← field notes

The Proxy Cat-and-Mouse: How IPs Get Burned and Reborn

somewhere right now, in a list almost nobody outside the industry ever reads, a single address that has done nothing wrong all year is quietly being added to a blocklist. yesterday it loaded a hundred ordinary pages and every site trusted it. today a few too many requests went out through it, a pattern looked a little too mechanical, and a system far away made a quiet decision. nobody announced it. the address just started meeting more resistance everywhere it went.

that small, silent event is the heart of why proxies get blocked. it has almost nothing to do with the address itself and almost everything to do with the story attached to it. to understand blocking, you have to understand that an address has a reputation, and that reputation has a life cycle.

an address is a reputation, not a number

an ip address is technically just a label that tells the internet where to send a reply. but over time it becomes something more like a name. every site it touches forms a quiet opinion about it, and those opinions get written down, shared between systems, and traded.

a clean address is one no system has anything bad to say about. a burned address is one too many systems have decided to distrust. the number never changes. what changes is the accumulated testimony of every machine the address has ever spoken to, none of which ever asked its side of the story.

how an address starts clean

every address begins with no history at all. when a provider hands out a block of fresh numbers, those numbers are blank, and on the internet a blank slate is valuable, because trust defaults to caution. an address with no marks against it gets the benefit of the doubt.

so a clean address moves quietly. it loads pages without extra friction, signs up for things without being held for review, and blends in because nothing about it stands out. that frictionless feeling is exactly the resource that gets spent.

what it means to burn an address

burning is not one dramatic event. it is an accumulation. each time the address does something that looks a little off, some system somewhere lowers its opinion by a notch, and those notches add up across many systems that never talk to each other but all watch the same behavior.

at some threshold the address crosses from trusted to suspect. now a captcha appears where there was none. a request that used to return data returns a polite refusal. a signup that used to go through gets held. the address was not banned everywhere at once. it just slipped down enough lists that the modern web stopped giving it the easy path.

the burn is rarely announced. there is no error that says your reputation is low, because a defender who tells you exactly why you were flagged is teaching you how to dodge the flag next time. so the system stays quiet, and the burn shows up only as a slow, frustrating fog where everything takes three tries instead of one.

the traffic that does the damage

so what actually burns an address? mostly volume and shape. one home connection has a natural rhythm to its day, a few hundred requests across normal hours. when far more than that pours through a single address, all day, aimed at unrelated services, the shape stops looking human.

it is rarely a single offense. it is the relentlessness. a real person does not hit the same site hundreds of times a minute, open dozens of accounts in an hour, or fill a form with mechanical timing. defenders do not need to prove the address is automated. they only need to notice that whatever sits behind it stopped behaving like one person.

how blocklists get built and shared

a lot of this reputation lives in blocklists: shared rosters of addresses that have misbehaved. some are run by security companies, some by volunteer projects, some by large platforms purely for their own use. each one watches for spam, abuse, attacks, and fraud, and records the addresses responsible.

the part that matters is that these lists are not isolated. a small site you have never heard of can subscribe to a list built from complaints gathered everywhere else. so an address that earned its bad name in one corner of the internet can find itself distrusted on sites that never saw it do anything at all. the punishment travels far ahead of the address.

there are different flavors, too. some listings are narrow and surgical, naming a specific offender caught sending spam. others are broad and categorical, listing entire ranges known to belong to hosting providers or proxy networks, on the simple logic that ordinary people rarely browse from inside a server farm. getting off the first kind is far easier than escaping the second.

how reputation propagates across the internet

this is the quiet engine under the whole system. one operator flags an address, that flag flows into a shared feed, the feed is consumed by hundreds of others, and within hours a verdict reached in one place is enforced in a thousand places that simply trusted it.

reputation also spreads sideways, by neighborhood. addresses are handed out in blocks, and if a whole block earns a bad name, an address inside it can be treated with suspicion just for the company it keeps. a brand new address can arrive already tainted, not by anything it did, but by where it lives. on the internet, your neighbors are part of your reputation.

why operators rotate addresses

once you see an address as a wearing asset, rotation makes obvious sense. anyone running a lot of traffic knows each address has a budget of trust that gets spent with use, so instead of riding one until it dies, they spread the load across many and move between them.

the goal is to keep any single address below the threshold where it gets noticed. but rotation only delays the arithmetic, it does not break it. every address in the pool is still aging, and the rotation itself leaves a tell: the same accounts hopping between addresses in lockstep is a pattern a defender watching the wider view can see. the move that protects each address can give away the whole operation, which is why this cat and mouse never settles into a clean answer.

the cooldown: how a burned address heals

here is the part that surprises people. a burned address is not dead forever. reputation has a memory, but that memory fades. most systems weight recent behavior far more heavily than old behavior, because an address that misbehaved a year ago and has been quiet since is probably under new management or simply back to normal use.

so an address that goes quiet can cool down. complaints age out, listings expire, shared feeds drop entries that have not reoffended, and slowly the address climbs back toward neutral. nobody pardons it. it just stops being the most recent thing on anyone’s mind, and being forgotten is a kind of forgiveness. how long that takes varies: a light flag can fade in days, a serious listing in weeks or months, and a categorical range listing may never lift at all, because it describes what the address is, not what it did.

reborn under new ownership

the faster path back is reassignment. addresses are a finite, managed resource, constantly returned, reclaimed, and handed out again. the connection that gave up an address last month may use a different one today, and that old number now belongs to someone entirely new.

so a burned address can be reborn simply by changing hands. the history that haunted it belonged to the previous holder, and a fresh holder inherits whatever reputation is left. this is also why an ordinary person can find their brand new home connection treated as suspect. they did nothing wrong. they just moved into a number whose previous tenant left a mess that has not finished fading.

the economics of clean addresses

now you can see why clean addresses cost money. clean reputation is both scarce and extremely useful, and scarcity plus usefulness equals price. an address nobody distrusts can do things a burned one cannot, so clean addresses command a premium and burned ones get discounted or discarded.

that turns reputation into an economy with constant churn. fresh addresses are acquired, used until their trust runs thin, rested or recycled, and brought back. the price of an address at any moment is really just the market’s guess about how much trust it has left. and the market rewards whatever is hardest to obtain, which is the thread the next breakdown picks up.

why defenders rely on reputation

from the defensive side, reputation is a powerful signal precisely because it is so hard to fake. an address cannot declare itself trustworthy. it has to accumulate a clean history over real time, through real behavior, and there is no shortcut that convincingly manufactures that history.

that is the whole strength of it. anyone can spoof a label, copy a fingerprint, or mimic a browser, but you cannot fast forward a reputation. history is expensive to build and cheap to check. a single reputation lookup, almost free to perform, quietly carries the weight of everything an address has ever done.

the loop that never ends

step back and the pattern is a loop. clean addresses enter, get used, get burned, go quiet or change hands, cool off, and re-enter as clean again. the same finite pool of numbers cycles through reputation over and over, like a deck shuffled and dealt and shuffled again. nothing is truly created or destroyed; an address is just a slot that holds a reputation for a while, loses it, and gets a new one.

so this is the shape of the whole thing. defenders lean on reputation because history is hard to fake. operators answer by treating addresses as a renewable resource, spending them, resting them, recycling them, always sourcing fresh ones. neither side wins, because the loop itself never resolves. every clean address is a temporary advantage walking toward its own expiration, and every burned address is a future clean one, waiting out its memory or its next owner. the addresses keep cycling, the lists keep updating, and the war goes on, quietly, address by address.

The Hidden Internet takes apart the systems that quietly run the modern web, explained from the inside. No products, just the machinery. Subscribe on YouTube.

watch
The Hidden Internet on YouTube

Every field note starts as a short documentary. Watch the systems in this piece explained on screen.

subscribe →
read on
More field notes

The rest of the series on how the modern internet detects, tracks, and sorts the traffic that reaches it.

browse the archive →