Sometimes the Value of Extended Reality is the Device, not the "Reality"

So I did this, recently. The value actually had nothing to do with extended reality--artificial or virtual--but in the use of the goggles instead of a dedicated stationary appliance. In other words, the appliance moves to the human, not the human to the machine. 

This sort of thing happens quite often with new technology. A device or app is developed to solve one set of problems or provide one set of values. But then it gets used in ways not in line with that creator vision, in ways that provide value, but were not specifically imagined. 

Yes, the goggles were "better" in terms of user experience. But it was the goggles, not the software or app experience, that accounted for the increased value. 

Automation is a Good Thing, But Effectiveness Might be Better than Efficiency

Connectivity provider operations are complex, almost anyone would agree. Perhaps everyone would agree that simpler, less-complex operations are desirable. The issue is how much less complexity is optimal?

Zero touch is one possible goal: the ability to provision any service with as little friction as possible. But there are larger questions. 

One difference between wholesale and retail business models is that wholesale operations are less complex by design. Wholesalers sell only to a relatively small number of other business customers who want to use those assets to sell products directly to customers. 

The reason mobile virtual network operators exist is that their business models allow for lower-cost operations. By avoiding capital investment in networks, they can focus their spending on marketing and sales. 

Automation in various forms helps both wholesalers and retailers, of course. But if the purpose of automation is to lower operating costs, there are potential business levers to pull beyond automation as such. 

Among recent trends beyond wholesale are co-investment in facilities, joint ventures or spinning off facilities, often to private equity firms.  

Telenor, for example, has just divested 30 percent ownership of a new infrastructure company, Telenor Fiber AS.

A consortium of investors led by KKR includes Oslo Pensjonsforsikring as a co-investor.A bit unusually, the new fiber company’s sole customer will be Telenor. Most other wholesale entities want to be able to sell access services and capacity to all comers. 

The motivations for such asset sales often include raising cash for other purposes such as reducing debt or reducing the expected cost of new infrastructure. 

All such deals, however implicitly, do raise questions about the value of access infrastructure assets. Such assets are an essential feature of modern computing and application usage. Less clear is the strategic value of digital infrastructure ownership. Mobile towers provide an essential function, but no longer are universally viewed as “must own” contributors to business value. 

Many mobile operators have concluded they can acquire radio sites as an operating cost, rather than investing capital in owned networks and facilities. Some fixed network operators seem to be concluding that full ownership of access networks is not necessary. 

In other markets, such as Malaysia, structural separation (retail mobile operators lease network services from a third party) might be acceptable, but with mobile operators also owning portions of that asset. 

More automation of operations--for both wholesale or retail operations--obviously is a good thing to the extent that it lowers operating costs and boosts customer satisfaction. But more automation is an example of the distinction between efficiency and effectiveness. “Efficiency” means doing what you do with less cost and overhead. “Effectiveness” means doing things that matter. 

Automation, ostensibly a matter of efficiency, also can raise questions of effectiveness. Where does greater value lie: automating present practices or automating different processes. In the past, we used to describe such choices as “digitizing your business practices without redesigning them.” The often preferable option was to change business practices using new technology. 

Choices to automate often should also include a deeper examination of what needs to be done, and why. Perhaps the greater value comes from reimagining what has to be done, and by whom.

Business Advantage is the Goal of Every Form of Virtualization

Business advantage is why we hear so much about various forms of connectivity network virtualization. And, increasingly, virtualization operations will be assisted by artificial intelligence, moving beyond today's applications for customer support operations.

Open networks are supposed to lead to lower costs. Virtualized networks are supposed to lead to greater agility and flexibility; the ability to create new services including self-managed or self-provisioned services. 

AT&T made a big commitment to “virtualization” of its network over the past decade, essentially completing that project in 2020. Lots of mobile operators are looking at ways to virtualize their radio access networks. Verizon has done the same. It is partly an evolution of software architecture and partly a matter of network architecture. 

source: Redhat 

Core network operations are shifting to “cloud native,” which essentially also is virtualized, in an effort to shift away from proprietary, closed software to open or extensible systems and more-modern use of hardware platforms and architectures, including use of hyperscale facilities instead of “owned hardware.”

In a broader sense, efforts to virtualize and go “cloud native” are the latest efforts to create more flexible, agile platforms that we can speak of as supporting “bandwidth on demand” or “feature on demand” capabilities. Automation might be another term some would use. 

Separation of control plane and data plane is a foundational principle for data networks these days. Essentially, the control plane is the architecture for making routing decisions (which path to take). In traditional networking, the control plane manages the flow of packets through the network while the data plane handles packet forwarding. 

With software-defined networking (SDN), the two planes are separated so that the control plane can manage the entire network, rather than just individual devices. That has implications such as allowing the separation of control logic and servers from forwarding logic and devices. -----

That allows lower-cost operations, as expensive controllers can be centralized while router nodes are dispersed and running on generic servers rather than dedicated appliances. 

source: Snapt 

The data plane architecture is the way actual “bearer traffic” or end user data is forwarded from one interface to the next. Such approaches are inherently decentralized, increasingly virtualized and based on cloud computing principles.  

source: Marc's Blog 

So it is not surprising that we hear so much about connectivity provider virtualization and cloud native architectures. Software-defined networking is another term sometimes used to describe modern data communications principles. 

The ability to use open radio access networks, separated base station controllers and radio sites and generic hardware (bare metal) all are related to the separation of control and data planes. 

All those also are based on the fundamental “layering” of all modern software. 

source: Guru99 

Is There a Moore's Law for Artificial Intelligence?

You might wonder whether it is possible to create a rate-of-improvement algorithm for artificial intelligence as we have algorithms for computing power and home broadband speeds, such as Moore’s Law and Nielsen’s Law. 

It is easier to illustrate the increase in computational power to support AI for the simple reason that we can use parallel processing hardware to support AI.  But AI progress depends on lots of other things.

For AI applications, training time matters. 

source: IEEE Spectrum; MLPerf 

To be sure, deep learning and neural networks might be said to increase algorithmic efficiency. That might bend the Moore’s Law curve higher. But artificial intelligence also depends on large datasets to train machines, so chip level performance is but one input. 

In other words, AI--especially machine learning--is driven by multiple inputs, not just chip performance. Moore’s Law, by itself, suggests increases in transistor density of roughly a doubling every year to 18 months.

In the case of machine learning, software and computer architecture, plus the ability to access bigger data sets, can produce a seven-fold 11-fold rate of improvement.

source: Discover magazine,; Jaime Sevilla, University of Aberdeen 

The point is that, at least for machine learning, progress has been faster than Moore’s Law.

AI Enables AR and VR, Which Enable Metaverse

Artificial intelligence is a crucial enabler of augmented and virtual reality, both of which are enablers of eventual metaverse experiences. AI, AR and VR all are examples of applying better technology to enhance the realism of media and content experiences, a process that is a century in the making.

Not all observers believe metaverse environments  will actually happen. Others might argue it just makes no sense. Even optimists warn that it will take a decade or more to develop.

But there is another way to look at movement towards full metaverse experiences that suggests it will arrive: realism. Defined as the experience that “you are there,” realism approaches “real life” experiences: three dimensional, interactive, using multiple senses. 

Think about the experience of participating in a sport, watching others play a sport live in a stadium, watching it on television or listening on radio, viewing a photo of the game or hearing somebody talk about a great play during that game, reading a story about that game or viewing an artist’s rendition of a key moment in a game. 

The point is that there are degrees of immersion and realism, and that the degree of realism has tended to improve in the eras of electronic media. 

source: Prezi 

“Real life” experiences are the ultimate in realism: you actually are there, actually doing something. All media are efforts to portray activities such as surfing, playing baseball or cooking in ways that simulate “being there and doing that.”

Storytelling is perhaps among the oldest forms of media or content. But it requires use of imagination on the part of the listener. Live theater performances were early developments in realism, as “you had to be there” watching real people in real roles. 

Electronic media extended participation in experiences, but with less realism. Radio was sound only, and monaural. Stereo arguably improved realism. “Surround sound” added more realism. 

Film was more immersive, but without sound. Movies were more realistic after sound was added. Film arguably became more realistic after the addition of color. 

Television began in black and white, then added color. Digital media improved realism by increased image definition (first DVDs, then Blu-ray). Broadcast television switched from lower-resolution NTSC to HDTV. Now we are moving to 4K and 8K. 

Think about 3D movies or TV as efforts to further extend immersion and therefore realism. 

Videogames initially were quite two dimensional. Today’s games are graphically much more realistic and immersive. 

In that framework, our movements toward metaverse are the next stage of improvements in realism and immersion. Persistent, three-dimensional environments where “you” and other people and objects “are there” is simply part of the on-going development of media realism. 

Of course, there are other angles. In manufacturing, realism might mean that whole processes can be modeled and replicated in real time. For videoconferencing, realism might be the sensation that all participants are in the same room. 

Metaverse will get traction as it proves to enhance realism in media experiences. And that is why, despite the hype and the long journey to commercialize it, metaverse ultimately will arrive. 

Humans have adopted all forms of media that provide higher realism.

How Big is Metaverse Market? Depends on What You Count

One issue with estimating potential metaverse revenues is that metaverse may be a feature of current experiences and products. So one can infer or attribute “metaverse” revenue to any product that might happen to use the capabilities. 

One might also differentiate between service revenues and metaverse hardware or appliance revenues. ABI Research appears to focus on the latter. 

Between 2021 and 2030, the number of active users in virtual spaces requiring headsets will climb from 118.3 million to 1,048.8 million, at a  compound annual growth rate (CAGR) of 27.4 percent, according to ABI Research.

Virtual spaces headset  revenue will grow  from US$4.2 billion in 2021 to US$44.8 billion in 2030, the firm estimates. 

ABI appears to model its results by estimating revenue in existing product categories that are obvious because they require the use of virtual reality headsets. Nearly $5 billion could be earned in 2030 by customers paying for headset-enabled experiences, for example. 

Roughly the same amount of revenue is anticipated to be generated by VR-enabled videogaming. Smaller amounts will be required to support advertising functions or cloud content. 

The biggest growth, though, should happen in the video collaboration area. ABI Research forecasts that total collaboration revenue (collaboration and related services revenue and dedicated video collaboration hardware revenue) will increase from US$24 billion in 2020 to US$64 billion in 2030. 

source: ABI Research 

That might be the only area where services revenue also is significant. That will include revenues generated by meetings where attendees are virtual. 

In that sense any estimate of “metaverse” revenue faces the same issue as forecasts for other products that are a combination of hardware, software and services  revenue: 5G private networks; local area networks; software-defined wide area networks, edge computing or internet of things.

Execs from Nvidia and Equinix on Scaling AI