Newsletter on LinkedIn

During 2023, I've lost patience with some of the more outrageous statements I've been seeing in my industry. There's a growing number of bald-faced lies, and I've started calling them out publicly.

Yes, I know that marketing and lobbying requires a measure of hype, exaggeration and glass-half-full predictions and estimations. There are areas where statistics or semantics can be skewed, but may contain a grain of fact.

Some problems lie with simplifications or lack of understanding made by non-experts. While politicians are generalists and can sometimes be excused, the wide use by the industry of economists to calculate supposed costs or benefits of networks is often deeply flawed.

Worse, I see many examples where purveyors of dodgy stats and talking-points cite (and amplify) each other's nonsense. You know the sort of thing - a laughable analysis of "GDP uplift" from a technology gets blended with questionable forecast traffic volumes, then multiplied by other flawed numbers to imply huge extra CAPEX or spectrum needs.

Nobody stops to ask users, application developers or enterprises how networks in the real world actually work or get deployed, or whether all the clever AI, automation and virtualisation we're also hyping might reduce the costs.

Few people really look at headline numbers or arguments to see what is cherrypicked, misrepresented, show correlation not causation, or just don't pass a simple "smell test" for being realistic.

Anyway... here's a dozen of the most egregious falsehoods in telecoms. There's plenty more, but this seemed an appropriately festive number to list. I'll add some links in the comments as well. Here's to a more honest and truthful 2024!

Lie 1: Autonomous vehicles & robotic surgery need 5G

Let's start with an easy one that long been debunked, although I still hear it repeated by some in telecoms or policymaking circles, as well as some self-appointed futurists who should know better.

No, AVs don't need 5G, although they may use it for certain functions if it's available and inexpensive. All the claims of TB of data created per hour are irrelevant - 99.9% never leaves the vehicle. Most of the processing & AI inferencing is done onboard. And cloud-driven 5G AVs would need perfect coverage and capacity, which doesn't exist - good luck in a tunnel, a car-park or a road in a remote region. Or on a highway in a lane sandwiched between two trucks.

5G is somewhat more important for remote-driven vehicles where a human needs streaming video of the surroundings.

As for robotic surgery - well for a start, most hospital operating theatres are Faraday cages deep inside buildings, so they'd need dedicated wireless systems. And the remote surgeon - and the robot - are likeliest to use fixed broadband and a fibre LAN connection, not wireless.

You remember the video of the banana undergoing "5G surgery"? That was a proven lie, with a full analysis by The Verge. It didn't use 5G.

Lie 2: There is an "investment gap" for broadband networks

2023 has been a year of preposterous lies from lobbyists, some telcos, industry groups and even government agencies, trying to concoct arguments for new regulations on cloud or content companies funding networks - or perhaps more government subsidies. A common refrain is that there is a "gap" in funding that needs to be filled by someone else.

The problems are that the majority of investments required for fixed and mobile broadband deployments are either:

• Covered by market forces and normal commercial investment plans, helped by existing government funding programmes.

• Exacerbated by arbitrary or poorly-defined "requirements"

• Not driven by growth in traffic volumes

• Based on old or inaccurate metrics and statistics of current network status & investment schemes

My friends at organisations such as Stratix consulting have done a good job at debunking some Europe-specific claims. (Link in coments). Let's have some more informed & truthful debate in 2024.

Sidenote: some investments needed won't be done by the normal telcos anyway. Indoor wireless owners, local FWA, fibre altnets, private networks, satellite systems, neutral host & infraco / towerco CAPEX weren't even in the discussion - if there is a new pot of cash, it's another lie to say that legacy telcos are the only or best recipients.

Lie 3: 5G networks are (or will be) "ubiquitous"

One of the common mistaken assumptions among policymakers and some others in telecoms is that 5G networks - and their headline capabilities like gigabit speeds and millisecond latencies - will be ubiquitous.

Clearly, that's not the case, and was doubtful from the start of 5G. Rural areas, small communities, indoor coverage & capacity, full connectivity on railways & trains, industrial zones and others often lack public network coverage, even from 4G. Even where there's a 5G logo lit up on a phone, that may just be a thin layer of sub-1GHz spectrum or dynamic-shared 4G/5G.

To be fair, this is perhaps more an example of ignorance rather than lying in many cases, but it should have been clear to even the least tech-aware person by now that the concept is itself proven untrue.

It's also very consequential - application and device developers are the ones that have been lied to, as much as end-users. Imagine spending money on developing AR/VR game or headset, expecting low-latency 5G everywhere it will be used. This type of hype has victims.

It's also quite amusing to see the latest ITU IMT2030 (6G) recommendations specify that ubiquity only refers to a defined coverage area, not nationwide: "The term 'ubiquitous' is related to the considered target coverage area and is not intended to relate to an entire region or country". Well, that's one way to avoid lying in future - redefine the word itself to mean something else that's easier an more convenient.

Lie 4: Mobile data traffic is growing "exponentially"

Exponential is a mathematical term referring to an accelerating growth rate. In almost all cases, mobile data traffic growth is now slowing.

Most predictions suggest it settling down to maybe 20% year-on-year in mature markets. And even that is mostly driven by MNO-driven decisions like offering FWA fixed wireless services (which have 20x the consumption of normal smartphone use), or inappropriate pushing of unlimited plans or bundled / zero-rated video.

I still see telco policy people deliberately overestimating traffic forecasts, either to make arguments for more spectrum, or when lying about so-called "large traffic generators" or risible "fair share" schemes.

Lie 5: Network traffic is "generated" by content / cloud companies

I've previously written a full LinkedIn post calling out the term "large traffic generator" as a clear lie. It's been one of the most-read that I've published this year, which presumably implies I generated a lot of data traffic personally.

The reality is that Internet users generate traffic. They request movies, play games, scroll timelines, download software updates, and read articles. The fact I watch a GB of video traffic from a popular site rather than a niche one is irrelevant - although the larger one is more likely to have its own CDN.

There's a small amount of "push" data such as autoplay follow-on videos, but adverts are usually tolerated as an alternative to fees. It's not a big deal - and in any case, network costs are linked to peak traffic levels, not total volumes.

Lie 6: "Voice" is the same as "Telephony"

There are 1000s of forms of voice communication. Phone calls are just one highly-specified application, with specific behavioural, technical and regulatory characteristics. Telephony is

Push-to-talk, in-game chat, karaoke, audioconferences, voice assistants, podcasts, audio captioning, Alexa "drop-ins" and many others are not telephony.

Generally, telcos don't do "voice" in general. They only do telephony, plus a few extra bits like voicemail. A few have their own voice home assistants.

(Historical sidenote: audio streaming using phone networks has been around since the 1880s. If you're a telco complaining about "OTT" media content, you're 150 years late).

Lie 7: All Wi-Fi use on smartphones is "offload"

The term "offload" really involves Wi-Fi traffic that would otherwise have gone over cellular networks, but an automated systems pushes to Wi-Fi instead. It excludes data that the user (or the OS) deliberately selects Wi-Fi for.

If I watch a YouTube video on my sofa on my phone via Wi-Fi, that's not offload. That's just me using my home broadband.

Maybe only 5% of smartphone Wi-Fi data is genuine offload. And even that could translate to a smaller amount over cellular, because both user and app-developer choices often mean extra "elasticity" - higher resolution or frame-rates, extra usage because it feels free/unmetered and so on.

Lie 8: Interoperability is always beneficial

Interoperability for infrastructure can be very useful - we all benefit from devices that work with wireless or fixed broadband networks, and are tested and certified to be compatible with each other.

But interoperability for applications and services is much more mixed. As long as users can "multi-home" and have several different calling, messaging, social or gaming platforms, it's not necessary to have interoperability.

Yes, there can be competition concerns, but that doesn't imply a need to regulate for a lowest-common denominator set of features, with a wide array of unintended consequences. Interop should be on-demand (especially if customers explicitly ask for it), not assumed to be an ideal scenario and mandated by regulation.

Yes, I'm thinking of the EU DMA's stupid rules on messaging interop, but also a more general argument about all communications apps & services, as well as fields like AR/VR/metaverse. There's still a role for proprietary network technologies and architectures in some cases too - openness is great, but it's wrong to say it should be universal.

Lie 9: QoS or performance can be guaranteed "End to End"

Almost every time you see the term "end to end", it's a lie, especially in reference to network quality or performance.

In reality, nobody controls all the components of a network or application path, outside very niche and specifically-engineered situations. At best, they can oversee and prioritise activity between two arbitrary points in the middle.

Think through a videoconference session. It involves at least two networks, plus interconnections. Multiple devices, with multiple components like processors and memory. A cloud platform and maybe a CDN and 3rd-party elements. Maybe a last-metre connection via Wi-Fi or Bluetooth or a USB cable. Maybe one user is deep in-building or in a moving vehicle.

End-to-end QoS can be created if there's one network, IT or OT manager in charge of a local system. It can be offered on one fibre connection and a few boxes, or perhaps across one boundary. But truly end-to-end? Nope.

Lie 10: "Digital" tech is new, cool & all that matters

Yes, I have a bee in my bonnet about the word "digital". I cringe at phrases like "digital transformation", or "digital infrastructure". It's like going back to the 1980s and hearing "information superhighway" or "multimedia".

We've had digital communications since Morse and the telegraph in the 1840s. Digital computers since the 1940s. Digital phone switches since the 1980s. The WWW since the 1990s.

Digital is old.

Don't get me wrong, it's still useful. (Obviously). But so is fire and the wheel.

What irks me is that much of the coolest stuff in tech today isn't digital - it's quantum, or neuromorphic, or biological, materials-based, analogue and many other domains of innovation that don't depend on 1's and 0's.

The new & cool stuff is mostly Post-Digital.

Lie 11: Mobile enables huge CO2 savings

There are many claims that use of mobile (or 5G) leads to huge savings in carbon footprint - fewer flights because of videoconferences, wireless traffic controls reducing congestion, connected solar panels and batteries, and so on.

Most of this is a massive exercise in double, or 10x-counting. If I do a videoconference with a client, there's at least 2 access networks, interconnects / long-haul fibre, two end-devices, two screens, two cameras, lots of chips, a cloud platform, a video app or browsers, mics etc.

Being generous, one of the access networks (probably fixed+WiFi rather than mobile anyway) maybe accounts for 5% of the total savings. And only a handful of trips are replaced with video, in any case - most video sessions are incremental, or replace a lower-energy voice-only call.

Lie 12: "The data and KPIs prove XYZ"

One of my big campaigns this year has been about the need for good metrics, not easy metrics. I've written several posts about it, and an entire report for my friends at STL Partners.

In telecoms we get huge volumes of reported data - average network speeds, 5G coverage, spectrum prices in $/MHz-pop, homes passed, minutes of use, GB of data traffic, numbers of messages sent, and myriads more.

The problem is that many of these data points are used because they are easy to collect, easy to regulate & have a lot of historical comparisons. But they're often useless, or sometimes worse-than-useless.

They get cherrypicked, used out of context, used to "prove" the need for policies that they don't actually support, and don't represent the reality on the ground.

Take mobile coverage, usually cited as a % of residential population that can get a signal outdoors. Yet most will agree that many 5G uses are for non-residential applications and locations (eg farms, ports, industrial zones), or indoors. And coverage doesn't mean performance or "full 5G" capabilities.

Then there are all the many tiers of lies about network data traffic, used to "prove" the costs of network CAPEX, energy use, need for taxing cloud/content companies, arguing against net neutrality and so on. Aggregate traffic is largely irrelevant - in reality costs are driven by initial build / coverage area (even at zero traffic) and sometimes peak rates. Energy usage isn't linked to traffic mostly, either.

We all know the phrase "lies, damned lies, and statistics". So let's try to collect and use better stats in future.

Conclusion

Yes, this is a bit of a rant. And yes, a lot of people will object to me calling them liars. But let's be clear, most of those I'm referring to have jobs to do, that sometimes need "messaging" to be massaged.

And I haven't even got onto the dodgy term "OTT", mentioned RCS (no it's not SMS2.0), laughed at cryptocurrency-powered "DePIN", called out exaggerations about the role of satellite direct-to-device, or ridiculed the criticism thrown at shared spectrum, CBRS or private networks.

So if you recognise yourself in this, tell the real telco Pinocchios - the lawyers, lobbyists, marketeers and headline-writers - that you're not willing to peddle untruths or half-truths in future. Don't let them put your nose out of joint, or make it grow longer.