It is not a question of "if" but "when" AI voices will dominate public online spaces. There are many reasons for why AI will become the majority speaker in public spaces: monetization potential by operators, political motivations by corporate and state bodies, etc. And there is no question as to whether AI voices will be indistinguishable from human voices in the future : they already are.

In 10 years or even 5 years from now, I cannot imagine a situation where in any currently existing online space, it is possible to discern an actual human from an AI agent posing and operating as a human, for whatever motivations.

If this is a scenario that concerns you, there are two steps you can take now to "inoculate" yourself from the coming epidemic of AI, as it were:

1. Establish a cryptographic identity -- there will be a date in history, perhaps assigned arbitrarily, that will be called the date at which new "users" cannot be distinguished between AI and humanity. Creating a public and private key TODAY and distributing the public key around will serve as a proof of humanity pre-dating the AI takeover of the internet. One public key should be created for each online identity -- and despite our side of the internet's general tendency toward anonymity, a public key for a "real name" should definitely be created and published as well. When necessary, "online identities" can and should be cross-referenced against an operator's "real identity" -- given that all communication in a private-key public-key environs is encrypted, anyway, with "plausibly deniable" text, it should not be a concern what is addressed in a state of forward secrecy.

Purchasing cryptocurrency remains a viable means to "establish identity" as it

(a) requires a financial stake to occur, and (b) is linked with a cybernetic identity, with public + private keys, which give people a contact address and a means of private, personal communication.

2. Purchase a domain name that is obvious to the greater community in identifying you in some way, shape, or form: even if AI agents may dominate Twitter, Mastadon Network, et al, paying real money to a quasi-governmental agency is a measure of stability, if not a proof of identity. As with point 1, this should be done sooner, rather than later.

Even without encryption tools at his disposal, as long as someone is renewing a domain name, an objective log of ownership of the domain name continues: a sudden shift in domain name ownership should set off alarm bells in cautious watchers.

To conform with my principle of vaccination against AI-fication of the WWW, grab my public key for the email address 52 at 4x13.net here: https://4x13.net/me/key.txt

Inspired by my friend vulonkaaz's "chan-like" site, //paperchan.club, an imageboard that only allows user-drawn images to be uploaded, I've began work on a new experiment I am tentatively calling Voxchan[^1] that only allows recorded audio to be uploaded. Paperchan is an exciting community for a number of reasons. Due to depending on original content that takes some level of effort and creativity, a lot of cool art pops up there every day, and the steady stream of cool art getting posted encourages more cool art to keep getting posted.

I don't think I've ever seen a voice clip-based platform before. Sure, there's been group calling apps like Mumble, Skype, Discord, et all for a long time, and there's plenty of places to upload songs, but a forum where people post by speaking seems to be entirely novel. Vocaroo encourages anonymous audio uploads but it's more of a file-hosting service than a community.

Is voxchan an anonymous site? This is a fun problem to think about. People's voices are largely unique, but I feel someone is unable to be "doxxed" by posting with their voice alone.; and, with a little effort, people can also alter their accent, tone, and other bits of personally identifying data to some degree.

There's a hypothesis I have that making posts depend on voice will have a positive effect on the quality of contributions. On an imageboard or textboard it's pretty easy for 2 or 3 guys to try and derail every thread but appear as the voice of the community. Samefagging takes significantly more effort to do via voice; most haters and losers have awful voices too so I think they would rather not put in the effort it takes to keep up a trolling campaign. And their parents would probably yell at them for making too much noise on the computer.

There are a lot of lonely people in the world, especially in the world of chan-like sites, and I think that being able to listen to the discussions instead of just reading more text could have a positive effect for them. People who are otherwise scared to do "voice reveals" in communities like imageboards, Discord, IRC, et al may also be more willing to speak in an anonymous forum.

Only time will tell how it will play out. A public release may be ready in a few weeks. If you're interested in trying out the in-dev software, drop me a line.

[1]: Vox being Latin for "voice", "sound"

I recently had a friend end up in the hospital, and he was able to bring a phone with him. He didn't have XMPP on his phone and I didn't want to install Discord.

The solution? Email!

I have plenty of old email addresses from friends. We've left Facebook and we've changed numbers but our email addresses are here to stay. The only kind of person who is unable to keep an email address is so mentally unwell, they probably aren't worth conversing with anyway.

Outside of work or school, I don't think anyone uses email. It's a damn shame. Not everyone has Telegram/Xmpp/Discord/Instagram, and not everyone wants to share their phone number (for Whatsapp/Signal). But I think virtually everyone has at least one email address. Making a new email address takes minutes and after it gets added to your phone or desktop it's as easy to use it as any other account.

Emails can be as long or short as you want, much like a blog post. And, generally speaking, there is no expectation to answer an email immediately.

One way to categorize the modern internet channels is by speed, where

• "fast" web - discord, tiktok, livestreaming
• "moderate" web - xmpp, irc, fediverse, reddit
• "slow" web - discussion forums, RSS, wikis, email

Probably the most controversial categorization would be calling fediverse a moderate-speed channel despite the fact that posts are often coming in at the rate of 30 an hour or more. But the tendency of the network is to post and contribute mindfully. The typical fediverse post will be shorter than a typical blog post, but will be more meaningful than a chat blurb on Discord or shitpost on 4chan. In the same way, emails generally are more pleasant and meaningful to engage with than fediverse discourse.

This is an open invitation for people to email me. Fediverse and Gikopoi may be faster ways to get ahold of me, but maybe you'd enjoy getting back into writing email. I always enjoy receiving and sending emails. You can find my email address on the contact page.

Related discussion: It's good to send emails to friends, Pohon BBS, 2024-01-12.

A rare but persistant gikopoi bug has been troubling me for years, and today, I finally found the cause.

Gikopoi is a free chat game with streaming options. 99.9% of the time it functions perfectly well. But once in a blue moon, people attempting to stream video/audio results in a strange bug: others are unable to grab the problematic stream and attempting to grab the stream results in a silent failure. Additional open streams also become broken once this bug is triggered. The streams are not broken from the start but become broken within a few minutes of beginning.

• Restarting the Giko server doesn't fix it
• Restarting the Nginx web server doesn't fix it
• Restarting the Janus media server doesn't fix it

I still don't know exactly what triggers the bug. It seemed like only certain players trigger the bug, and I couldn't find any useful data in logs to fully understand the problem. But today, I found how to resolve it:

In the Gikopoi server directory, a file "persisted-state" exists that keeps a json of all users with data like their ID, name, last message, character, etc. This is useful for cases when the server resets -- everyone active resumes playing like nothing happens.

For some reason, when streams break on gikopoi, shutting down the server and clearing out this file makes it work again. So I'm suspecting that streams may break if users fail to exit the game properly... or maybe players being logged on for too long triggers the bug... I honestly don't know. I have a saved copy of persisted-state I can compare against a new copy of persisted-state the next time streams break.

I really enjoyed playing an old Flash game known as Gikopoi. However, circa 2021, when Adobe pulled the plug on Flash, Gikopoi's days were numbered. What happened between then and now is a long story, but long story short, I now host the most active Gikopoi international server, and I set up some services to support the community, even writing a few from scratch.

This post will look at the various services that are not the actual game/chat of Gikopoi.

Continue reading...

This is just a brief review of known Gikopoi blogs that exist on the internet. If this post hurts your feelings, then I hope it motivates you to improve.

Akai.gikopoi.com

This is a really great blog. There is a strong focus on text and the headers & footers don't take up a disgusting amount of space. Frequent updates, open source, cute icons, and there's even an Atom feed. Tying the blog into the author's game and including trackers for life milestones make it even better.

The design will be polarizing -- my wife, who is not a "techy" finds the design frightening and unpleasant. I think it's a breath of fresh air.

Verdict: 9.2/10 -- just docking points so you have motivation to work harder -- I believe in your potential!

Shaddox.neocities.org

If akai.gikopoi is extremely web 1.0, Shaddox.neocities is extremely web 2.0: there are lots of gradients, animations, advanced CSS, and JS enhancements in play here. Both sites are iconiful but Shaddox's modern presentation gives off a more professional smell.

In contrast to akai's blog which covers a bit of everything, Shaddox's blog focuses more strictly on "otaku culture". The use of images in articles such as in his tutorial on "general streaming" are delivered to strong effect. Waifus and radio improve the reading experience.

Despite the slow release schedule, every article so far is quite high quality, and it's easy to get alerted to new posts thanks to the built in RSS feed.

Verdict: ⭐ ⭐ ⭐ ⭐ -- open sourcing the code would be an easy fix towards improving this blog's rating. A favicon could also add some more personality to this already quite delightful blog.

Temple-moss.neocities.org

A different kind of web 1.0 blog than the akai project. Text and background colors clash, the date format is European for some reason, and the only image is way too big. Updates are few and far between and there is no way to subscribe to new posts. This guy could spend a bit more time on his blog.

Verdict: 👎 -- you have a lot of soul, invest more of it into your work!