When a platform blocks the whole carrier range

Address level friction is messy. You rotate, the reception changes a little, one draw sails through, the next catches a check, a third gets a soft refusal that eases off if you slow down. Every address has a different past behind it and the responses reflect that.

A range level rule has none of that texture. It is completely flat.

One customer rotated eleven times in an afternoon and got eleven responses that were identical word for word, arriving at the same speed each time. That uniformity is the signal. It means the decision was made before anything about his particular address was consulted, and it means he spent an afternoon paying for draws out of a bag that was never the variable.

This is a different subject from how a platform scores an individual address, which turns on history and behaviour and is covered elsewhere. Here the unit is the whole block, or the network that announces it.

The signature, before you spend money

Two rotations is enough to tell the two apart.

If two fresh addresses give you two different receptions, you are still looking at address history and the usual playbook applies. If two fresh addresses give you the same page, at the same speed, with the same wording, stop rotating. Everything after that point is noise, and the eleven rotation version of this story is what happens when nobody sets a stopping rule in advance.

Write down the exact response text and the time. A week later you will want to know whether anything actually changed, and memory will hand you a flattering version.

Four tests, about twenty minutes

Each of these rules out a different wrong answer. Run all four, including the one that feels redundant, because the redundant one occasionally comes back with a surprise.

Unrelated target, same line. Point the line at a news site or a weather page. If that loads normally, the modem, the registration and the line itself are out of the picture. Most people skip this because they already know the line is fine. Some of them are wrong.

Same target, different carrier. If you only hold lines on one network, borrow a phone, turn wifi off and open the page over mobile data. A different carrier means a different block and a different ASN. A clean result here is the strongest single piece of evidence you will collect.

Same target, other address family. A mobile line will usually give you both. Some rules were written against one and never copied across to the other. This succeeds less often than the carrier test, and when it does it tells you the rule is narrower than you feared.

Same target, no mobile involved. Home broadband, the office, anywhere fixed. This separates a range problem from a problem with your account, your automation or the thing you are actually doing.

Read the four as a set. Line works elsewhere, target works from another carrier, target works from broadband, target refuses everything from this range: that is a range level rule and there is not much left to argue about.

If the target refuses you from broadband too, close the proxy tab. The address was never the variable.

Two practical notes on running them. Use the same client for every test, because swapping from a browser to a command line tool changes more about the request than the address does, and you want one variable moving at a time. And run them close together. A range that lifts overnight will hand you a contradictory set of results if the four tests are spread across two days, and you will spend the afternoon reconciling readings that were both accurate when they were taken.

One caveat as well. Test more than one surface of the platform. I have seen a range that could not get past a login page while the public product pages served perfectly from the same line, and the reverse, where reading was fine and anything that wrote data got stopped. A rule can sit on one surface and leave the rest alone, so testing only the page that annoyed you leaves you with half a picture.

What is actually being applied

Large platforms hold rules at more than one level of granularity. A single address is the finest. A block of addresses is coarser. The ASN, the registered network that announces those addresses to the rest of the internet, is coarser again and covers everything the carrier has.

When sustained abuse arrives from a range, chasing individual addresses is a losing race, because the carrier hands those addresses out and takes them back all day long. The cheap response is to move the rule up a level.

From where the platform is standing that is a defensible decision. It is also completely indifferent to you. You have never met the people whose volume triggered it, you cannot see what they did, and nothing anywhere separates your requests from theirs at the moment the rule is applied.

These rules also come and go. I have watched a range go cold, stay cold for weeks, and start working again with no announcement from anyone. I have watched one return a fortnight after it lifted. There is no published list, no notification, and no lookup that reports the current state. You find out by trying.

I want to be careful about motive here. I do not know why any given range went cold. I can see the effect on my own lines and I hear it from customers, and past that I am guessing like everybody else. Anyone claiming to know which platform blocked which range and why is telling you a story.

Three things that help

Move that workload to a different carrier. This works most often and it costs the most, because your accounts have been living on one network and moving them is a visible change on their side. An account that appears somewhere new after months of consistency has some explaining to do, and some platforms will care. Confirm the alternative actually loads before you buy anything.

Change the work instead of the address. If the rule is reacting to a pattern, then reducing concurrency, slowing the pace or approaching a different part of the same site sometimes puts you back inside what gets tolerated. This is guesswork and I would rather say so. It costs an hour to test and nothing else.

Wait. These lift. If the job is not urgent, park it, test again in a week, and get on with something else. I have had customers do exactly that and come back to a target behaving perfectly, having changed nothing.

Waiting sounds like defeat and it is often the correct call on the money. A week of a line sitting idle costs you the line. A week of buying replacement lines, moving accounts onto them and discovering the new network has its own quirks costs you considerably more, and you may end up waiting anyway. Decide which one you are doing deliberately rather than drifting into the expensive one because it feels like action.

And the one that cannot work

Rotation hands you another address out of the same range, announced by the same ASN, which is the thing being refused. Being turned away at a door and rejoining the same queue produces the same conversation with the same person.

The reflex is understandable. Rotation is free, it is one command, and it fixes a lot of other problems. That is exactly why it eats afternoons: the cost of trying it again is so low that nobody stops to ask whether the eleventh attempt differs from the first.

Two draws. Then test.

What a provider can honestly do

Less than customers hope, and I would rather set that expectation before you buy than after.

When a target refuses lines across an entire carrier, I cannot fix it. There is no relationship with the platform, no form, no appeal, and no version of this where I call a carrier and ask them to have a word.

What I can do is notice early. I hold separate accounts on more than one Singapore network, so when somebody reports trouble I can test the others in ten minutes and tell him honestly whether the problem travels or stays put.

Then I can move people. If there are spare cards on another network I will swap a customer across, and that is a genuine new address, which puts the account history question straight back in their lap before I touch anything.

And I tell the next person who asks. If one network is cold at a particular platform this month, I say so before the sale, because selling somebody lines that cannot reach the one site he needs is a refund with a delay on it.

What nobody can promise

A range nobody has ever abused does not exist. Every carrier block on earth is shared with real subscribers doing whatever real subscribers do, and a large share of the volume leaving any mobile range belongs to people I will never meet.

That crowd is the reason a mobile address is worth having. Their ordinariness is the cover you are paying for, and their occasional bad behaviour is the exposure that comes attached. Both arrive in the same package and no seller can separate them.

The habit that replaced my dashboard

For about a year I treated every one of these as a hardware conversation. Somebody reported a target refusing him, I checked signal, registration, throughput and round trip time, found everything healthy, and replied that the line was fine.

True, and useless. He did not care whether the line was fine. He cared whether the page loaded.

What changed my approach was a customer patient enough to run the carrier test himself and send me the result: one target, three networks, one refusing and two loading. A quarter of an hour of his time answered a question I had spent three days failing to answer with instruments.

Now that is the first thing I ask for. One target, two carriers, tell me what you see.

If you want lines from someone who holds separate accounts on more than one Singapore network and will tell you plainly when the network is the problem rather than selling you another rotation, that is what I run.

Get new guides and videos first — join the Telegram channel.