domains. Management of buildings for example. It isn't inexpensive, and it has returns only if tied to some control. They have input points as well as output points. They are called control systems.
As for monitoring, great idea. Say you have a grocery store with refrigerated cases. Temperatures are read, and if they are outside a specified range, an alarm is raised. Someone the other side of the continent phones the manager to tell him the deli case is warm. They get dozens of calls like this every day. Included in the dozens of spurious alarms is one that occurs once and says leak detector alarm. Five days later half the store is alarming because the system is down.
Too much information and no understanding. I think the current US administration is facing this hard wall in almost all their policy initiatives. The ones who have decision making authority are getting too much information strained through some algorithm that was designed long ago and whose advantage was cheapest, first or the only one that worked. the systems and management trained the local folks who know to keep their mouth shut, or the locals are stupid and poor so would be of no help in any case.

Big data is seeking unicorn farts. The low hanging stuff, supply chain, warehouse, etc has already been done. The remaining stuff is hard because each application is different, specialized or small. No payback.

Regulatory agencies would love it. Imagine all the minute details of people's lives and businesses they could get their fingers into. When you are an apparatchik the apparat just needs more power and information.

We don't need to look at fiction to see how this works.

This is alarmism. You have no idea what the Internet of Everything looks like. Nor are you aware that privacy is a legal fiction that is fast losing it's ability to function.

To give a glimpse, the Internet of Everything means that we can correlate your body's statistics in the lead up to your most recent heart-attack with millions of other heart-attacks so that the best procedures and medications are known before you even enter the emergency room.

Yes what you say is true . I suspect that it will with near 100% certitutde also bring with it unimaginbly totalitarian societies

Have the wrong opinion, do the wrong thing and everyone knows is a quick way to a stagnant society

Wall Street Journal called it "A Nation of Sullen Parnoids" but the real result withh be much worse, a world of petty little DPRK's

The gated communities of the future will be those that cut you off from all this surveillance.

The NSA opponents want us to live in poverty.

I award you 1/2 of a Stewart Baker for this.

This sounds like the basis for a frightening dystopian sci-fi novel.

Hazel - See Accelerando by Charles Stross.

Hazel, read Accelerando by Charles Stross. And if you hate powerpoint, read the Jennifer Morgue too..

But a network of sensors in a factory, for example, hooked to powerful data analysis capacity, could greatly improve the productivity and flexibility of production, and perhaps lead to a rebirth of manufacturing in the U.S.

It doesn't sound as if the authors of this paper are actually familiar with modern manufacturing plants. Because modern factories already have a substantial network of sensors that are hooked up to a fairly large array of PLC's, switches, routers, servers, sensors, databases, web servers, etc. I'm not sure at what level they are talking about, but the site I'm at currently historizes over 3,000 device outputs per second. And that's only what gets logged. The controls systems I/O count is measured in the hundreds of thousands of tags per second. There are far more computers than people in a modern plant.

Moreover, until recently the technology to build inexpensive wireless links that could reliably and cheaply transmit large amounts of data did not exist.

Most modern plants have multiple plant-wide industrial (high power) wireless networks. However, industrial gear is still quite expensive, so there could be some low hanging fruit there. But keep in mind that modern plants have a vast infrastructure of ethernet and various control nets.

But the actual production process itself within factories and at construction
sites has remained mostly disconnected from the Internet.

No, sorry that's just wrong.

The Internet of Everything, just sounds like a buzz word. What exactly does the Internet of Everything do, again? (Other than help sell GE products).

Comments for this post are closed