Aurexbt

What the world forgot

[Aurexbt @Aurexbt]

@0xSyrex love you bro

[Aurexbt @Aurexbt]

A software backdoor can survive after the malicious instructions that created it have been deleted from the source code.

In his 1983 Turing Award lecture, published the following year, Unix co-creator Ken Thompson explained how a compiler could learn to lie.

Start with the login program. Modify the compiler so that whenever it sees login being built, it secretly inserts a password bypass. Inspect the login source afterward and everything looks clean.

That trick seems easy to remove: inspect the compiler itself. Thompson added a second rule. When the infected compiler builds a new compiler, it copies both hidden rules into the new binary.

Now delete the malicious compiler code and rebuild from clean source. The backdoor returns, because the old compiler quietly places it back during compilation.

The attack does not live in the source developers read. It lives in the tool that translates source into the machine code computers execute. Recompiling is not a cure when the compiler itself is compromised.

Four decades later, the same problem sits beneath software supply chains: how far down through compilers, build tools, and downloaded binaries does your trust extend?

Clean source can still produce a poisoned program. The first tool you trust may be the one that taught everything after it to lie.

[Aurexbt @Aurexbt]

https://t.co/7a5RCxb7x7

[Aurexbt @Aurexbt]

@DenlitL thanks

[Aurexbt @Aurexbt]

@0xdiego404 It’s so good bro

[Aurexbt @Aurexbt]

The web was not the original dream of hypertext. It was the version simple enough to escape the laboratory.

In 1960, Ted Nelson began designing Project Xanadu, nearly three decades before Tim Berners-Lee proposed the World Wide Web.

A link on today’s web points one way. The destination usually has no idea who linked to it. Delete or move the page, and the connection can die.

Nelson imagined something stranger. Links would work from both ends. Quotations could remain connected to their original source. Every version of a document could stay visible instead of being silently replaced.

He called that last idea transclusion. You could reuse a passage without creating a detached copy, then follow it back to its original context.

Imagine every screenshot, quote, or viral claim carrying its own path to the source. Attribution would not be a courtesy added later. It would be part of the structure.

Xanadu never became the global system Nelson wanted. The Web did, partly because its simpler architecture was easier to build, release, and spread.

We call broken links, lost context, and copied fragments normal because the system that shipped first taught us what a link was allowed to be.

The Web connected documents. Xanadu wanted connections that could remember.

[Aurexbt @Aurexbt]

https://t.co/7a5RCxb7x7

[Aurexbt @Aurexbt]

The fastest signal physics allows can travel only 11.8 inches in 1 nanosecond. Grace Hopper carried that distance into an NSA lecture as a piece of wire.

This was 1982. Hopper was explaining why computer speed eventually stops being an abstract engineering problem and becomes a problem of distance.

A second can move a clock hand. A billionth of one is too small to see, so engineers talk about nanoseconds without feeling their cost.

Hopper asked an engineering building to cut her a nanosecond: the maximum distance light could cross in a vacuum during that time. The answer was 11.8 inches.

Then she asked for a microsecond. The result was 984 feet of wire. She joked that programmers should hang it over their desks so they could see what wasted time looked like.

The lesson is brutal. Every extra inch between components consumes time. Every satellite message must cross thousands of miles. Faster computation is partly a fight against geography.

That is why chips pack components close, data centers chase latency, and distance can matter even when code is perfect.

A nanosecond sounds like nothing. Hopper turned it into a ruler, then showed that even light runs out of room.

[Aurexbt @Aurexbt]

https://t.co/7a5RCxb7x7

[Aurexbt @Aurexbt]

In 2013, Bret Victor spent 33 minutes predicting programming's future while dressed as a computer engineer from 1973

Every prediction was already at least 40 years old

Sketchpad let people manipulate graphical objects directly in 1963. Carl Hewitt's PLANNER explored programs built from goals and constraints in 1969. Engelbart's NLS and Xerox PARC's Smalltalk explored spatial representations. The Actor model described concurrent computation in 1973

None of this was science fiction. It was working research

Yet in 2026, much of professional software is still written as instructions in flat text files, read from top to bottom, then executed to see what happened

Computers changed beyond recognition. The basic conversation between programmer and machine barely changed

Victor's argument was not that those early systems had solved programming

His closing line was blunt: "We don't know what programming is."

It was that the field stopped treating programming itself as an open question. Text became the default, then the default became invisible, and alternatives started looking impractical simply because they were unfamiliar

AI coding is now reviving one of those abandoned paths. Describe the outcome in plain language and let the machine construct the steps

What feels unprecedented partly echoes an idea researchers were already exploring in 1969

The unsettling question is not how far programming has advanced

It is how many possible futures disappeared because one version shipped first

[Aurexbt @Aurexbt]

https://t.co/7a5RCxb7x7

[Aurexbt @Aurexbt]

IN 2013, A MAN IN A 1973 ENGINEER’S OUTFIT WALKED ON STAGE AND SHOWED THE FUTURE OF PROGRAMMING - THE PROBLEM IS, IT ALREADY EXISTED

Bret Victor wasn’t predicting the future. He was showing ideas that already existed decades earlier - and were forgotten

-> Code should react directly to what you touch
-> Programs should describe goals, not manual steps
-> Software should exist in space, not endless text files
-> Concurrency should be the default

The strangest part? The industry didn’t reject these ideas because they failed. It simply chose another path - and forgot there was ever a choice.

Maybe the biggest limitation of programming isn’t technology. It’s the way we learned to think about it

AI is reopening one forgotten door. The others are still waiting

Read the story of the future we buried ↓

[Aurexbt @Aurexbt]

https://t.co/7a5RCxb7x7

[Aurexbt @Aurexbt]

https://t.co/7a5RCxb7x7