Hacker News Reader

Comment by makecheck | Hacker News Reader

We have a ridiculously-backwards model on the web where you essentially pay for what you use (via your data plan, and via forced ads prior to promised content) without having any way to know in advance what it will end up costing you to display content. Heck, you don’t even know if the content will display correctly after all that loading. Worse, there are many ways to trigger loads accidentally, meaning you may want none of the content but you end up paying for it through your data plan.

We desperately need absolute maximums enforceable in the browser, reversing the firehose. I want to opt your site in to more data use, after I trust your site. And I expect sites to work within my limit or not receive visits.

View on HN

So while you are fiercely criticizing sites which apparently understand and leverage the reality of zero-marginal cost digital data, you assume that data caps are a necessity in a digital world.

Workers don't shovel extra connectivity into the towers when you go over your data cap. So why are you focusing on data flowing over the nearly-always-on connection you have through your data plan, rather than the bullshit terms you agreed to when you signed up for a data-capped plan in the first place?

Edit: remove unnecessary word, clarification

Data caps are like your city bus. You can buy a bulk number of rides per month at a discount, and everything over that is billed at a regular rate.

The bus doesn't magically add more seats when you make extra trips on it, but it still costs more to take the bus three times a day than twice a week.

Even though the marginal cost of your body on the bus is minimal, having the fee serves to reduce crowding (demand at that price point) as well as fund the infrastructure. Data caps are the same.

Now, you can argue that data caps are surrounded by deliberately misleading marketing and that overage charges are unnecessarily steep. But you seem fundamentally opposed to the concept of data caps.

Both of the public transit systems I use on a regular basis sell me monthly unlimited-use passes.

That’s nothing more than marketing.

Do you ride more because it’s unlimited? When you get to your stop for your office or dinner or whatever, do you go back home and then back again before getting off? Do you go all the way to the end of the line to minimize your cost per mile, then walk back to your actual destination?

Of course not, you use it as efficiently as possible to do the thing you used public transit to actually _get to_.

Do you consider the money you spend for your unlimited pass as rent for housing or office space? Do you just stay on the bus all day, working on your laptop and holding meetings?

Unlimited transit passes are similar to unlimited vacation plans: people normally need and use much less than 100% of the resource offered. People also don’t normally push 100% bandwidth 24/7 through their unlimited plans.

If everyone decides to use their unlimited transit pass in this way, they will go away or increase in price accordingly.

I definitely use public transportation more often instead of walking or riding my bicycle than I would if I had to pay per ride.

That wasn’t my point though, you still only use it as much as you need it.

That is true. Point taken.

Of course not, you use it as efficiently as possible to do the thing you used public transit to actually _get to_.

The comment I replied to claimed transit systems don't sell unlimited-ride passes. I was pointing out that some of them do. What you're arguing is they wouldn't have the capacity to handle it if every person bought such a pass and immediately ceased all activities other than riding transit 24/7, which may well be true, but is also a non sequitur.

Though since you ask, it would be a bit difficult for me to "maximize" use of my Caltrain pass in the way you're suggesting, since going "all the way to the end of the line to minimize your cost per mile, then walk back to your actual destination" is not realistic -- the line is ~80 miles long.

I did not claim that, please read more carefully.

> Data caps are like your city bus.

Hi fwip,

I don't have any knowledge about the economics or logistics of city bus lines.

I do understand something about Transfer Control Protocol, Javascript's event loop, and soft realtime performance in Linux.

Can you make an analogy between data caps and any of those concepts so I can understand what you're getting at?

The terms are not bullshit. The network does not have the capacity to service everyone at their maximum last-mile data rate, all the time. Bandwidth is a large, but finite resource. There are several ways of addressing this:

-You can divide the total capacity up equally and give everyone dedicated connections that aren't very fat. Nobody likes this, and it's very inefficient because most people aren't using their full bandwidth most of the time.

-You can charge by the byte, on top of the base fee, so that using the system more costs more. It's fair - pricing directly reflects resources consumed - but nobody likes the idea of being nickel-and-dimed whenever they click a link.

-You can do the above, but include some large chunk of data (which is quite cheap) into the base fee, that most people won't go over. Call it a "cap". This works well.

-You can promise "unlimited" data, but start sending nasty letters and/or slowing the connection to a crawl when they go over some prescribed "fair use" amount. This is a data cap in all but name.

Basically, all-you-can-eat stuff is always sold on the basis that you will consume some reasonable amount, because no resource is infinite.

Browsers should start offering the option to deny all cross origin resources too.

It would be easy to make an extension to do that but I suspect most websites would no longer work correctly. Many websites have their own API server on a different domain, many host images on another domain, most use a CDN of some kind which hosts CSS, JS and sometimes HTML on another domain, etc.

Even a simple limit would probably help a lot, e.g. “at most one alternate domain” (your main CDN, and not sketchy-analytics.com or unnecessary-ad-malware.net).

Just use one of the many ad/tracking blockers. Websites will still work. Blocks sketchy-analytics.com. Why make this more complex than it needs to be?

Because when I go to use a website its an interaction between me and them, I never want any third party involvement. Allow requests to subdomains but nothing outside that.

Use noscript then.

Me using noscript doesn't make a dent in the usage patterns of the public at large and it doesn't inform others that there is a better way to browse.

Without any decent weight behind it, there would be no incentive for people to build their software to actually respect the user and they would continue serving a broken application.

Much like browsers slowly boiling site owners by displaying sites as insecure if they don't have any encryption, a similar effort could be made to stop websites handing visitors around like the town bicycle.

Yes, “shaming” would be a relatively easy improvement too. Browsers should feel free to display big, red, scary-looking logos like “this page load has consumed an unusually-large portion of data/battery/whatever”.

uMatrix works on all browsers that matter.

Out of the box Android and iOS, no it does not. And that's a (the?) cash cow for online advertising.

I'll admit the mobile version of uMatrix on FF could work a lot better especially considering I have to disable and enable it in order to get it working at times. That doesn't diminish the fact that I like having the option to do so. Most of the time if I don't want to spend to much time configuring a website I just open it in Chrome.

I use a VPN with a hosts file that is composed of EasyList etc for my phone. It saves me money and battery.

Same here, but is performance of the server OK? Because I've had bad performance issues with ~1k lines in hosts(5).

I've thus written a script that I run daily for my local unbound(8) daemon -or bind: https://gitlab.com/moviuro/moviuro.bin/blob/master/lie-to-me

I don't notice any speed differences. I'm loading stuff via 4g mostly so I think the network speed is the main bottleneck. I'm using etherVPN on a tiny arm server in the same continent as me.

Reply on news.ycombinator.com