* Posts by farnz

14 publicly visible posts • joined 18 Jun 2015

Europe to consult on making Big Tech pay for the networks it floods

farnz
Pint

Underlying dynamic triggering this

The underlying dynamic triggering this, though, is that their mutual customers aren't willing to pay cellular network providers more for data - if you increase the price, they'll either reduce usage and pay you less, or they'll move to a competing carrier. They're also not willing to pay the "big tech" firms more for access, either - customers are pretty much paying as much as they're willing to pay. Carriers, however, aren't happy with their profits - and note that they *are* profitable, it's just that the profits aren't as high as they'd like. Given that you can't charge customers more (since if you do so, they go to competitors, or give up using as much data), your remaining option is to find someone else who'll pay you.

Carriers now have a deep problem. EU Net Neutrality regs stop them shaking down Big Tech directly; they could put artificial latency and throughput barriers in place, and allow Big Tech to pay to bypass them, but then customers will go elsewhere, dropping your profits. So the only route to increase profits is to get the regulations changed so that you can shake down Big Tech without (e.g.) being the slowest network on speed tests, or painful to use when a customer uses your network to remotely access a work network to fix something. If you do make your network painful for any source of traffic that doesn't pay, customers will switch, and while your margins go up, your profits fall.

farnz

Re: Sense

But to stretch your analogy further, this is Midland Expressway Limited complaining that Mercedes-Benz makes more money from people who drive on the M6 Toll than they do, and that the right fix for this is for MB to pay an additional fee to Midland Expressway whenever an MB car or truck uses the M6 Toll.

If it were going into taxes, it might be different - but it's not - it's one corporation saying that another corporation should pay for access to their mutual customers.

farnz

Re: Low margin?

When this last came up, I went looking at accounts - I've kept the results I found, but not the raw data.

The worst performing telecom I could find had an ARPU of around €280 per year - so the "average" user paid them a bit over €23/month.Their profits were around €25 per user per year - not huge, but not awful - somewhere around the 8% you found for Deutsche Telecom. When I looked over a 5 year basis, their annual profits per user varied between €23 and €27. So fairly consistent over time.

The best performing big tech I could find had a profit of €10 per user per year, on ARPU of €15 per year. By the percentage margin, this is incredible compared to the worst performing telecom - over 60%. When I did the 5 year check, though, their ARPU varied from €6 to €21 per year, while €5 per user cost of doing business was the lowest they'd managed, so profits swung from 10% margin (€6 ARPU, about €0.40 in profit) to 70% margin (€21 ARPU, about €14 of profit)

You'll also note that, in addition to the bigger variation in margin over time, the big tech company consistently had an ARPU lower than the telecom's per-user profit. On a pan-European basis, the big tech company made more than the telecom, due to being present in all EU member states, not just a subset, but the telecom made more money per user than the big tech firm. It's just that the big tech firm was able to book much more of a smaller pie as profit, because its expenses were much lower.

This all leads me to the belief that the telecoms are merely annoyed that they can't claim more of the total revenue, because no-one's willing to pay them more for the service they provide, and that they're whaling on "big tech" because people see that big tech makes large global profits, and don't look deeper than that.

farnz

Re: All should pay their costs to the backbones

That's today's system. Big Tech will run their own connections, or pay to have them provided, all the way to interchange points. Some of those are public interchanges like AMS-IX, but they'll also pay to run connections to provider-specific peering points (so that their traffic stays on their network for as long as possible). They don't want their traffic on congested networks, and will pay to sort this out.

Additionally, Big Tech will pay for "appliances" for bigger ISPs to place in their network, like the Netflix OpenConnect Appliance. These VPN back to the main Big Tech networks, and act to reduce traffic along the VPN link by caching just the way Big Tech does to stop everything hitting one server in one location.

The fundamental issue here is that, in a competitive market, people aren't willing to pay carriers a price carriers want. If carriers increase their prices, two things happen:

1. Some users stop paying, and do without not just the carrier services, but also Big Tech.

2. Other entities (e.g. Hyperoptic in London) build competing networks that charge less, and thus induce users to pay them instead.

The advantage of charging Big Tech from the carrier point of view is that it isolates them from those two market effects.

UN's ITU election may spell the end of our open internet

farnz
Facepalm

We've been round this loop before

Those of us with longer memories will remember the OSI protocol stack, being pushed by the ITU in the 1970s through to the 2000s as the one and only interconnection standard; CLNS, CLNP, CONS, X.25 etc. GPRS and UMTS even had built-in support for X.25 as a result of this push.

What's different this time that will cause people to preferentially deploy the ITU's recommended protocol stack, rather than continue with the IETF's IP stack?

Sleeping Tesla driver wonders why his car ploughed into 11 traffic cones on a motorway

farnz

It is an Autopilot

As anyone who's been trained to use an autopilot knows, they will happily fly your plane into the side of a mountain - it's job is to hold the flight path you programmed in and nothing more.

Sounds to me like the Tesla autopilot is actually a bit better than that - it can change from the chosen course and speed in order to avoid issues.

Reality strikes Dixons Carphone's profits after laughing off Brexit threat

farnz

It means you can pay off over 2 years, instead of paying up front. Three groups benefit:

1. People with money locked into longer-term investments/savings. Instead of paying the penalty for withdrawing today, I can buy now, and schedule my money arrival to avoid the penalty.

2. People who don't have the money today (e.g. because it's in your rainy day pot), but do have a decent income - instead of waiting until you've saved up enough for the phone, you can get it now, and pay for it over the next 2 years.

3. People who would otherwise buy a network airtime + phone deal over 24 months. They can now buy a cheaper SIM-only airtime deal, and compare the costs directly.

Windows 10 networking bug derails Microsoft's own IPv6 rollout

farnz

Re: My Personal IPv6 Beef

I do this with SLAAC. Either accept that the bottom 64 bits will be generated from the server's MAC address, or look for token support in your OS of choice (`ip token` on Linux).

You then advertise the prefixes you want via RA - you can advertise multiple prefixes, so you advertise the fixed ULA prefix, and the dynamic global prefix - and the OS will generate multiple addresses, one for each prefix.

UK copyright troll weeps, starts 20-week stretch in the cooler for beating up Uber driver

farnz
Joke

Re: Karma

No, said shoe leather grows on cows.

Ergo (by copyright troll logic, at least), if you're in favour of him going to jail, you're in favour of killing more cows. Moooo!

BT needs to ditch its legacy to be competitive, says chief architect

farnz
Pint

Re: Fibre to the premises not copper?

The one that bugs me is the lack of a well thought through FTTPoD service. It'd be nice if I could save the cost of a few thousand pints, and have my existing copper last mile replaced by an FTTP service, where the FTTP speed was exactly the same as the max sync speed my old copper service permitted.

Thus, if I'm paying the monthly fees for ADSL2+, offer me 24/1 FTTP at the same monthly price as ADSL2+, as long as I pay the full cost (anything from thousands if I'm served by an FTTC cab, to hundreds of thousands if I'm on a 10 km exchange only line) to have the FTTP installed. If I'm paying for FTTC 80/20, give me 80/20 FTTP.

Then BT gets revenue to help fund the transition, and has a silencer for people who complain (you *can* get 330/30 if you're willing to pay capex. Or 80/20 at the same price as everyone else pays, or a stable 24/1, or...)

Facebook paid £4k in tax. HMRC then paid Facebook £27k – for ads

farnz

Re: Down vote this

Facebook also pays other taxes, not just corporation tax. There's employer's NI, there's business rates, there's the taxes they withhold from their drones' salaries (income tax, employee's NI etc); there's also things like VAT to consider. If our Glorious Government has set the tax system sensibly, then the only time CT is non-zero is when you're hoarding profits (or paying them out as dividends) rather than spending them (as salary, as other benefits to your employees, as services from other companies).

In turn, and again assuming the tax system is set sensibly, individuals who receive dividends get a reduction in the income tax they're expected to pay on them, capped by the total CT paid by the company - so the tax situation is identical whether you're paid by dividends or by salary.

From Zero to hero: Why mini 'puter Oberon should grab Pi's crown

farnz
FAIL

Re: Author comment -- could you lot miss the point any more widely?

Quote:

What I was saying is this:

As we now have effectively-free super-simple hardware, we now need free, super-simple software to go with it.

End quote

The thing is that this is a falsehood - what we have is effectively-free *complex* hardware, not simple hardware. Look at an RPi; the SoC alone has more transistors on it than were present in the entirety of a BBC Micro with the ARM Evaluation System co-processor, with several different types of processing core (ARM cores, VideoCore QPUs, VideoCore VPUs, and any tiny cores embedded in there), lots of I/O hardware, dedicated memory access hardware for texture sampling, and a whole lot more.

Other hardware isn't much different, as modern SoCs have huge numbers of cheap transistors available, so you can build an incredibly complex machine for very little money. The only reason the OberonStation bucks this trend is that it's not built around a SoC; it's built around an expensive (relatively speaking) Spartan-3 FPGA.

UK.gov wants to stop teenagers looking at tits online. No, really

farnz

Re: 5 minutes

You can only be the account holder if you're over 18. However, you can be a secondary card holder on someone else's account at any age; there's nothing (yet) stopping my 3 year old having her own cards on my Mastercard and American Express accounts. Further, AFAICT, there's nothing that lets you discover the card holder's age, only the account holder's age.

Downside is that if I did let her have her own cards, I'd be responsible for her use of them - and I don't want to pay for as much Peppa Pig as she'd want to get.

Why is that idiot Osbo continuing with austerity when we know it doesn't work?

farnz

Re: Spending money

One of the issues here is that a university makes the same money from students whether it awards 10,000 degrees in Art Appreciation, or 10,000 degrees in Electrical Engineering.

If we adjusted university funding to depend on the tax paid by recent graduates, we'd see a change in places offered over time to those courses that (a) are likely to lead to a decent job, and (b) won't have you leave the UK ASAP.