All you need to 'fix' this is a fast loading news website that gets enough paid subscribers to earn enough margin from subscriptions that you can pay for a news staff, an office, and various overheads.
That is a longish way of saying that 99.9% of the overhead in any modern web site can be traced almost entirely to the mechanisms by which that web site is attempting to extract value from you for visiting/reading.
If people would visit with a 56K modem and deal with a 3 - 10 second page load, then that is the bar. And any spare bandwidth you might have is available for the web site to exploit in some way to generate revenue. The more bandwidth between you and them, the more ways they can come up with to exploit that bandwidth for additional surveillance, ads, or analytics that will get them more money.
When you are the customer, which to say it is your purchasing of a subscription or articles is the only revenue the site needs in order to survive, then the things that retain you as a customer have the highest priority (like fast page load times, minimal bandwidth usage).
But when you are a data cow, a random bit of insight into a picture much bigger than you can comprehend, a pixel in a much larger tapestry, or an action droplet in a much larger river of action. Well then there isn't really any incentive to make your life better, as long as the machine we have milking you for data can get even a couple of molecules more of that precious data milk without scaring you out of the barn. Well we'll build right up to that limit.
Hilariously, the New York times tries both: you get five or so article reads (with shitloads of tracking), and then you have to pay to read more per month.
But if you're paying, the pages don't load any different. You're paying to be mined.
In paper newspapers this is the norm. You can read the front page for free, have to pay to get the rest, and the rest is still filled with ads.
The difference is the tracking. I don’t think ads are really the problem. It’s the tracking that bloats pages and intrudes on privacy, and the tracking doesn’t need to be there because other media have ads without tracking and manage just fine.
There’s a race to the
bottom here. Tracking earns more revenue, so to be competitive you have to do it. Most sites won’t stop tracking until forced to by either the basic infrastructure of the web or legal requirements. I hope GDPR will lead to the disappearance of tracking, but so far most sites seem to pretend tracking is compatible with GDPR.
I agree in theory. However, I haven't noticed the slowness in their website and the ads are well done and blend in with the webpage. They may have to redesign/rearchitect their whole website to get what you are asking for.
It should be noted that traditional newspapers include ads alongside news content and no one complains. In fact people used to sift through the Sunday NYT simply for the ads.
> the ads are well done and blend in with the webpage
That's called "native advertisement" and it's supposed to trick you into thinking you're reading a genuine article instead of an ad. I actively avoid sites that do this.
> In fact people used to sift through the Sunday NYT simply for the ads.
Back when people couldn't google for stuff, and the ads were useful because they mostly came from local businesses you actually needed once in a while.
Moreover, there is a tipping point, beyond which the ad blends too well and becomes plain deception. This is even more of a problem for journalistic publications. See also: native advertising.
Indeed. In print, you had dedicated pages to advertisement. Not optimal, but much easier to ignore. Nowadays, you never know if an “article” is simply a marketing agenda.
Yes because the signs people put around the "native ads" literally stating that it's an "ad" or "sponsored content" are easily ignorable. Get real. People pay for a newspaper with ads in it. You still have to sift through it. Ignore it? You literally have to turn the page or spend 5 minutes pulling the ads out of the newspaper if you don't want them and in some cases there isn't a way to escape the ad with print because part of the article is there with the ad. Ad block won't save you then.
Maybe they could send multiple letters and I could have some kind of client that shows me the headlines and allows me to open the articles I'm interested in.
Over 20 years ago, the San Jose Mercury-News offered exactly such a service.
Called Newshound, it let you set up to face sets of keywords (in the basic $5/month subscription), and it would email you the plain text of every article that matched the criteria, whether generated within the publisher network or from wire services.
In reply to your edit. You can use something like umatrix to block almost everything (even css) and reader mode will still work, I do this for most newspapers and works quite well.
And when I first got online in the early 90's, my ISP had an additional subscription option to get real newspaper-type articles delivered in a special usenet newsgroup hierarchy (can't remember the name of the news service itself though).
Like any community, that's defined by the people in it. Some groups are excellent, some groups are mostly dead, some display varying levels of toxicity. Overall my experience hasn't been a bad one.
Given the lack of popularity and commercial support, combined with the complexity to connect a client with an available server (compared with downloading an app from the store), I'd expect that they are populated mostly by old tech people or passers-by from universities; and any special interest group would have small user base coming from that demographic profile. Are my assumptions correct?
> That is a longish way of saying that 99.9% of the overhead in any modern web site can be traced almost entirely to the mechanisms by which that web site is attempting to extract value from you for visiting/reading.
Well, that and the fact that front-end developers just can't seem to exist without pulling in hundreds of kilobytes, or even megabytes, of JS libraries. You actually don't need all that crap to serve adverts, or really even to do tracking: people managed without it in the 90s. It's just that it's more work to get the same effect without a buttload of JS in this day and age, and most third party tracking services involve their own possibly bulky JS lib[1]. The thing is, given the slim margins on ad-serving - whilst I don't condone it - I can see why people don't bother to put in the extra effort to slim their payloads.
[1]And if you have a particularly idiotic marketing department they might want, say, tracking to be done in three or four different ways, requiring three or four different libraries/providers. This is not merely cynicism: I encountered exactly this situation at a gig a few years back.
> "You actually don't need all that crap to serve adverts, or really even to do tracking: people managed without it in the 90s. It's just that it's more work to get the same effect without a buttload of JS in this day and age"
While I sympathize with this sentiment, this is also the entire history of computing in a nutshell. Moore's Law has driven us orders of magnitude beyond where we were when personal computers first came into existence; but Wirth's law[1] has kept pace. The laptop I'm typing this on right now has 8 GB of RAM, and that's already become pathetically tiny, pretty much the minimum viable for a consumer PC; I have to keep checking my memory usage or I'll spill over into swap (on a mechanical drive) and have to wait several minutes while my computer recovers.
Performance in computer applications fundamentally doesn't improve. Stuff gets prettier, sure, and applications do more. But things will still run about as slowly as they always have, sometimes a little worse. (There are exceptions - some things like loading programs from tape, or loading things from an HDD once SSDs were invented, were so painfully slow compared to their replacement that you'd have to actively try to write slow code to get anywhere near that performance.) It's ease of programming, flexibility, and freedom of design (in aesthetics and interface) that the advance of computing technology has always enabled. And all of those are extremely valuable in their own way, and can make applications genuinely better - even allowing qualitatively new things to come into existence that wouldn't have been feasible before - even if they don't run faster or take up less of your memory.
(To understand why, think about the development of - say - computer games since the 90s. For all that we mock poorly optimized games, how inaccessible would game development be if we required them to be coded as efficiently as Carmack built Doom? For all that mindlessly chasing "better graphics" has ballooned costs and led developers to compromise on gameplay, how many games simply couldn't be translated to 90s-era graphics without fatally compromising the experience? How many projects would never have been started if we set the skill floor for devs so high that a hobbyist couldn't just download Unity and start writing "shitty" code?)
(Or think about something like Python. Python is a perfect example of something that allows devs to massively sacrifice performance just to make programming less work. If we kept our once-higher-by-necessity standards for efficient usage of resources, something like Python's sluggish runtime would be laughable. But I think you, and I, and everyone else can agree that Python is a very good thing.)
(All that being said, I'm also fairly salty about having 8 GB of RAM and a mechanical hard drive rendering my computer incredibly painful to use as technology has marched on. Discord - which I use almost exclusively as an IRC chatroom with persistent while-you-were-gone chat history, embedded media, and fun custom emotes - is an entire Electron app that eats over 100 MB minimum; Firefox is eating 750 MB just keeping this single tab open while I type this. Even with no other applications but those open, Windows 10 and assorted background processes already push me to 5.7 GB allocated. Various Windows background processes will randomly decide they'd like to peg my disk usage to 100% for ten to fifteen minutes at a time, which I imagine is because spinning rust disks are considered deprecated.
I saw a discussion on HN a few months back about a survey of computer hardware, and one dev in the comments was shocked - shocked! - to find out that the typical user didn't have 16 GB and a 4k screen. That definitely rustled my jimmies a bit.)
> I saw a discussion on HN a few months back about a survey of computer hardware, and one dev in the comments was shocked - shocked! - to find out that the typical user didn't have 16 GB and a 4k screen. That definitely rustled my jimmies a bit.)
This is extremely common in dev circles, it's an area where we're completely detached from average users. Just to make the point, here is the mozilla hardware survey that shows >50% of users having 4GB or less: https://hardware.metrics.mozilla.com/ .
If we look at the more technical users on steam (https://store.steampowered.com/hwsurvey) then only ~15% of users have 4GB or less, along with 40% having 8GB.
Oh, I recognize that Mozilla survey as actually the specific one that user was talking about! Let me see if I can track down the actual comment thread; it's probably less ridiculous than I actually remember it being.
I'm really surprised at the reaction to the resolution. I like 1080p for movies on my (too) big TV but for coding I was more than satisfied once we got to 1024x768 and haven't thought of it since. My home coding machine is a cheap dell at 1366x768 and I've always been happy with it.
I agree with everything you said. I just have a somewhat different experience with Firefox on Windows 10:
>Firefox is eating 750 MB just keeping this single tab open while I type this.
I have 127 tabs open on Firefox Quantum 61.0.1 (64 bit). It uses ~ 1100 MB spread among 7 processes. I have 6 addons enabled (Decentraleyes, Firefox pioneer, I don't care about cookies, Tab counter, Tree style tab and uMatrix).
Why do you hate spinning disks so much? And no they are not considered "deprecated", they're really the only way to affordably store large volumes of data. It's an old, venerable, and still-very-useful tech
I'm also on a system with 8Gb of ram at the moment. Firefox is using up a hilarious 4.6Gb keeping a few dozen web pages open, but the entire rest of my Linux system, including Inkscape, qCAD, and SketchUp under Wine are using only a combined 907 megabytes. So it's possible part of your problem is just Windows 10.
My i3wm environment doesn't randomly start anything I don't ask it to start and runs very comfortably with 8gb of ram heck it would run fine with 4gb and no swap. Maybe you are running the wrong os.
A tiling window manager won't save you when dealing with Electron apps.
I run Linux and StumpWM on my desktop, and recently I had upgrade to 12GB of RAM, because it turns out 8GB is very easy to exhaust these days. I currently have 9.3GB tied up, mostly by browser processes.
Yeah, this. I'm mostly using swaywm instead of Gnome in order to free up about 1 extra gigabyte of RAM for apps, but that equals about one Electron app. The only Electron app I haven't eliminated from my daily usage, though is Patchwork, so it's not so bad.
Funny thing is I use my i3wm environment on 16gb of RAM and a 4k screen. I'm actually migrating to rat poison because it's so incredibly simple and basic that it has been making me drool. I mean, look at the source code. It doesn't get much simpler than that for a tiled WM.
Load the entire GHC garbage collected runtime just for my window manager? Isn't this the same philosophy that causes people to use Electron? And that results in unnecessarily large memory footprints and runtime performance penalties?
I don't think esoteric linuxes are really necessary--I'm running vanilla Ubuntu 16.04 with 16GB RAM and top three processes are only Crashplan (~800MB), Dropbox (~460) and Chrome (327 resident, 1380 shared, per htop, with 6 tabs running). My total usage at the moment is 4.04GB out of 15.4 available, again per htop. Some of the other numbers in this thread are baffling to me.
But I don't run any Electron apps, so there's that.
But, yeah, I guess I wouldn't be able to run this same workload with 4GB RAM. That's what lubuntu is for.
Oh, no, I'm aware these are very much Win10 specific problems and I look at Linux people with a not insignificant degree of envy. Unfortunately, I do quite like PC gaming.
Modern gnome is a pig, some of this could be tracked back to gnome-shell's use of javascript and css styling if you were so inclined. Most linux users don't really notice it due to the fact that most x86 machines are crazy fast.
OTOH, try starting a modern full blown distro on something like an rpi instead of raspian and you will quickly discover that you _NEED_ more than 4G of ram and a lot of CPU just to start firefox. Its even worse if you don't have hardware GL acceleration.
OTOH, the lightweight desktops (lxqt, lxdt, xfce) really are..
While it would seem convenient to simply switch operating systems from the popular, widespread options to...whatever i3wm is, practically speaking that is seldom possible.
I find that the best way to do this is to just VNC to a windows or Mac dedicated slave computer to do graphic arts work. VNC is so good nowadays that I can use my QHD phone screen as a second monitor for all my Adobe apps- and the lag nowadays is almost non-existent thanks to super fast wifi.
Premier or Final cut don't but
DaVinci Resolve and Lightworks do.
Obviously if your workflow and thus your livelihood depends on a particular tool you should run a platform that runs it but I would question why someone with money would pick windows over mac.
Those are commercial apps for people with real money. OTOH, there are a number of unbeatable free apps that don't have linux ports. Fusion360 comes to mind, its not solidworks level , but its light years ahead of freecad.
Smart marketing treads the fine line providing rich experiences for their users based on the average internet speed vs annoyance. If the service is free they should try to get 'some value' for the content.
As highlighted, some take it way too far by trying to extract 'maximum value' which ends up being counterproductive.
Hence the rant mentioning Molochian economy, under which we operate. And that reference explains in depth why this is a very hard problem.
I like the rant too. Except maybe the bit about sending content at the speed of humans - I for one would like to take lightweight, bullshit-free content as fast as it can be sent, to pipe it to further processing on my end, in the never-ending quest to automate things in my life.
I was more thinking about e.g. running a script to fetch 3 different lightweight sites, run some personal code on it and combine the data. If the script would spend 99.9% of its time waiting on IO because of "human speeds", I wouldn't be too happy.
That said, I would be willing to bite the bullet and accept speed limits across the board if it resulted in lean web.
I read virtually all of them. Most of them provide RSS feeds, some are a bit hidden but it's Googlable.
I started with the basics BBC, NYT, Guardian, The Intercept for general news, The Conversation for science news without sensationalism, and some tech blogs.
Then I just read most of it, and follow some links to find new sources. Most of the times news start with "As reported by X", or just a link, so you can discover new sources like that.
You can also browse HN (and n-gate for the highlights) and Reddit to discover new sources.
If you add so much you can't keep up, remove some, or change the feeds into section feeds. Most online newspapers provide them. I miss Yahoo Pipes, and have yet to find a simple hosted alternative.
There is also RSS Bridge for sites without feeds (Twitter, Facebook), but I still haven't found the time to set it up.
You can also add paywalled sources to read the headlines only. You can mark them as read from the index.
I don't think 3-10 seconds should be the bar. I spent years using a 14.4k / 28k / 56k modem.
That was during the mid and late 90s.
Browsing the web where you need to wait 3-10 seconds for everything to load is not a good user experience. It's a colossal waste of time, and today we have so many more reasons to view more pages compared to back then.
We should strive for an improvement instead of trying to stick with limitations from 20 years ago.
The real problem is people developing sites now give zero fucks about resource constraints. This is exactly like lottery winners who went from being poor to having 50 mil in their pocket but then end up broke in 3 years because they have no idea how to deal with constraints.
It's also a completely different type of person who is running these sites today. Back in the day you made a site to share what you've learned. Now you have "marketing people" who want to invade your privacy, track every mouse movement and correlate you to a dollar amount instead of providing value.
>Browsing the web where you need to wait 3-10 seconds for everything to load is not a good user experience.
Im a lone owner of a website. I have enough time to accomplish one of 3 things before January
>Finish my Finance App
>update 200 pages to have pictures load based on screen type so you can load in 3 seconds instead of 6.
>Collect and compile data to create 20 more pages, all of which my 3000 subscribers actually come to my page for.
Very quickly you can see why an extra 2 seconds of loading time is not on our mind. Its important to allocate resources effectively, changing my website to load faster is limited value added vs creating content that my users actually want.
> Very quickly you can see why an extra 2 seconds of loading time is not on our mind. Its important to allocate resources effectively, changing my website to load faster is limited value added vs creating content that my users actually want.
I understand. I'm also a sole owner of a website where I'm selling a product (video courses for software developers).
My priorities are to give as much value as possible for free and also sell some courses if I can.
According to Google's network tab the DOMContentLoaded time is about 250ms to load any page on my site (which are typically 1,000 to 5,000 word blog posts with some images). From the user's POV, the page loads pretty much instantly. Then about a second later Google Analytics and a social sharing widget pop up, but those happen after the content.
The interesting thing is I really didn't try hard to make this happen. I just stuck to server rendered templates and compressed my assets. I also made an effort to avoid heavy front end libraries and only add javascript / CSS when I needed to. I basically run a heavily modified theme based on Bootstrap with a couple of third party javascript libs (including jquery).
There's a lot of room for improvement but I haven't bothered because it seems good enough. It's very possible to get the perceived load speed of a page to be under 1 second without dedicating a lot of time to it.
The most profitable news paper in my country, didn’t have a website with articles or news stories on it until earlier this year. Their page was something from the 90ies (it was probably newer), and all it really offered was info about the paper and a way to buy it.
What they do, that other Danish papers don’t, is write lengthy meaty articles that takes time to read because they actually teach you something new. There was a story on Trumps connections to Russia, it was three full pages long, and we’re talking old school news paper format, so that’s what? 10 a4 pages worth of text?
The paper only comes out once a week, because it takes time to write, but also because it takes time to read.
I’m not sure where I am going with this, I just think it’s interesting how they’ve increased their subscription amounts while not really giving two shits about the bullshit web.
They may give two shits about the bullshit web now of course, having gotten a webpage with articles. I don’t know though, I’m a subscriber, but I haven’t visited their site yet.
Or what if, we allow tracking and ads, with native speed?
There is the problem of Network that won't / cant be fixed in any short term. Then the problem with Rendering and Reflow.
The first one being a long time before everyone gets 1Gbps internet, the 2nd being even if you have 1Gbps internet it will still be slow due to all the mini scripts.
What if the fonts were there in the first place?
What if the browser actively tack every mouse movement, links etc, bringing 80-90% of all the tracking scripts datapoint, and doing so natively, sending back the data to website as requested.
No more 3 - 5MB of Scripts downloaded per site, no more CPU running of these scripts, no more 1MB of fonts. And they don't cause the page to jank. You get Butterly smooth webpages while still getting Ads.
My biggest problem is with the idea of extending the Web via Javascript and everything should be Javascript only. Rather than extending the Web Browser native function.
Unfortunately this is an idea that Apple may not like, even if the data are anonymised.
FWIW, you have effectively just described the 'app' solution.
In that solution all of the tracking and analytics, fonts, and other 'baseline' content are part of the app, which then fetches the unique content (the few Kb of story text and Mb of images) and then renders it all locally. There is even some ability to do A/B testing in that setup.
The "App" itself is basically a browser with none of the non-content UI controls that browsers normally have, that can only go to a specific URL (the content supplier).
Precisely, but we don't want to be bounded by the App Store ecosystem. And we want to improve the UX of Website, which so far hasn't been great.
As a matter of fact, may be these API for tracking, analytics should be the same across Apps and Browsers.
We tech nerds keeps throwing out new terms, Web App, Web Pages, etc, but to our user they are all the same. They want to consumer information in Text, Video and Images, and in a fast and smooth way without Jank.
You say that as if it's easy. "All you need" is to have a product that users what to pay for. But I think, all the various tries and attempts has proven that users aren't really that keen to pay for content online, at scale.
They don't need paid subscribers. Physical papers and tv have been supported by ads without tracking for decades. Companies pay for space or time.
Internet allowed advertisers to track so now we have this BS. They had many years to fix this. But tracking is the business of many companies like google and facebook. Now lot of people uses ad blocker and it's increasing.
The price an advertiser pays to be a full page ad in the NYTimes and printed with the paper is on the order of $150,000, the price an advertiser pays to obscure your entire screen with an ad is as little as $1.00.
What you're missing is that advertising rates for television and print are several decimal orders higher than the rates for internet advertising. Why that is is more complicated than you might guess, but the economics of "printing" a newspaper by sending you the text is a couple of orders of magnitude cheaper than running a printing press. Between those two realities a lot of news web sites are being crushed.
If 600,000 [0] people see that $150,000 ad, the advertiser has paid $4 per impression. At $1 CPM, 600k impressions is $60,000, but as you said this cost is at the lowest end. An ad that obscures your entire screen might cost as much as $8 CPM or more [1] and now buying the newspaper ad is sounding like a better deal.
All you need to 'fix' this is a fast loading news website that gets enough paid subscribers to earn enough margin from subscriptions that you can pay for a news staff, an office, and various overheads.
That is a longish way of saying that 99.9% of the overhead in any modern web site can be traced almost entirely to the mechanisms by which that web site is attempting to extract value from you for visiting/reading.
If people would visit with a 56K modem and deal with a 3 - 10 second page load, then that is the bar. And any spare bandwidth you might have is available for the web site to exploit in some way to generate revenue. The more bandwidth between you and them, the more ways they can come up with to exploit that bandwidth for additional surveillance, ads, or analytics that will get them more money.
When you are the customer, which to say it is your purchasing of a subscription or articles is the only revenue the site needs in order to survive, then the things that retain you as a customer have the highest priority (like fast page load times, minimal bandwidth usage).
But when you are a data cow, a random bit of insight into a picture much bigger than you can comprehend, a pixel in a much larger tapestry, or an action droplet in a much larger river of action. Well then there isn't really any incentive to make your life better, as long as the machine we have milking you for data can get even a couple of molecules more of that precious data milk without scaring you out of the barn. Well we'll build right up to that limit.