Most active commenters
  • lxgr(6)
  • usr1106(5)
  • pavlov(3)
  • diggan(3)

←back to thread

98 points thunderbong | 28 comments | | HN request time: 0.812s | source | bottom
1. pavlov ◴[] No.42479172[source]
> “HTTP is also too inefficient for wireless use. By using a semantically equivalent, but binary and compressed format it is possible to reduce the protocol overhead to a few bytes per request, instead of up to hundreds of bytes.”

Around the turn of the millennium, there were numerous international committees and hundreds of millions of dollars spent by companies on this idea that we simply can’t use the existing internet on mobile phones, so there needs to be something else.

Of course for the companies it was mostly a plot to capture the web, which was uncomfortably open and uncontrolled. The mobile operators were used to charging 20 cents for sending a 140-character message and 1 euro for delivering a monophonic ringtone. They wanted to be the gatekeepers and content curators of the mobile web, taking a cut on every bit of content that flows to devices. (I remember vision PowerPoints where operators imagined that one day when video can be watched on mobile phones, they’d be making more money from each watch than the studios.)

“We must save 200 bytes on HTTP headers or the network will melt!” was just a convenient excuse to build a stack they could own end-to-end.

replies(7): >>42479197 #>>42479224 #>>42479250 #>>42479283 #>>42479399 #>>42479951 #>>42481549 #
2. diggan ◴[] No.42479197[source]
I don't know if you tried to use the web via 3G or even GPRS, but I remember I did, and it was terribly slow. Opera Mini/Mobile ran some sort of proxy service that made things faster (not sure how or what that was, I was too young to understand anything) and helped a little bit, but the best thing you could come across was dedicated WAP websites that basically were "website lite" versions some websites ran concurrently with their real websites.

And even so, loading a 0.1MB WAP website still took time. The pipes were really slow back then, and the devices not being like the pocket computers we have today.

> The mobile operators were used to charging 20 cents for sending a 140-character message.

In Sweden when I was young, it was pretty common for us to have monthly plans with unlimited text messages included (but not surfing, no one did that on the phone anyways). Even with that, WAP seemed to have served some sort of purpose, at least for me personally.

replies(4): >>42479245 #>>42479287 #>>42479638 #>>42481879 #
3. fulafel ◴[] No.42479224[source]
Yep, it was obsoleted before it could catch one. Sadly not so with IoT devices. Today's IoT device hardware would run HTTPS, JSON etc just fine, but there are all kinds of constrained weird protocols used, and lots of companies there have incentives to perpetuate it.
replies(1): >>42479822 #
4. usr1106 ◴[] No.42479245[source]
> Opera Mini/Mobile ran some sort of proxy service that made things faster (not sure how or what that was, I was too young to understand anything)

That is still running. The SymbianOS version of the Opera Mini browser still works.

From my Web server log:

    88.88.88.88 [20/Dec/2024:18:55:10 "GET /redacted HTTP/1.1" 200 75 r:- "Opera/9.80 (Series 60; Opera Mini/7.1.32444/191.361; U; de) Presto/2.12.423 Version/12.16"
replies(1): >>42479256 #
5. cherryteastain ◴[] No.42479250[source]
We're better off than that scenario, but not much better. Apple and Google ended up owning the phone operating systems so only the things that they deem acceptable are what the vast majority of users are allowed to use.
replies(1): >>42479330 #
6. diggan ◴[] No.42479256{3}[source]
I don't think their proxy server would use Opera Mini as the user-agent. What I seem to remember, was that they run this proxy which did the fetching for you, did some ridiculous compression or similar, and then sent you the compressed reply.

If I remember this correctly, I'd expect the user-agent to be something like "Opera Proxy" or "Opera Compressor", not the user agent of the browser itself. But again, I might remember this all incorrectly, was a long time ago and I was just a kid.

replies(2): >>42479290 #>>42479691 #
7. usr1106 ◴[] No.42479283[source]
> Of course for the companies it was mostly a plot to capture the web, which was uncomfortably open and uncontrolled. The mobile operators were used to charging 20 cents for sending a 140-character message and 1 euro for delivering a monophonic ringtone. They wanted to be the gatekeepers and content curators of the mobile web,

So how has this changed? Nowadays Google and and Meta are the gatekeepers. The business model has changed from billing the end customer to personal data prostitution. You sell us your private life and we give you "free" services to get even more personal data. Disregarding the ethical aspects: If you look at Google's profits and the money they can happily spend on paying fines to regulators, it's obvious that we have no functioning market economy.

In the old days one could still change between ~3 competing operators and one was typically competing on price. Nowadays you don't really have that option. Maybe every n years when you have to biy a new phone you can choose between Android and Apple, but it's a limited choice.

replies(2): >>42479304 #>>42479586 #
8. pavlov ◴[] No.42479287[source]
Both are true: WAP was technically a well-intentioned (if poorly designed) solution to a real problem, while also being a cynical power grab by the operators.

The problem turned out to be more short-lived than anyone imagined in 1999, and fortunately the power grab failed too. Steve Jobs hammered the definitive last nails onto that coffin. Mobile operators became the dumb pipes that was always their worst nightmare.

replies(1): >>42480868 #
9. usr1106 ◴[] No.42479290{4}[source]
I know the user causing those log entries personally. They use Opera Mini on an ancient SymbianOS phone.

Yes, it works like you describe. They use a compressed protocol between the client and the proxy. The DOM might not even be based on HTML, not sure about that.

replies(1): >>42479309 #
10. sabbaticaldev ◴[] No.42479304[source]
the fact that other companies succeeded doesn’t have much to do with the many that failed with terrible assumptions
11. usr1106 ◴[] No.42479309{5}[source]
Actually I would not have needed to redact the IP address. It's not the end user but Opera's proxy:

    82.145.211.80 [20/Dec/2024:18:10:36 "GET /redacted HTTP/1.1" 200 75 r:- "Opera/9.80 (Series 60; Opera Mini/7.1.32444/191.361; U; de) Presto/2.12.423 Version/12.16"
12. pavlov ◴[] No.42479330[source]
Do Apple and Google stop you from opening a web browser and navigating to any web site?

Because that was the trillion-dollar vision for mobile operator owned WAP portals around 1999. They would completely control access to online services on mobile devices. That was how they planned to get a cut on everything you view on a phone.

replies(2): >>42479744 #>>42480159 #
13. cyberax ◴[] No.42479399[source]
> Around the turn of the millennium, there were numerous international committees and hundreds of millions of dollars spent by companies on this idea that we simply can’t use the existing internet on mobile phones, so there needs to be something else.

This was not unreasonable. GPRS started to roll out around 2002. And it was quite spotty initially, to say the least. The phone hardware was also quite underpowered, good old Nokia 3310 had a whopping 2kB of RAM accessible to the software.

I got my first mobile phone in 2000 that had WAP-over-SMS, and it was quite useful. I could check the weather forecast, and my university had a nice WAP site with important notifications (scheduling changes, exam reminders, etc.)

14. orev ◴[] No.42479586[source]
No, this is a completely different scenario. There’s a huge difference between the companies who provide the pipes having full control over everything (the way it was back then) and what we have today which is that some companies have become defacto standard platform providers. It’s the difference between having to pay a toll every time you leave your driveway vs. paying for things at whatever store you decided to drive to.

Google and Apple happen to have devices that most people use, but if you want to you can buy USB adapters to use on a laptop, or some laptops have built-in mobile data.

I think people making this type of comment like “there’s no difference now” just were not around back then and have really no context on just how completely locked down the wireless carriers had things back then.

replies(2): >>42480370 #>>42480563 #
15. rollcat ◴[] No.42479638[source]
> Opera Mini/Mobile ran some sort of proxy service that made things faster [...]

In 2017ish, I've overheard from a friend who used to work for Opera, that back in their time they were using Presto (their in-house engine) on the backend to pre-render/optimise/compress pages. Think smart VPN/proxy.

Also, heck yes, Opera on "dumb" phones was an amazing experience - compared to the built-in contemporary browsers.

replies(1): >>42479671 #
16. lxgr ◴[] No.42479671{3}[source]
This was how I browsed the web on the go for the better part of a decade, before browser became good enough and prices (and download times) per megabyte cheap/fast enough to actually use the full web on mobile devices.

It was absolutely amazing. And it's still around! I have an install on my iPhone; while the app seems to be unpublished in most western countries' App Store, I can still use it just fine. I believe the Android and J2ME versions are still actively installable.

Although I just gave it a try, and the iOS version seems to not be the real deal anymore; it seems to use a media compression proxy and regular WebKit to render on the iOS side. The Android version still does server-side rendering.

replies(1): >>42481779 #
17. lxgr ◴[] No.42479691{4}[source]
> I don't think their proxy server would use Opera Mini as the user-agent.

It does, I just checked:

> Opera/9.80 (Android; Opera Mini/87.0.2254/191.361; U; en) Presto/2.12.423 Version/12.16

> What I seem to remember, was that they run this proxy which did the fetching for you, did some ridiculous compression or similar, and then sent you the compressed reply.

That's Opera Turbo, a feature of Opera Mobile (which is a full browser, HTML, JavaScript and all), which indeed compressed images and other media in a lossy way (and text losslessly, if it wasn't already at the HTTP level, I assume).

Opera Mini actually renders HTML (and executes JavaScript for a couple of seconds) server-side and then sends a binary version of that pre-render to the phone. I imagine it to be closer to SVG or PDF than to HTML and CSS.

18. lxgr ◴[] No.42479744{3}[source]
> Do Apple and Google stop you from opening a web browser and navigating to any web site?

Apple kind of does, yes. They do let me use exactly one web browser.

On my old Symbian phone, I had the choice of the built-in WebKit browser, Opera Mini, Opera Mobile, all with different rendering engines and their various advantages and downsides.

Now it's all WebKit frontends.

> That was how they planned to get a cut on everything you view on a phone.

Apple does get a cut of every web search I do, and I can't even use the search engine of my choice.

Not that I think the mobile phone operators and manufacturers of the early 2000s would be better custodians of user freedom on their devices, but I don't think it's correct to paint this as an unequivocal change for the better.

There was more design by committee back then (and the committees were as bureaucratic as it gets, just open any 3GPP or OMA specification and you'll never complain about anything W3C again), but there was also more pluralism of device design and UX aesthetics, and more than two implementations of everything, both by US companies sometimes struggling to empathize with users living in other countries and speaking other languages.

19. lxgr ◴[] No.42479822[source]
I wouldn't underestimate how verbose JSON over HTTPS is, compared to a binary protocol that maybe only sends a single symmetrically encrypted UDP packet per sensor status report.

Some IoT devices run on CR2032 Lithium cells, and even a single order of magnitude still matters there.

Network data usage can be a consideration too. Some IoT devices use satellites as a backhaul; you definitely don't want to (and in fact can't) run HTTP over something like Iridium SBD, for example.

20. ◴[] No.42479951[source]
21. cherryteastain ◴[] No.42480159{3}[source]
No, they are less direct but not less malicious [1]

[1] https://en.m.wikipedia.org/wiki/Web_Environment_Integrity

22. lxgr ◴[] No.42480370{3}[source]
> Google and Apple happen to have devices that most people use, but if you want to you can buy USB adapters to use on a laptop, or some laptops have built-in mobile data.

But if you want a smartphone like most people these days, you’re out of luck.

Things are almost infinitely better on the network side, and I don’t miss operators dictating when I get a firmware update or which app store I can use. But now Apple does that, and Google isn’t a really great alternative either for other reasons.

23. usr1106 ◴[] No.42480563{3}[source]
> how completely locked down the wireless carriers had things back then.

That was (and probably still is to some degree?) the case in the US. Here in this country SIM-locked phones have been forbidden for more than 20 years. Phone subscriptions have limited competition because the were only 4, later 3 operators. And the most aggressive price challenging service providers get acquired by the bigger ones.

But phones have fully free competition, you buy them from a domestic or international store, from your own operator, or from a competitor, just as you prefer.

24. diggan ◴[] No.42480868{3}[source]
The influence of Steve Jobs in Sweden was basically nil at that point in time when it comes to cellphones and internet on cellphones. What put the nail in the coffin there, was the launch of a new company (called 3 or "Three", was new in Sweden at least) which promised connection to a new network/stack called 3G. With 3G, you no longer needed to use WAP. We were all using Sony Ericsson and Motorola Razor (Razer?) when 3G arrived, and the iPhone was still years away at that point.
25. miki123211 ◴[] No.42481549[source]
And it's worth remembering that the company that changed it all was Apple.

They were the first[1] to ship a "real" web browser on a phone, and they were large enough to bully carriers into submissions and force them to provide them a raw data pipe, no "special" services.

When people at RIM (AKA Blackberry) saw this, they were outraged, as no carrier would ever let them get away with that. Their solution had to be far more integrated with carrier infrastructure, it was very far from Apple's "just let us do TCP/IP and we will handle the rest" approach.

[1] At least for "real" consumers, there was probably a way to get one on some phone somewhere before that, but it was obscure enough that nobody really did.

[2] source: Losing the Signal: The Untold Story Behind the Extraordinary Rise and Spectacular Fall of BlackBerry: 9781250096067: McNish, Jacquie, Silcoff, Sean

26. anthk ◴[] No.42481779{4}[source]
At https://f-droid.org you have a J2ME emulator.
replies(1): >>42482662 #
27. mxuribe ◴[] No.42481879[source]
I remember those days, and agreed things were slow...but i feel like at least the thought that went into trying to ensure code and assets implemented for mobile were better simply because devs at least tried to optimize things to improve/have the best experience possible (even with the slower pipes). Nowadays, i feel many devs fill as much javascript and framework libraries for the web experience, that we have so much bloat...and when some of these devs or their leaders are asked "what about the mobile experience for our mobile visitors/customers, etc.?"...they often respond with something along the lines as "Don't worry, we'll just send'em to the mobile app, which is managed by the separate mobile dev team....etc.". Or, maybe i',m too cynical these days? :-)
28. lxgr ◴[] No.42482662{5}[source]
I've also tried that with J2ME Loader (which is also on the Play Store), and itw works! But the native Android version still works for me as well.