Welcome to MemoryHole

I’ve been tinkering with web archiving for almost 10 years. This project is the latest iteration, and I suspect I’ll still be doing some version of it in 2036.

After many starts and stops, it’s finally good enough to share. So here’s MemoryHole.

What is it?

• It’s a personal web archive.

To start, there are two ways to save pages:

1. A Chrome extension that saves exact copies of web pages
2. An RSS reader (beta, maybe forever)

Why save pages at all?

If you’re like me and you sit with ideas for years, it helps to revisit the exact page you saw back then.

Think about an important book you read when you were younger. If you flip through the pages your younger self doodled in, you’ll feel something. I get a similar feeling from old books when I notice a dog-ear exactly where I left it. It reminds me how much I’ve changed. The web can do that too. When you revisit an old thread, blog post, or silly YouTube comment, you can feel the distance between who you were then and who you are now.

I want that experience for the forum threads, blog posts, and stray comments that have lodged themselves in the back of my brain. Eventually, maybe it extends to podcasts, audiobooks, videos, and other non-text media.

Locked gates

You’d think digital documents would be easy to copy and share. They aren’t. Old web pages from the ‘90s are relatively easy to preserve: save them to disk and open them in a browser. There are no stylesheets, no JavaScript, no iframes—just tables for layout, more or less as intended.

About 20 years ago, pages became interactive, JavaScript became required, and—worst of all—you now often have to log in to view public content. That complexity is increasingly used to close off the web. I sometimes have to solve a captcha on my phone just to read an article I was looking at minutes earlier on my laptop. I pay for some of these subscriptions, on devices I own, and I’m still not in control of how the page reaches me.

The frustrating part is that these barriers mostly apply to people. AI crawlers still move through the open web at industrial scale, often with more access and fewer consequences than the rest of us.

In no particular order, here are a few signs of where things are heading:

• Cloudflare demanding AI labs pay per crawl
• Reddit is apparently now profitable because it’s licensing user-generated content to AI labs.
• xAI raised billions after it trained on Twitter comments

Meanwhile, responsible organizations like the Internet Archive are slowly getting locked out.

People add value by thinking, writing, commenting, and sharing with one another. But we have very little say when that material changes or disappears. If this trend continues, more of the web will feel like a pay-per-view event—just without the spectacle.

What’s the point in trying?

I’m not trying to stop the AI train. We’re already living with it whether we like it or not. Our RRSPs and 401(k)s, and public pensions * are part of the same bet.

I’m also not trying to make you angry. Eventually this mess will settle into something more normal. I just want to make sure that when more gates close, the things you wanted to keep are still within reach.

What about MemoryHole?

It’s an app with a funny name. It’s not going to fix the whole situation. I just hope it helps people reflect on the past.

The goal is to offer a service that lasts for years. I expect the web to get more locked down over time, and I don’t want the things that matter to disappear before we think to save them.

A lot of what we archive will be forgettable. Some of it will be worth keeping. I don’t want to end up in a future where a chatbot tells me what was worth remembering.

Free offer

On a more practical note, if you've been laid off under the banner of "AI automation", I'm sorry. If a free subscription would help, reach out to free@memoryhole.app and I'll set one up.

Even if you never use it, I hope the gesture helps a little.

A few unrelated notes

CPP Investment Board shenanigans

To Canadians, I'm sorry to be the one to tell you this. Lending $400M to xAI isn't even the worst thing the CPP Investment Board has done. 20 years ago they moved from mostly passive to active management, and the results have been rough. This shift cost us $100B in subpar returns, but I guess that's the way she goes..

It's also wild that these execs still get paid bonuses while underperforming their own benchmark. It would make more sense to pay them to do less damage.

Don't get too stressed out about your retirement because the 2025 Actuarial Report has modeled that CPP is sound for the next 75 years. You can trust the Office of the Chief Actuary. They're boring, independent, and they don't get paid bonuses for doing a bad job.

AI disclaimer

No AIs were harmed in the writing of this post.