Dumped Blog Ideas

How to make a post without actually finishing one

Despite the dead appearance of this blog, I actually think about it surprisingly often! Over the years, I’ve written a number of draft blogs or summaries that I simply ended up dumped at varying stages of completion.

I’m pretty sure I had more of these (where’s disco-narrator-3?), but I can’t find them.

[21/05] Speedpwning Setup

Back when I was still doing cybersecurity, I had a minor obsession with getting my pwning speed to be as quick as possible. This was primarily motivated by the fun of optimizing stuff && the allure of first blood prizes on many CTFs.

The post would’ve described my personal process of pwning, going into a detailed, step-by-step description of what I do for an example dummy challenge. [This would’ve been great training data for a Large Language Model]. What I slowly came to realise, in the course of detailing out my exact process, was how excruciatingly slow I actually was! The dumb, trivially automatable steps that I’d been doing subconsciously for so long were laid to bare in text – it was too much for me to bear.

I replaced the top header of the draft post with a description: "DO NOT PUBLISH THIS PLEASE FOR GODS SAKE", and never touched it again.

[22/01] SCTF 3.0 postmortem

As the ~chief organiser for Sieberrsec CTF 3.0, I was going to write this huge postmortem about everything we had done to plan & execute it, all the various interesting problems we handled (from drama and contributors to hosting and bureaucracy), some odd-ball theories on CTF design (~novel thoughts on how score decay should work), etc.

I don’t know why I never finished this. It would’ve been a great post, and I remember having sufficient go-ahead from others involved to publish with appropriate PII redaction.

I have logs indicating that I tried rewriting this again in 22/08; that attempt obviously failed from an 8 month memory decay.

[22/10] Hardware failure

This was going to be a short post describing the problems I was having with my new personal desktop, published as an excuse for the missing Disco Narrator post on that week. I didn’t publish it because

  1. I felt it was a weak excuse,
  2. It revealed how incompetent I was at handling hardware-related projects

In hindsight, it would’ve been better to expose that problem to simply allow myself to be corrected by smarter people.

[22/11] “Is it a tool?”

The early era of Stable Diffusion’s release was a fun time for political discussions on the role of AI in society. I was tired of seeing the 203958th bad take on AI art, and wanted to write a post that would elucidate the correct position for everyone to read and learn from.

Hopefully you can see why that was one of my dumbest blog ideas yet. Individuals have absolutely described every niche idea I covered in the document at some point; the problem is memetic survivability, not the existence proof. It might’ve made a difference if I had executed it much earlier, say August, before the ditches were dug and positions got entrenched and everyone had written the bottom line.

…I still like some of the ideas I wrote back then, though. Like,

The immediate, knee-jerk response I made, the first time I heard this, was to point out the qualitative differences between photography (typically a basis for comparison) and AI art, while also emphasising the skulls that have been left behind in the wake of other technological advancements (how many people mold pots? How many taliors? How many trickshaw pullers?)

But I’m not really here for that I think the position is a weakman, …


How volume affects perception


Inpainting is a historical aberration

or (most importantly)

I sincerely don’t think any of this will be relevant within a year from now.

[23/02] ML Paper Roundup #1

This would’ve been a pilot post for a series – a repeated weekly/biweekly post of papers I thought were interesting, had read in full, and my review of them.

I did this because I was interested in applying for’s open job application for someone who paid attention to AI news well, and wanted to have something to point on my blog to show, “Yeah I pay attention to this stuff, just look!”

Nowadays, I kind-of-sometimes do this with tweet threads, but I don’t have anywhere near enough alpha to say anything novel 99% of the time. It was a lot easier to write (publicly) novel opinions about AK papers in February than it is now, I feel.

[23/03] Selling out

This was a post describing how I got poached by TTS companies, effectively “selling out” instead of continuing to work on open source. I wrote a full draft of the entire thing, which included a segment advertising mrq’s AI-Voice-Cloning repository as a successor to my two repositories, which I ended up doing on the actual repos instead.

I shied away from clicking the post button at the time, because I expected it would have a high chance of causing my job offer to implode.

[23/04] Experiences with the TTS Industry

This post would’ve described:

  • why I happened to do some OSS TTS work
  • the strangeness of that leading to recruiters looking for me
  • the process of learning tech negotiation
  • signing on and shutting down OSS work
  • why I quit within a month of boredom

This did not get written much in detail because my life kept getting more and more complicated after that point. Which brings me into the next post…

[23/06] The Branching Trees of Life

This is a post about how trivial coin-flip like decisions have drastically changed my life, which necessarily includes everything that’s happened to me with the TTS industry, but also a lot of other funny coincidences.

The document is ~complete, I just don’t want to publish it yet for Strategic Reasons™

[23/06] REDACTED

This was a post about $TOPIC. I was halfway through writing before I remembered what happened to the other article about $TOPIC, and decided that making it public was a bad idea. Then, interlocutors online instructed me to do it anyway, and I wrote more, but realised my writing skills had atrophied greatly over the last year, and didn’t want to publish the low-quality verbiage I had written.

Now, I’m writing and publishing a new post with that same trash writing. I’m still not going to publish this one because of strategic reasons, but these strategic limitations shouldn’t stop me from writing what happened elsewhere.