YouTube broke for 40% of the UK population after rolling out 60 FPS videos
thehftguy.com
thehftguy.com
Many apps assume 100% internet connectivity at all times. i.e. not even minimally functional when offline.
They assume a certain speed, latency and packet loss (zero).
The adage should be: plan for the worst, then adjust for reality.
It seems that OP defines downgrading to 720p60 (?) as "broken".
I do support bringing back the 1080p30 stream though.
Edit: Tested in Chrome with the "Slow 3G" throttling profile. YouTube automatically falls back to 240p with some buffering.
TFA is claiming ~40% got a downgrade in resolution.
The 20% who had a connection 4-8 Mbps and could do 1080p got downgraded to 720p60. The 20% who had a 2-5 Mbps connection and could do 720p got downgraded to 480p.
Obviously it's not clear cut, there's a continuous spectrum where things will load much slower and start having breaks in playback until you really have to drop the resolution.
Well, we should also consider households having more than one person. Two people watching YouTube requires double the bandwidth, going into 12 Mbps+ territory. Truth be told, this is affecting more than 40% of the population.
It's clever to push lower quality, but here they push to a lower quality than what was achievable so far before the "upgrade". So from a user standpoint, YouTube have downgraded the service overnight for all low speed connections.
The more frustrating being for users that could sustain 1080p yesterday and are "upgraded"(?) to 720p60 today. Roughly same bandwidth but much lower resolution for what? better frame-rate? Not sure why anyone would prioritize frame-rate over resolution with all our fancy HiDPI screen.
When you watch a video with text or small content, like a gaming video for example, these have menus and texts, it stops being readable when downgraded from 1080p to 720p (or worse). It's broken.
When you live with flatmates or family, you're getting crazy load time and erratic pauses now because of the contention. There's not enough headroom to buffer ahead with the bandwidth use increasing 50% overnight. The video is dropping as soon as somebody else opens something, it's effectively broken.
The algorithm will see contention and downgrade the quality, only to upgrade it again a minute later when it sees a bit of headroom, only to fail again. There's a lesson here on network contention. Worse case is with 2 people browsing youtube, the 2 players are fighting each other for bandwidth.
Wait, what? Why would 1080p30 be gone?! I don't like 60fps, it looks weird, so now it's either 480p OR I'm forced to watch 60fps?! WTF?!
It only looks 'weird' because you're not used to it.
We're talking about frame rate.
(Does it even make sense to "upgrade" a 30/29.97 fps origin video to 60fps? I don't think so.)
Check this for example, you should only have the choice between 60 FPS options. https://www.youtube.com/watch?v=LXb3EKWsInQ
Videos that were submitted in 30 FPS remain as they were. It doesn't make sense to create new frames out of nothing indeed.
D:\_learning>youtube-dl.exe -F https://www.youtube.com/watch?v=LXb3EKWsInQ
[youtube] LXb3EKWsInQ: Downloading webpage
[info] Available formats for LXb3EKWsInQ:
format code extension resolution note
249 webm audio only tiny 57k , opus @ 50k (48000Hz), 1.89MiB
250 webm audio only tiny 75k , opus @ 70k (48000Hz), 2.53MiB
140 m4a audio only tiny 130k , m4a_dash container, mp4a.40.2@128k (44100Hz), 4.84MiB
251 webm audio only tiny 148k , opus @160k (48000Hz), 4.96MiB
394 mp4 256x144 144p HDR 88k , av01.0.00M.10.0.110.09.16.09.0, 30fps, video only, 2.34MiB
278 webm 256x144 144p 97k , webm container, vp9, 30fps, video only, 3.38MiB
160 mp4 256x144 144p 111k , avc1.4d400c, 30fps, video only, 2.87MiB
395 mp4 426x240 240p HDR 191k , av01.0.00M.10.0.110.09.16.09.0, 30fps, video only, 4.33MiB
242 webm 426x240 240p 233k , vp9, 30fps, video only, 6.89MiB
330 webm 256x144 144p60 HDR 243k , vp9.2, 60fps, video only, 7.96MiB
133 mp4 426x240 240p 245k , avc1.4d4015, 30fps, video only, 5.86MiB
396 mp4 640x360 360p HDR 403k , av01.0.01M.10.0.110.09.16.09.0, 30fps, video only, 8.67MiB
243 webm 640x360 360p 435k , vp9, 30fps, video only, 12.85MiB
331 webm 426x240 240p60 HDR 504k , vp9.2, 60fps, video only, 16.65MiB
134 mp4 640x360 360p 633k , avc1.4d401e, 30fps, video only, 15.49MiB
397 mp4 854x480 480p HDR 735k , av01.0.04M.10.0.110.09.16.09.0, 30fps, video only, 15.80MiB
244 webm 854x480 480p 857k , vp9, 30fps, video only, 23.18MiB
332 webm 640x360 360p60 HDR 1059k , vp9.2, 60fps, video only, 35.72MiB
135 mp4 854x480 480p 1496k , avc1.4d401f, 30fps, video only, 30.03MiB
398 mp4 1280x720 720p60 HDR 1552k , av01.0.08M.10.0.110.09.16.09.0, 60fps, video only, 35.98MiB
247 webm 1280x720 720p 1697k , vp9, 30fps, video only, 46.97MiB
333 webm 854x480 480p60 HDR 1988k , vp9.2, 60fps, video only, 69.11MiB
399 mp4 1920x1080 1080p60 HDR 2683k , av01.0.09M.10.0.110.09.16.09.0, 60fps, video only, 65.43MiB
302 webm 1280x720 720p60 2742k , vp9, 60fps, video only, 76.27MiB
136 mp4 1280x720 720p 3220k , avc1.4d401f, 30fps, video only, 57.62MiB
248 webm 1920x1080 1080p 3249k , vp9, 30fps, video only, 84.22MiB
298 mp4 1280x720 720p60 3483k , avc1.4d4020, 60fps, video only, 92.61MiB
137 mp4 1920x1080 1080p 4337k , avc1.640028, 30fps, video only, 103.11MiB
334 webm 1280x720 720p60 HDR 4524k , vp9.2, 60fps, video only, 162.28MiB
303 webm 1920x1080 1080p60 5018k , vp9, 60fps, video only, 134.10MiB
299 mp4 1920x1080 1080p60 5794k , avc1.64002a, 60fps, video only, 163.70MiB
335 webm 1920x1080 1080p60 HDR 6920k , vp9.2, 60fps, video only, 252.07MiB
400 mp4 2560x1440 1440p60 HDR 8626k , av01.0.12M.10.0.110.09.16.09.0, 60fps, video only, 223.85MiB
271 webm 2560x1440 1440p 9621k , vp9, 30fps, video only, 249.96MiB
308 webm 2560x1440 1440p60 13357k , vp9, 60fps, video only, 388.12MiB
336 webm 2560x1440 1440p60 HDR 16917k , vp9.2, 60fps, video only, 607.18MiB
401 mp4 3840x2160 2160p60 HDR 18167k , av01.0.13M.10.0.110.09.16.09.0, 60fps, video only, 460.07MiB
313 webm 3840x2160 2160p 18852k , vp9, 30fps, video only, 628.46MiB
315 webm 3840x2160 2160p60 26568k , vp9, 60fps, video only, 956.39MiB
337 webm 3840x2160 2160p60 HDR 30646k , vp9.2, 60fps, video only, 1.05GiB
18 mp4 640x360 360p 655k , avc1.42001E, 30fps, mp4a.40.2@ 96k (44100Hz), 24.53MiB
22 mp4 1280x720 720p 1669k , avc1.64001F, 30fps, mp4a.40.2@192k (44100Hz) (best)But yeah if you're assuming every user has the same bandwidth and latency a Google office (in SV) or employee has you're going to have a bad time
Depending on the app it's easy to trigger JS errors with latency problems and some things happening before others
If you look at this 2019 report, they first say that only 1% has less than 10 Mbits on page 20, then 13% have less than 10 Mbits on page 22.
Problem, metrics are like are so out of touch with reality, they're just made up. People actually use ADSL (copper) and it can't do more than 10 Mbps if you're a mile away from the internet center, that's the case for way more than 13% of the population. You could think that people have switched to fiber but they have not, I live in central London and it's hit and miss if you can get fiber at all in an apartment block.
This might be anecdotal but when we switched to encoding videos in 60 FPS on a channel I was involved in ~4 years ago, some users complained that their Android TV sets (afair a Sony model was mentioned) weren't able to play our new videos stutter free anymore.
30 FPS should stay the default option. At least YouTube should make the FPS setting manually configurable independently from resolution.
Seems to happen most frequently with slideshow video content, not sure why.
It also has a pseudo-market for broadband services through LLU. The network service and billing are done through various providers which compete on price. The ADSL line is monopolized by BT Openreach, who combine the efficiency of a state monopoly with the pricing of a private monopoly.
The government periodically gives BT money to improve speeds. BT pay this money to shareholders and then fail to improve speeds. Both parties are apparently happy with this arrangement.
One of the candidates at the last election promised free high-speed broadband to the country. Media commentators agreed that this was ludicrous pie-in-the-sky.
It has proven much easier to roll out fibre into villages and smaller towns where DSL is no where near as obvious as the solution and there is room to place the equipment easily.
What irks me more than anything is the fact that builders are making big blocks of flats and running normal copper cabling into them. Ones that opened just 5 years ago genuinely had ADSL back to the local telecoms office and only recently got upgraded to DSLAMS in their basements to fibre. It really doesn't save a lot of money to put copper in instead of fibre at this point and to not have planned at the very least VDSL is just quite ridiculous. These types of builds have just made local issues with broadband that much worse.
You'll find the same problems elsewhere in Europe dealing with population density and rolling out the infrastructure for this stuff. I know plenty of people in Sweden, France, Germany who struggle with exactly the same issues, despite better headline average speeds etc.
> "hollowing out"; services that get used by the small elite fraction of the country that lives in the right bits of London are good, everything else is quietly ignored.
Fundamentally wrong given that central London is in a large proportion (often > 50% by ward) social housing. Or are the millions of people who live there the "small elite" too? Or perhaps it is the case the high population density urban centres are naturally the easiest place to build infrastructure to benefit the most people.
> BT pay this money to shareholders and then fail to improve speeds. Both parties are apparently happy with this arrangement.
Clearly this isn't true. BT OR has for many years been separated and highly and transparently regulated. Perhaps you would like to explain more your implications of corruption that isn't visible to anyone else?
https://www.ofcom.org.uk/__data/assets/pdf_file/0019/114814/...
https://www.bbc.co.uk/news/technology-23194287
https://br0kent3l3ph0n3.wordpress.com/2013/09/26/dcms-misman...
(admittedly those aren't all recent, but that's how we got in this situation over the past decade). It's not just me but the Commons Public Accounts Committee.
However by now in 2020, I believe the majority of the US has access to internet over cable tv lines/docsis instead of dsl with speeds more like 100mbps, with up to 1 or 2 gbps available in big metros for $70 a month. I know many rural areas are still underserved, and of course not every household is subscribed to internet service, but it seems to be available in most places.
I’m really surprised internet in the UK seems to be an order of magnitude or more slower for such a large percentage of the population.
I roam around in South east Asia, not the developed states such as Singapore.. In Vietnam I get 60mbps, 16Mbps upload at home for 15 USD per month. I go to coffee shops they are all equipped with 40 to 100Mbps. Malaysia offered 300Mbps for $50 per month. 5 years ago.
And all plans are truely unlimited. Whats more? 4G/edge is commonly covered across the country, 8Gbytes pay as you go for.. 5$
The telecom companies in the UK, US and most developed countries are simply treating customers like cows and protect their oligopoly via corruption/lobbying of officials.
Edit: We had a problem with our BT connection recently and I almost got a 4G connection from these folks: https://www.4g-internet.co.uk/
Edit: Actual fibre connectivity was going to be a few thousand just to get a quote... :-(
Most of the population are wired for fibre and have been for a few years now. This typically starts at ~35Mbps, ~70Mbps is very common, and a few providers will do 200-300Mbps but in more limited areas. Some suppliers do 1Gbps, but these are often limited to apartment blocks in cities or towns that invested heavily in infrastructure.
This quote is misleading:
> Found some data about internet speed in the UK, wasn’t easy to come across, that puts 45% of UK housesolds have < 8 Mbps connection.
The graph below shows that this is a survey of ADSL2+ users where the technical cap is 20Mbps. They obviously can't get any better. Anecdotally this matches my expectations.
I think the reality of the current state is that most of the population can get fibre, but many haven't bothered upgrading from their ADSL/ADSL2+ packages. When I've done price comparisons there's typically not a lot of difference in price mid-market, but there are some very low end providers who will do ADSL/2+ for ~£10, where typically home broadband/fibre is £20-40.
Lastly, with things like home internet it's not useful to talk about area coverage.
> I moved to a newer home earlier this year and like most places in the UK (even in London) it only had broadband internet aka slow ADSL over copper.
Yes, most of the UK does not have fibre, but a significant part of that is fields. In reality most of the _population_ have access to fibre, but of course they are not evenly distributed. Area coverage makes a bit more sense when talking about wireless/cellular services because people move about.
There are a few areas of London that were wired up in the late 90s/early 00s in such a way that they are very hard to re-wire for fibre. The Surrey Quays area comes to mind as one. Those areas have essentially skipped fibre and are moving straight to 5G.
For those not from the UK, FTTC is a product where they run fibre connections to the local cabinet - however from there it runs over the end users phone line using VDSL2. For historical reasons this line may be aluminium instead of copper, and is often long with lots of splits and joins. As you can expect, this can degrade performance a great deal, and there's basically nothing an end user can do to get this upgraded. For PR reasons, these connections are named "Superfast broadband" and the regulator notes that 95% of the population can get one. However in practice due to old and badly maintained phone lines, real speeds are often much lower than the "minimum 30Mbps" they claim.
0: As per https://www.increasebroadbandspeed.co.uk/broadband-statistic... only 8% of the population has access to "real" fibre to their actual premises.
In practice, most UK premises now have at least FTTC available, which can give realistic download speeds of up to 80Mbps depending on how close to the cabinet you are. FTTP is the technology currently being worked on in a lot of areas but will obviously take a long time to become the norm if only because existing housing stock will need new cables to be physically installed. If you're willing to have that disruption and to pay the premium, there are also cable companies that will install dedicated lines for you with much higher speeds even if FTTP isn't yet available in your area using BT's network.
Describing 80Mbps for FTTC as realistic is also a bit of a stretch. If you live in an area with modern phone lines and happen to be close to the local cabinet, this may be your experience but isn't that representative of the population as a whole. Per https://labs.thinkbroadband.com/local/ the mean download speed for FTTC is only 34Mbps, which given that some users are getting speeds of 80Mbps should tell you a lot about how poorly it performs for others.
In any case, even if 34Mbps is the real world average download speed for the type of fibre connection most homes would be using today, that's still much higher than the figures in the article would have suggested for a typical UK home Internet connection speed.
Some takeaway points: Firstly, median speed is always lower than mean, often significantly so, as the mean speed is inflated by users with very fast connections. Secondly, the bottom 20% get very poor speeds, often in the ranges mentioned by the article. Perhaps the article should have said 20% rather than 40%, but I'm not sure that really changes the sting of the title much? Thirdly, and I think this is in some ways the most important takeaway, some users getting excellent connections appears to have no correlation whatsoever with users overall in that area. For example in the City of London, a very wealthy area generally, the top 20% get 94Mbps and the mean user gets 56Mbps. Howwever the median user only gets 39Mbps, and the bottom 20% get only 6.5Mbps - which is definitely little enough to experience the problems in the article.
And for the record, I too live in a very old village, some miles outside the nearest big city, and I pay for the best connection possible to get, which gets me actual speeds far below the "minimum advertised speed" it's supposed to reach. The problem is down to aging aluminium phone lines, which Openreach refuses to replace.
Eventually you have to realize that as a person you either have good internet or you don't. The poor experience for the 40% who have no access to fiber and don't live right next to the DSL cabinet is very real. We could debate whether the proportion is closer to 30% or 40% or 50% but that doesn't help with the issue.
To say that 4.5Mbps is typical is completely incorrect.
Where are you finding those please? I'm currently paying £18 for TalkTalk and I'm looking to switch to something cheaper.
There are other problems as well - arguably there is no functional market for internet here. The UK is quite a small and centralised but for whatever reason cannot deliver infrastructure to "rural" areas that are only 50 or so miles from the largest cities in the UK, e.g. Wales.
I have co-workers who have moved into new builds with only copper to the building.
I live in the UK with only overhead wires on my street (probably 60-80 years old copper wiring). Those wires are eyesores, and also break regularly since tree branches cut them in storms. Given Openreach's incentives, forget about these wires ever being replaced by fibre.
I have to bond all available connections to me (broadband, 4g, bt wifi hotspot; thinking even of adding a high gain directional wifi antenna to a nearby mall) to get some borderline decent speeds here
Yes, https://www.speedtest.net/result/10279588044. I share this connection with 5 other people in the evenings, thankfully only 2 of us use it during the day so video conferencing is passable most of the time.
Lived in the UK for five years and yes, it was bad. Virgin Media is the worst culprit. It marketed its broadband as 5G but has unreliable bandwidth. Not to mention this predatory behaviour. [0]
I live in the city centre and I don't even have fibre. Meanwhile all the new corrupt housing developments all get "fibre" as a perk for buying a house on land that floods.
But fellow city dweller here. I'm stuck on Virgin Media coax where I live, despite being about two streets and direct line of site from the BT exchange.
16mb ADSL2 is the best OpenReach can give me as a consumer.
In what fashion do you call this corruption?
Buying off councils, planning permission. You name it.
Yes I do, doesn't take a genius. Do you even live in the UK? I've been nice and even included the search for you.
https://www.google.com/search?q=housing+scandels+2019+uk+flo...
https://www.imdb.com/title/tt9764140/
https://www.theguardian.com/world/2019/oct/13/profits-from-p...
https://www.councilhousescandal.co.uk/
https://www.bbc.co.uk/news/uk-england-46279048
And for real experience, I lived next to a field that was repurposed for housing development and because the "deed" was lost, ie owner paid off by housing development company, it is now all houses.
Not forgetting the case of the local council selling off the library of my town to another housing development.
Pasting a load of hastily googled things doesn't advance your prejudice induced argument one iota.
https://www.fastmetrics.com/internet-connection-speed-by-cou... https://www.speedtest.net/global-index
In both of these Poland ranks similarly badly (31st/43rd) despite entirely different data sources. France ranks slightly better but still pretty bad. Note that the first site lists median speeds, while speedtest.net shows averages.
I dont think there is a single reason.
https://chrome.google.com/webstore/detail/improve-youtube-op...
Default fixed Video Quality to always match your Screen and bandwidth: always 720p, 1080p, 4k, 8k ? Or simply 480p, 240p? avoid 60 frames (vs. 30 fps)?
H264 codec (h264ify to speed up Player / save CPU electricity. H.264 encoding is fast, efficient in GPUs built before 2016 without VP8 VP9 )
Side note: ouch on those internet speeds :( I think I’m too impatient for that now.
Anyway for people suffering from this problem there's always https://youtube-dl.org/. It has some advanced options to download the right stream or even just grab the URLs to play the stream directly.
It means if you play a video at 2.0x, and then you click pause 1 minute into the video, that pause will take 1 extra minute before having an effect! I figure it's because my screen is only 60 fps, and it still displays 1 video frame per display frame, and doesn't skip video frames as it should.
The bugs been around for 2+ years...
I live in one of the biggest cities but like everywhere we have a shitty connection
I could have decent video on chat if companies cared enough to use AV1 for slow connections. Youtube is slowly migrating. The only video chat I can find so far is Duo and nobody uses it for work
Outside of some very specific things like gaming etc, I don't think I've ever watched a YouTube video and wished that it was 60fps. Even if there was a ton of bandwidth and CPU available on the client side, I'd prefer if they spent it on putting more information into each of the 24/30 frames.
They'll have diced the metrics by user device (to see if specific devices perform poorly at those frame rates), ISP (to look for incompatibilities with ISP QoS settings targeted at youtube), country, and a bunch of other metrics. They wouldn't have launched until the vast majority of users saw a better overall experience, and all regressions were at least understood.
You're right that some changes within Google aren't properly AB tested, or are pushed out at the direction of management whatever the metrics say, but this one I guarantee was.
From all of the above, I can conclude that the majority of users in your position probably prefer 720p60 over 1080p30.
FTFY
I think its good to push boundaries but making my x220 no longer feasible for YT no.
Thinkboardband have "superfast" (>24 Mbps) at 96.83% UK household/business coverage. 91% Openreach >30 Mbps coverage. Basically "Fibre" (FTTC/VDSL/G.fast/Cable/FTTP) has a 98.73% coverage atm. - https://labs.thinkbroadband.com/local/
VDSL (Pretty much the cheapest of the list above - though Virgin (Cable) have slower speed packages priced to compete with VDSL) comes in 3 flavours. Upto 38Mbps, Upto 50Mbbps and upto 80Mpbs (The upto depends on the copper cable quality and distance from modem to the Streetside VDSL cabinet).
The data they list is for ADSL2+ connections which max out in ideal worlds at 24Mpbs (When ADSL2+ was the fastest I could get I was getting 16Mpbs). Its dependant on the quality and the distance of the copper cable between the modem and the exchange the line is connected to, a much further distance)
With the push from ISP's to get more people onto VDSL or better lines the price between VDSL and ADSL is next to nothing (A cost differnce of £2 per month for my ISP).
I know many households that have been auto upgraded to VDSL by their ISP's (If you have a modem from recent years then the upgrade can happen without anyone even visiting the home).
For an example of an alt to ADSL. They live in London, they have decent 4g/5g coverage, a no contract pay per month unlimited data sim from Three cost £35 p/m and less if you commit to a contract period. I was happily pulling 100Mpbs down on that plan and using 2TB+ per month with 0 complaints from Three. Granted you have to have the luck of the draw of being in decent range of an non over utilised mast which would be harder in a more dense area like London.
Yes we still have blackspots (Mainly people who live a distance away from their closest cabinet) in the UK where connectivity is still shit and there are pushes and grants from the goverment to reduce those blackspots. I just don't buy the 40% of the UK number quoted. I think they mean that YT broke for 40% of the people still on ADSL connections.
To OP (if they are reading) buy a 4G router off eBay (They are cheap and slap a Three simcard in it and give it a spin to test speeds in your area), OK you are now going to be CGNAT'ed (on ipv4) but if that is a problem (it was rarely a problem for me) there are ways around that too using a £5 per month VPN - I lived with using a 4G connection for 12 Months just fine. I Feel bad for the people who live in the real sticks and don't have even that option to consider.
The bad parts of using a 4G conenction from Three? At the time I couldn't get IPv6 but towards the end of my time with them that was more down to my hardware than Three's problem. A slighlty higher ping then using a fixed line connection but I was happily playing online FPS's on the connection. And when I wanted to host a IPv4 server I had to use something like ngrok or spin up a VPN and tunnel to that and then use the VPN's ipv4 address. ngrok is easiler to set up but VPN will give you a static IP.
Oh and (if like what happened to me) if your local mast get knocked offline for what ever reason (A car driving into it for example... Yup that happened to me) you will have to use a mast thats further away which now has more users using it so speeds will get slower. On the plus side for that if soemone crashes into your local cabinet (Also happened to me but years prior) you are going to get knocked offline with no fallback until the cabinet operater comes out and repairs it.
I am reading. I see these reports come up regularly about 9X% being covered by fiber and yet every building I lived in and visited in London, except for one, did not have fiber available. Same experience for coworkers and friends so it's not just me.
I honestly wonder who makes these reports and how they come to these figures? They're pulling numbers out of their ass.
About 4g, I tried it once 3 years ago and it was really bad. The speed was no better than copper (the tower wasn't close enough sadly) and it had issues during peak times (evening 6 to 10 pm is the worst), I assume because of serious contention and oversubscribing. I think mobile is a perfect example of something that's great when it works (right time right place) and that's terrible when it doesn't. Considering that I depend on internet for my job, I'd rather have ADSL that works 100% of the time even if it's slower in average.
I also depend on the internet for my work too (And for my home entertainment - Games, NF, Amazon Prime, YT, etc etc etc). The reason I used 4G over fixed line is because at the time I was living in a rental place caring for my parents and my shithole of a landlord refused to allow me to have my own fixed line installed. It was eaiser to just use 4G then deal with the battle as I was only going to be living there for 12 months max.
Total complete downtime on 4G over the 12 months I depended on it for me was 2 days (1 of which was a nationwide outage - I forget the reason for the other day. I did have someone drive into the local mast - that was fixed within a week iirc but during that time I still had a connection, just a bit slower then normal). I've had similar downtime (days per year) on my fixed line providers over the past 20 years I've been paying for my own connection. (I had someone drive into my local phone cabinet back in 2016 which took it out completely https://i.imgur.com/CBiLSV7.jpg that took Openreach a week to fix)
The reason I stopped using 4G for my day to day internet is because I moved back and I could get a faster connection via fixed line.
The reason I say to try 4G is because a decent 2nd hand 4G router is going to cost less then £25 on Ebay and Three's pay as you go mobile broadband costs at most £35 for a month (No comittment pay as you go - but if your inital tests work out well for you then you can commit to a longer contract and lower the monthly bill) so even if it sucks for you you are only out £60 but if you wanted to try for even less you can just buy their 4GB payg sim for £10 and do a speed test on your own phone (though I don't think Three allow tethering on that plan - but doing a speed test on your phone will give you a rough idea of the speed you should be getting).
You can use your own phone as a modem however you are gonna be limited by the phones USB/WiFi speed and its a bit more of a pain in the ass to set up somehting like OpenWRT to share your phones connection with the rest of your house and when you leave the house with your phone your house loses internet connectivity - but using your phone would be a decent speed test for you before dropping the coin on a 4G Modem/router.
But I will say try before you commit to a 12 month contract because even though I was getting 100Mbps at the rental place on 4G when I go to my friends place I only get 70~ but as its wireless you can try the modem is differnt places around your home to see where you get the best speed. In the Rental place I was on the first floor and I placed the router on the window sil closest to the mast (There are websites that will list the towers around you.)
EDIT: The reason I say Three is because they were the best network for me at the time, you might have better luck with one of the other networks. But that is why you test with their no contract options first before commiting to a cheaper contract.
EDIT2: For me, my local mast was not over subscribed so I had no issues even during peak times but that is location dependant - mobile providers have been upgrading their backhaul in recent years (mainly in prep for 5G) so honestly it is worth reconsidering. I know I would be if I was stuck with a ADSL line (That or be planning to move. Connectivity is the new "good transport links" when house shopping, The house might be beautiful and a nice location and at an awesome price, but if the net is shit I aint moving in.)
I had Three on my mobile, don't remember the exact number but the speedtest was single digit. Moved to giffgaff when the contract expired and they tried to increase price and to force me into a new 24 months contract.
Checked a map. The masts are around the corner for both.
E.g.
https://www.testufo.com/framerates-versus#photo=eiffel.jpg&p...
It is a little unfair because you'd hope film makers use a high duty cycle exposure (I can't remember the technical name, but basically it will blur the result), but I've seen plenty of films that don't.