Mid-band 5G is another story. With better MIMO and such, the efficiency improvements seem to be more like 50-80%. Not only that, a lot of new spectrum is being added to networks and it's TDD (time-division duplexing) rather than FDD (frequency-division duplexing). T-Mobile and Verizon have launched their 2.5GHz and 3.5GHz networks with 60MHz of spectrum and people are usually seeing speeds around 300-400Mbps which is significantly faster than LTE (and significantly faster than the average home broadband connection). A lot more mid-band spectrum will be coming over the next couple years.
Likewise, initial rollouts of 5G were in non-standalone mode. That means that LTE is running the show and the 5G is providing its bandwidth. That also means that ping times are driven by the ping times of the LTE network. 5G standalone networks drive down ping times a lot. Network and handset support is a bit iffy right now, but it'll come along.
I think it's also easy to forget how LTE was a minimal upgrade over HSPA+ when it initially came out. Why not just use dual-carrier HSPA+? Even in mid 2013 (2.5 years after Verizon's LTE launch), they were averaging 10.25Mbps while T-Mobile's HSPA was 7.66Mbps (via PC Magazine). Only a third faster? Why bother, right? Of course, that hides the fact that LTE drove ping times down a lot from 3G networks which meant a much better experience for users. It also hides the fact that LTE had a lot of room to improve over the years and now we see LTE speeds 4x higher than 2013. When you're comparing a newly launched product to the last generation that you've spent years working on, the new product can often seem like a minimal upgrade (even if it has a lot more runway).
T-Mobile's mid-band network is the most deployed in the US and their average speed has risen to 91Mbps (via Ookla/Spedtest.net, as of 3-6 months ago) - not their average 5G speed, their average speed. That's with less than half their network usage going over 5G and even within that 5G usage, a lot of it is probably over the low-band 5G network (which is more like LTE++). Their average 5G speed is 187Mbps and that's averaging in both the low-band LTE network and mid-band LTE network (again via Ookla, as of 3-6 months ago). I think that going from 40Mbps to 190Mbps is a pretty big difference.
Not only that, it's a better real world experience. Ookla found that 13-20% of the time (depending on carrier), LTE was below 5Mbps down and 1Mbps up for people. T-Mobile's 5G speeds were at least 25Mbps down and 5Mbps up 82% of the time. So with LTE, you're going to get at least 5Mbps 80% of the time and with 5G, you're going to get at least 25Mbps 80% of the time. Plus, 5Mbps up means a nice video call compared to 1Mbps. 5G is offering better stuff in the real-world and this will get better as the networks mature - carriers are still upgrading towers.
Now, if you're on AT&T, they have almost no mid-band 5G. If you're on Verizon, they've just launched a decent amount of mid-band 5G (30% of Americans covered, although people are reporting that their maps are hugely over-estimated), but it's limited compared to T-Mobile and only available on their premium plans. It might not seem like a big deal if you aren't seeing mid-band 5G.
I think 5G is also starting to put pressure on home internet providers. T-Mobile gained more home broadband customers in Q4 2021 than anyone (wired or wireless). A lot of rural customers are getting real broadband for the first time and upgrading from things like HugesNet with restrictive data allotments. Starlink "isn't a big deal" if you already have good broadband. If you live in an area without it, Starlink gives you a modern internet connection. Likewise, wireless home internet will offer new options to many rural customers as well as put competitive pressure on broadband monopolies.
I think part of Tim Bray's skepticism is: do we need more speed on mobile? In some ways, I share that skepticism. However, I'd also add: do you need gigabit fiber? He notes that he has fiber at home. We often don't know what we'll use speed for until we have it. Mobile data was around for a long time before the iPhone and while some people had found some uses for it, most of us didn't see utility. Then Apple releases the iPhone and all of a sudden the world starts reorienting around data. It can take time for someone to make a killer-app for a capability.
T-Mobile just launched a 5G innovation center to help people figure out what to do with 5G. Sometimes technological capabilities precede someone figuring out what people will want to do with it. I mean, we had broadband connections capable of 1080p video for years before Netflix started streaming. Today, streaming is a way of life.
Bray also says that "I don’t personally know anyone whose life has been changed by 5G. Yes, family members with newer phones occasionally report that the status bar says “5G”, but I don’t hear that they’re having a different experience." A big part of that is that carriers rolled out 5G before most 5G capabilities were really there. If you're on a low-band 5G non-standalone network, it can be hard for it to feel any different. Heck, my device often performs better with 5G off if there's network congestion (since so many people are using 5G in my area and my carrier isn't really good). But that's because every carrier wanted 5G to pop up on your phone even before 5G's capabilities were really in your hands. Carriers pushed things out a bit quickly in part because they wanted phones to start supporting 5G (for when the real capabilities of 5G got deployed) and because it offered useful marketing.
Bray also says that 5G started in Spring 2019. That's not wrong, but it can be misleading. Verizon launched a small amount of millimeter-wave 5G in Spring 2019. Millimeter-wave spectrum typically has a range of 100-200ft and is blocked by almost anything in its path. T-Mobile launched low-band 5G in December 2019, but again low-band 5G in non-standalone deployment isn't really that different. T-Mobile really started launching mid-band 5G in Summer 2020 with Verizon following 1.5 years later in January 2022. However, it takes a while to upgrade a network. If you have 60,000-80,000 towers that you're upgrading and you're upgrading 1,000-1,500 per month, it's going to take a while. For T-Mobile, we know that it's going to be around 3.5-3.7 years to get to 90% of Americans covered. Verizon started deploying mid-band ahead of their launch (since they knew the spectrum they'd have before they were allowed to use it), but they're staying it'll still be 3 years to get to 75% of Americans covered (and given that they started deploying 6-9 months before their launch, we can say 3.5-3.7 years to get to 75% of Americans covered). Basically, we're starting to see some great capabilities some of the time, but we probably have another year or two before it becomes really common and gives people more "wow" moments.
So I think if one is skeptical of 5G, one needs to talk to people that have been using "real" 5G (rather than 5G pushed out for marketing reasons) and one might need to wait a few years as networks get deployed. However, there are already so many people that 5G is making a difference for and that number will grow over time.