I've been reading "Darknet" via Kindle and it has this idea of "autonomous companies" that do high frequency trading, shuffle around money, and eventually even start "replacing" people by email and video chat.
As we have more and more of these systems tied together - usually via APIs - I think it's getting more more likely where things can "just work" transparently.
If I need gas for the car, it can go get it, billing to my account on file. If I need diapers for the kid, the system detects that, orders them automatically, delivers them automatically, etc. There are few humans involved anywhere.
It's a fascinating and has society-shaking implications.
I agree that will be a start. But knowing that you need more diapers isn't particularly revolutionary - just order the same ones, we are almost there - just have to connect the cabinet to Amazon dash.
What will blow my mind is when we have computers that can take an idea - "shit, I just had a kid, I need diapers" - and successfully mimic my purchase criteria and find the diapers that I would've found in my own research process. It wouldn't need me to define my purchase criteria - it would already know me well enough to successfully predict in most cases.
At some point we'll have mind readers. Be careful of thinking you want a jelly donut - Postmastes is already on the way.
> At some point we'll have mind readers. Be careful of thinking you want a jelly donut - Postmastes is already on the way.
For years, Target has been trying to find likely pregnant women based on shopping patterns. There's a strange anecdote where a daughter received pregnancy-related ads and it turned out she was actually pregnant! NYT article discussing this: http://www.nytimes.com/2012/02/19/magazine/shopping-habits.h...
Ambient Intelligence is the phrase a lot of people are searching for when talking about this stuff. There's also a bunch of stuff about Second-order Cybernetics and Adaptive and Collaborative Multi-Agent Systems people should seek out and read.
I've co-opted some terms and am now speaking of this as Second-order IoT. Where previously we were concerned about our personal networks and systems, this next wave will be concerned about how these systems work together and enlist one another to carry out some task or contract autonomously. Chaining these contracts is where the cool stuff comes in. Observing and influencing these systems and letting them create efficient networks on our behalf is the reason we should be building this.
Agents are what people are talking about when discussing bots, and agents are basically the AI program that can conscript if not completely become part of the operating service on a device. Like little plug 'n play souls that carry along information about you and what you need within some context.
For instance, when you step out of your autonomous vehicle, it can use a service (Playa) to look up physical interfaces offering recharging. It will be able to negotiate and transact a recharge on your behalf, then contract itself out to an uber-like service while you sleep to help pay for itself.
I'm building the infrastructure/platform to support all of this, giving rise to a fully autonomous global logistics and supply chain -- I call it Playa (http://getplaya.com/) It's at first an open service exchange (marketplace) for autonomous intelligent agents to learn from us, and act on our behalf with other services and agents (both digital and physical). - If Artificial General Intelligence is the ruler, then Playa is its throne.
If you're interested in building this vision of the future, please reach out. I'm happy to grab a coffee in San Francisco and chat more about the current state as well as long term social and political implications of this if you'd like.
The next step would be doing these interactions - service registration, search, negotiation, usage, payments, certification etc. without a trusted, centralized intermediary. Have a look at the https://ethereum.org project to see how this could work out. If these kind of distributed systems for messaging, data distribution and storage as well as consensus scale, intelligent agents will be able to interact with services and each other faster, efficiently and more secure than before.
I definitely think ethereum is a cool project. Unfortunately, I'll be a contrarian and say I don't think the answer is a completely distributed system. I think there is a few mixed options, but the totality must collapse to a central authority at some point.
I'm not a maximalist either. Distributed networks are always less efficient than hierarchical systems at specific tasks. Some projects, such as the Bitcoin lightning network or Ethereums' Raiden therefore try to build faster payment or state networks on top of the distributed systems with them as point of last resolve in case of conflicts of cooperation or availability. Doing the inverse of this might also lead to interesting results.
Imagine we gave life to inanimate objects and those objects could speak to each other and us. As you described, our car would speak to a charging station. The charging station would advertise "I can charge you for a fee!" And the car could say "charge me here is money". Forget the "internet" of things, these objects only need to speak to other objects in their "domain". Imagine it like this: you walk down a street towards a sitting dog. Once you get close, the dog looks at you and observes you walking by. Once your sufficiently far away, the dog looks away and you're no longer in its domain.
Have you thought about the " language" for this inter object communication? Essentially the "api". Words like "charge" and "you" and "money" in my example.
Exactly, this is a component of what we're planning. We'd need agents constantly updating their interfaces, locations, and all sensor data simultaneously.
Also building some corpus of ontologies for these systems needs to exist -- and we're hoping to make it a community effort (a la freebase), as well as the recipes for actions between disparate systems or at least some reasonable way to address/connect the commonalities between interfaces.
Next will be developing artificial intuition, being able to connect interfaces in new and thoughtful ways to produce some interesting/undiscovered/unsupervised outcome.
> Google of course needs a vibrant web for its search engine to remain useful
So why did they build Android on top of Java then? Wouldn't it have been much smarter to build Android on top of Javascript+HTML where suitable, possibly extending the standards where needed?
All these "apps" nowadays represent a move away from the web.
I don't think cpu were ready back then. Sure you had that blackberry thing using the equivalent of the modern embedded webview but you couldn't do much heavy lifting with it.
Two main issues back then were scrolling perrformance and memory constraints.
We're working at image heavy microsites at lookcast and I can confirm memory issues and composition performances are still constraining what we can do on devices.
That would have made the performance far worse in particular on low end devices. And there were no sufficient, complete and useful HTML5 standards that could allows web apps to compete with local software.
Extending the web standards would simply have been too much work and caused too much trouble.
Arguably not. But even with this use case, the Web has got significantly more complex. Local Storage, local databases, GPS access, gesture detection to name a few. That all hits Web performance hard especially on earlier mobile devices.
A company worked on Android for several years, then Google bought it.
Google has ChromeOS, with a HTML based shell. And there were some news articles that Google is merging ChromeOS and Android - we will see if those speculations were true.
In addition to this. J2ME and Symbian (the two biggest mobile platforms when Android was being developed) were both based off Java so it made sense to use it.
Exactly, the very first iPhone didn't even have installable third party apps. 'Why didn't they sue webaps' ignores the fact that it was tried right from the word go and in a fight with native apps that was skewed in web apps favour, they still lost massively.
Chrome OS is of course an attempt to build an OS entirely on HTML and Javascript, but you don't see it on phones because the memory, CPU and power requirements are too high. Eventually these problems will likely be solved, FirefoxOS was another attempt, but if we can't even do it right now there's no way it was a viable option back in 2005.
I find this sort of stuff creepy since it's always built with the concept of giving you the idea to buy something even if you didn't really want it. It's nothing like Futurama brain slug but it's still just as disturbing. It's probably why I'm always unsubscribing from email advertisements from various companies I tend to buy things from (since I'm not a regular customer for anything but utilities, rent, and groceries). I want my choices of consumption to be driven more by necessity than just some slick ad or product placement. Hell, I didn't own a smart phone until just last year. How's that for weird?
> The hope is that this will not only create compelling user experiences, but also unlock access to the tens of billions of ad dollars that are currently spent on TV.
This is simply not gonna happen. I used to think that as well -- until I did a stint working with one of the biggest sellers of TV ad inventory.
It's a completely different game. TV ads are bought and sold in enormous, packaged deals. There's very, very little transparency in the TV advertising game. Everyone on both sides knows Nielsen ratings are total crap with very little bearing on reality, but they trust each other even less, so Nielsen it is. All anyone knows is that TV is great for brand advertising, and you can get a rough estimate of how many eyeballs saw your ad on a given night. With brand advertising, all you care about is the number of brand impressions you make on your target consumer in a given month to keep the brand value fresh in their mind (i.e. "Ford trucks are Built Ford Tough").
TV ads are sold based on personal relationships between reps at advertisers and publishers. That's it. The ad rates you can get for ads sold through traditional TV sales channels are 30-40% higher than digital ads on a per-impression basis. The reason for that? The buyers are more senior, have larger budgets, and less accountability.
That TV ad money is literally just disappearing as audiences scatter to various media platforms -- these older, more senior ad executives don't understand digital so it's left to the younger guys who have a much larger culture of accountability and measuring by the metrics. The Internet has created so many opportunities for advertising and so much transparency around pricing for those ads that nobody can justify TV-like rates for digital ad impressions.
Think of the things that advertise on TV: they're nearly all high-value goods or services that need to establish a brand reputation. They can (and do) do that online for a lot less per impression, and they don't need to run an equivalent dollar amount of impressions because the online publishing space has created so much inventory they can't even sell it all (creating so-called "remnant" inventory filled programmatically by algorithms at rock-bottom CPMs). But only the "premium" publishers (i.e. anyone with a non-advertising revenue stream like New York Times) can afford to limit their inventory in such a way as to demand higher CPMs: everyone else is in a race to the bottom just to survive.
But it's gotten to a point where ad rates are so low that for premium publishers able to succeed with a subscription model, they may not even be a material amount of money. See: Netflix, Amazon Prime Video, etc. In 5 or 10 years, we may get to a point where all "premium" video services are delivered through a subscription model without advertising. That TV money has to go elsewhere, but it's likely to be spent on non-digital forms of advertising (e.g. outdoor/billboards, corporate sponsorships, product placement, etc.)
tldr: Online will never see the type of ad money that TV generated because the balance of power has shifted from the publishers to the advertisers. The next generation of ad buyers will be much more savvy, and the digital publishers have created far too much ad inventory. There's a tragedy of the commons situation with publishers that will continue to drive down CPMs while inventory increases, because if they don't then they will die. Because of low ad rates, it's not really worth it for premium video publishers to compromise their user experience by subjecting their paying customers to advertising -- thus driving the ad money from the big brands out of premium video content and on to other, non-digital forms of brand advertising.
Usage demographics are moving off of TV, so advertisers have to follow. It's not like how it sells will be different, it's that retailers are going to go to TV sales people less.
There will be fewer TV advertisers, and the money that in the past would have been spent on TV ads will be spent on digital ads.
I don't think it will be a 1:1, I think it will be a 1:.5 for a while, putting a lot of those people you talk about out of business completely - just like travel agents.
I agree that TV advertising will go away, but keep in mind when I say "advertiser" I'm speaking of companies purchasing advertising slots (not the ones selling them - they're typically referred to as publishers). The ones selling TV ads will definitely go away. Digital ads have a totally different sales model and level of "prestige" that is unattractive to the type of advertisers who make up the bulk of the national TV advertising market where the majority of the billions are spent. Most TV advertising falls under the category of brand marketing, where you just want to get your company's name out in front of consumers X times per week so that they remain familiar with your value proposition. If you see 10 ads for "$5 Footlongs" in a week, Subway will be at the top of your mind when you are hungry and want to eat for like $5. Digital impressions are much easier to ignore, so you risk overexposing your brand in order to get the desired effect across all consumers in the market.
Basically, if you have a $100 million advertising budget, you're going to try to spend that in as few places as possible while getting the most bang for your buck. Digital is already a component of total ad spend for these guys, and they're probably not going to increase digital ad spend significantly because most of them are already at the point of diminishing returns on their digital spend. There's just too much inventory and not enough advertisers, so the pricing pressure in digital leads to a race to the bottom for publishers who don't have other revenue streams.
My hypothesis is that this money will end up in two places: outdoor advertising (billboards, bus stops, etc.) and on actual content production. Why make an ad that nobody wants to watch when you can hire RoosterTeeth or PewDiePie to make entertaining content that prominently features your products? The problem for the advertising platforms is that this money bypasses them completely and ends up in the hands of the production houses.
Not to say that digital advertising won't continue to be a component of total marketing spend -- it will be, and in a big way. But not to the same level that TV is today, because the power in the relationship has switched. Digital ads aren't a limited commodity like TV ads are - you can only show so many 30 second spots during the Super Bowl. That means advertisers can apply significant downward pressure on pricing, which is much harder for them to do when there is a limited supply of ad inventory that has to be sold in large chunks.
All of this leads to a situation where the publishers have less of a monopoly, and thus less ability to dictate pricing. Which means prices will go down until margins hit commodity levels (3-5% above cost).
I think your analysis is spot on...if you are looking backward.
Here are the points I would argue:
Digital ads have a totally different sales model and level of "prestige" that is unattractive to the type of advertisers who make up the bulk of the national TV advertising market
True currently. You assume however that TV is going to be the digital entertainment interface of choice for the foreseeable future. In fact it's overwhelmingly mobile already and in the next 10 years (as an AR developer) we see that moving toward AR displays with complete replacement for a significant portion of the population by 2030.
Basically, if you have a $100 million advertising budget, you're going to try to spend that in as few places as possible while getting the most bang for your buck
I suspect these TV budgets will drop, as new interfaces reduce cost to on-board, and then rise again over the longer term. So whereas you would line item 50M for TV, maybe that drops over the next 15 years to 5M as TV eyeballs go away and competition for share grows.
Digital ads aren't a limited commodity like TV ads are - you can only show so many 30 second spots during the Super Bowl.
Here is where I have the most disagreement. When you reduce the channels in which people interact, there is extremely limited quantity. The overwhelming bulk of digital advertising dollars go to Google and Facebook. We have been talking for years about who controls "virtual ad space" [1].
In the end the argument to me comes down to the actual entertainment interface. People are moving away from TVs already and trends are pointing to TVs going away all together in the not-too distant future. My company is working hard to accelerate that process and we see a lot of competition forming over the last few years that is doing the same.
> True currently. You assume however that TV is going to be the digital entertainment interface of choice for the foreseeable future. In fact it's overwhelmingly mobile already and in the next 10 years (as an AR developer) we see that moving toward AR displays with complete replacement for a significant portion of the population by 2030.
I actually don't disagree with this at all (AR is absolutely the future), but I think that people will be incredibly averse to advertising in AR interfaces outside of an entertainment context. As seen with mobile, the device manufacturers will be the gatekeepers -- and the gatekeepers who put consumers first will win. Apple didn't win mobile because they had a better phone; they won mobile because they defined the product space in a way that gave them the type of margins that enabled them to put the user experience first and foremost. They didn't need a second revenue stream from advertising or apps to succeed. It wasn't an accident that the first iPhone didn't have either of these capabilities -- apps were an obvious idea (and had existed on PalmOS and Windows Mobile), but Apple wanted to get the UI out there and make the user experience great before introducing 3rd parties whose interests would cloud their decision-making. And the original iPhone was a legitimate leap forward in mobile UI at the time as a result.
I actually don't think that TV is going to be the digital entertainment device of choice even in the near future -- my entire premise is that the ad dollars that are disappearing from TV are simply drying up along with the medium and won't necessarily be spent on advertising in the future. Mobile is a big part of that, but mobile video is a completely different market than TV video and the advertising model is very different as well.
> Here is where I have the most disagreement. When you reduce the channels in which people interact, there is extremely limited quantity. The overwhelming bulk of digital advertising dollars go to Google and Facebook. We have been talking for years about who controls "virtual ad space" [1].
I'm actually speaking about it on a "per impression" basis. Impressions are the bare minimum measurement you can have in any form of advertising -- it's a measurement that has some sort of correlation to the number of times your message was seen by a consumer. Impressions are qualitative as well -- a 30 second ad spot is a much better quality of impression than a banner ad or even an interstitial.
There's also a difference in a "lean in" experience (typical digital experience where there is some interactivity or user action expected) versus a "lean back" experience (turn on a TV show, sit down and watch it). Users in "lean back" experiences are much more accepting of "high quality" advertising, and it becomes easier to convey a nuanced brand message through imagery, sound and voiceover. That message can then be reinforced and shaped over time through more passive advertising (display ads both online and billboard-style). Users engaged in "lean in" activities typically ignore advertising because they're looking for the next opportunity for interaction (or become impatient if they're told they have to wait until an ad is finished playing).
Overall, my hypothesis is that unless publisher inventory can shrink dramatically on a per-impression basis, advertising is not going to be a viable revenue stream for anyone but Google and Facebook. The companies that win will be the ones that can do it without advertising, and do so by putting their users' needs ahead of any others. If you think of it that way, sites like Google and Facebook simply become the filter that the user trains to their preferences, and when they are interested in a product / category, that algorithm is trained to know the user's interests and present relevant options. Users are outsourcing the decision-making to algorithms because there is too much choice -- and it's going to fundamentally transform advertising (if not eliminate it entirely).
Watch "Detroit Muscle", "Truck Tech", etc. on TV. Those are shows that show you how to hack your car. The sponsors of the shows provide the tools and parts that the mechanics use. It is advertising, but in a way that people want to see. It's a very organic form of product placement.
The old TNN network[1], (before it pivoted to Spike) programming was probably a pioneer of sponsored content - besides that kind of gearhead fare, it was chockablock full of hunting and fishing programming that was basically just advertising for Remington, Browning, Bass Pro Shops, etc.
yep, entirely an ad. I was once their customer (2006ish), and paid about $1000 per minute. Was not worth it. Would not do again the way that it was done. Counterintuitively, I think I would have been better off to spend more and dominate the focus on the show.
In this light I think just producing my own content for Youtube is the smarter strategy. I have to acquire new skills, but who doesn't when they run a business?
> In this light I think just producing my own content for Youtube is the smarter strategy. I have to acquire new skills, but who doesn't when they run a business?
This is the route it's going. Larger brands will hire professionals and celebrity hosts, but smaller guys can get in on the game as well (i.e. the "Will it Blend" videos).
Sure, you'll probably have to do some ad spend to get some eyeballs to your videos, but it won't cost anywhere near what TV advertising does.
TV remnant inventory is usually sold at the local level and is nowhere near as large a percentage of of the pie as it is online. It's also hard to "create" more inventory on TV - not so online; which is exactly what publishers do when their CPMs start to drop. They'll add more slots for banner ads, interstitials, split articles into multiple pages... This increases total revenue for a while, but ultimately crushes your CPMs.
There's a floor on TV advertising of about 16-20 minutes of advertising per hour. More than that and you start to turn your users off in huge numbers.
As we have more and more of these systems tied together - usually via APIs - I think it's getting more more likely where things can "just work" transparently.
If I need gas for the car, it can go get it, billing to my account on file. If I need diapers for the kid, the system detects that, orders them automatically, delivers them automatically, etc. There are few humans involved anywhere.
It's a fascinating and has society-shaking implications.