Although I've been publishing things on the internet since I was about eleven, it took me until I was seventeen to get reliable access to a computer of my own and nineteen to set up my first Tor hidden service, back when v2 addresses were a thing and one had a hope of memorizing the URL to share to others. Meaning, up until then, I was sharecropping on someone else's server: Blogger at first, then WordPress once I realized how evil Google was, then Neocities until I was harrassed off one day.

If I knew back then what I do now, then obviously I would have never put anything on the clearnet to begin with. What creative ideas did I let wither inside me because I had to constantly toe the line of being grounded and electronics confiscated for something innocuous? (So my brothers can openly joke about porn and violence against women and my parents refuse to do anything, but I wrote one poem about my ex-girlfriend that I had when I was fourteen- and while it wasn't very good by my current standards, I certainly wasn't wishing violence against the teenager- and they took all my devices and isolated me from my friends for a week...?) I can only guess how much better at writing I would be today if I hadn't had to constantly bite my tongue.

But Tor hidden services require a server, being that Tor is a client-server model. And while I cannot remember whether or not Orbot had hosting support back in 2014 when I was using it to get past my middle school firewall, I can attest that not only does Orbot have hosting support nowadays, but there are also many more options for publishing content on the "dark web" with nothing more than a phone and a little technical expertise.

For this post, I attempted to do as much as possible with that same device I spent so many hours typing into the WordPress editor on: a Motorola Moto X, Verizon edition, Android codename "ghost". It has not seen system updates in about five years, forever stuck on Android 5.1, and I have neither the bandwidth nor the patience to attempt to compile an abandoned version of LineageOS just so I can have Android 6.0 instead. For apps that require a more recent version of Android or otherwise crashed on the Moto X, I instead used the phone I had after it, a Samsung Galaxy Note 3 with the latest nightly of LineageOS (meaning some minor point version of Android 11 at the time of writing). For client-server darknets, I used ServeIt, available on F-Droid, as the backend server; it only supports static hosting from a directory of files, meaning no dynamic applications, but that would have been all I needed back then.

Tor

Once upon a time, there was a "Tor Browser" app named Orweb which was little more than an Android WebView wrapped in a very poorly-made UI. That eventually got abandoned and replaced with Orfox, which was slightly better and also allowed for niceties like bookmarks and tabs. (I am going off my memory from about a decade ago, so don't bother emailing me if I got something wrong.) Nowadays we are a bit more spoiled with an official Tor Browser. However, it seems to crash a lot, so I prefer to browser Tor hidden services on Android using Privacy Browser instead.

Orbot is best in class as far as "Tor on Android" goes, and I know of no other (non-proprietary) solution that doesn't come coupled to a browser. In Orbot, you can set up a hidden service by going to the meatball (three dots) menu, pressing "Onion Services", then "Hosted Onion Services". Ensure that "User services" is selected and then press the plus button in the lower-right corner. "Local port" should be the port ServeIt is running on, and "Onion Port" should be 80 as that is the default HTTP port. (You can make it something else, but then you would need to share that as part of the URL.) Long-pressing the service URL once created will show an option to backup the keys or copy the URL to the clipboard.

I2P

There are two I2P apps on F-Droid that I know of and have used. The primary one is slow and buggy and features a built-in browser that is even worse than Orweb was back in the day. Therefore I recommend using i2pd instead. Any browser can be the "I2P browser" if set to use localhost port 4444 as HTTP proxy; the aforementioned "Privacy Browser" has a preset in the proxy menu to use I2P without messing with port numbers.

Unlike Orbot, i2pd stores all of its settings and keys in the filesystem instead of in the app settings. (This may or may not be a security risk depending on your threat model.) In your device's home directory will be a folder named "i2pd". Inside is the file tunnels.conf. If your file manager has the option to manually open the file as text, use that and open it in your favorite text editor; otherwise temporarily rename the file to tunnels.txt to edit it.

At the end of the file, add the following:

[server]
type = http
host = 127.0.0.1
port = 8888
keys = http.dat
							

Please note that port 8888 in this example is what ServeIt was running on. If yours is on a different port, then change the number in the above config accordingly.

If you renamed the file, change it back to tunnels.conf. Go back to the i2pd app, press on the meatball menu and then "Reload tunnels", then "Open Web Console" and "I2P tunnels" to get the new URL of your eepsite.

Yggdrasil

There is a Yggdrasil app which I have successfully used to navigate Yggdrasil sites before on my main phone, but it continuously crashed on both my Moto X and Note 3. On my main phone (a Samsung Galaxy S9) it was hard to see the full IP address... and the request to ServeIt from a different device also running Yggdrasil on the same network (so autopeering would have occurred) timed out. As Yggdrasil runs on Linux by creating a new virtual network adapter so that it can coexist with an existing network stack (instead of using proxies for access like Tor) and none of my devices are rooted, it does not make sense to attempt to compile and run it in Termux or UserLAnd.

Lokinet

There exists a dump of Lokinet binaries, including APKs for Android, run by the developers... but I was immediately greeted with the following message:

Builds from this server may work, may fail, may wipe your hard drive without asking, and are entirely unsupported.

Being that Lokinet has a personal reputation of being clunky and inelegant at best and downright non-functional at usual, I did not go on with high expectations. My hopes were not bolstered by the fact that half of the Android binaries were named "anrdoid". Very classy. Truly the attention to detail I want to see in an anonymity network.

The Lokinet app white-screened and then crashed on the Moto X. It ran on the Note 3 (at least, it said it was running...?) but no pages loaded.

Freenet

There is a version of Freenet mobile on F-Droid that I've found works wonderfully on my current phone. As it uses the same FProxy interface as the desktop version, everything works the same, from adding bookmarks to managing peered friends to browsing sites. Although the app ran on the Moto X, it hung at "Starting up". It also seemed to make Orbot in the background go berserk, constantly crashing and bringing itself back up again.

As for publishing, jSite, which I use to publish freesites on the desktop, is a no-go unless one wants to spend several hours setting up a Java runtime environment and an X server and a desktop environment in either Termux or AnLinux. While I could probably do this on my current phone with little issue using desktop mode and my wireless mouse and keyboard, for most of the Moto X's useful lifetime (the battery has been shot and nonfunctional for a few years now) I did not have even so much as a Bluetooth keyboard and was stuck to tapping on the screen.

The Freenet devs, however, have provided a simpler method of freesite publishing for those unable to access an X server. Run the following in Termux:

termux-setup-storage
pkg install git python3
git clone https://github.com/freenet/pyFreenet
cd pyFreenet
										

To upload a single file. run the following command: python ./fcpupload ~/storage/shared/path/to/file

(You can also just use the FProxy interface from a browser.)

To make an updatable freesite:

python ~/pyFreenet/freesitemgr add
python ~/pyFreenet/freesitemgr listall
python ~/pyFreenet/freesitemgr update NAME_OF_FREESITE
										

Please note that it may take a great while depending on the CPU speed of your phone, your internet speeds, and the size of the freesite.

ZeroNet

There have been several ZeroNet apps for Android. The most feature-packed is ZeroNetX Mobile, which does not seem to be on F-Droid. Please keep in mind, however, that the latest public build as of writing is from December 2021. It hung on first start on my Moto X, but worked on the Note 3.

After the app downloads all assets, the UI should start in settings. Make sure to enable "Public DataFolder" and "Disable Battery Optimization".

After initialization and first start, go to ZeroHello and then open the left side menu. Wait for all zites to finish updating, then click the meatball menu by "Scribe" and then "Clone". If you're not good at HTML and want an in-browser editor and built-in comment support, you're done and can get to writing. If you want to drop your own HTML files into the zite, open your file manager and go to Android/Data/in.canews.zeronetmobile/files/ZITEHASHHERE and edit/add/delete to your delight. (Just leave content.json alone, as that contains some of your private keys.) Go back to your zite in whatever browser you're using and hit the "Sign & Publish new content" button. It may fail since the zite is new and has no seeders, but that can be fixed by sharing the zite URL with some others who may be interested in your content.

IPFS

The official IPFS client requires Android 9.0 or newer, so it can't run on my Moto X. Termux to compile the CLI binary won't run on the Moto X either since it requires Android 7.0 or newer. I did not attempt to compile it in UserLAnd, and back in 2014 I wouldn't have known how to compile anything anyway. The IPFS app in my experience does not run well on any device and has issues with retrieving data, even from peers on the same network as it. (Plus the QR code scanner makes an annoying beep.)

To compile IPFS in Termux, run the following:

pkg install golang build-essential file git
git clone https://github.com/ipfs/go-ipfs/
cd go-ipfs
make build
./ipfs init --profile=lowpower
												

From there, all IPFS commands work as usual, but you must find a way to run the daemon in the background.

Hypercore

Gateway Browser, which attempted to be "Beaker Browser but for Android", was last updated a year ago, and the developer never provided any publically available binaries, testing or not. The repo was so large that I had issues downloading it from my home connection; I had to clone it on my VPS and then move the git tree elsewhere and download the other files through SFTP to reduce it from half a gigabyte to only a few megabytes.

The README's instructions to compile it were absolutely awful and vague. When I finally got npm and Node.js working on my machine, I ran node ci to install the dependencies for the app, and at the end I got:

76 vulnerabilities (2 low, 34 moderate, 24 high, 16 critical)

I love unmaintained software!!

The build failed. What did you expect from Node.js?

Hypercore, the protocol Beaker Browser started using once they abandoned Dat, apparently provides its own CLI... unsurprisingly also written in Node.js like so much other "next internet" garbage. This was slightly easier to install, although I don't see how my early-teenage self would have even known how to look for this.

UserLAnd failed every npm command with "Illegal instruction". AnLinux in a Termux install got as far as running npm install -g @hyperspace/cli before npm crashed at the end with the message "Exit handler never called!"

Hosting on mobile comes with two obvious challenges:

1. When using a client-server model, the phone will likely have less "uptime" as far as the network is concerned than an actual server running somewhere. Not everyone has the benefit of a persistent Internet connection, much less one fast enough to handle more than a light load of traffic, or the battery life to safely have one's device connected to a network and pumping data 24/7. While mobile data can help alleviate network availability, not everyone has an "unlimited" data plan, and even "unlimited" ones often have a set amount of data at high speeds usable every month before the user is kicked to a subpar speed for the rest of that billing month, which may not be fast enough for hosting purposes. These mobile servers would be even more vulnerable to DDoS attacks. Being that phones are used for a hell of a lot more than just hosting and the trend of using chat apps instead of traditional SMS, network congestion and increased battery drain caused by a flood of traffic to one's hidden service could lead to a life-threatening situation.
2. When using a peer-to-peer model, not all mobile networks are friendly or accomodating towards peer-to-peer traffic. In some networks, whether mobile data or WiFi, said traffic may be filtered or blocked or even illegal in some jurisdictions. With ZeroNet, for instance, a VPN can be used to circumvent any blocks for the purposes of viewing content, but until one gets a substantial amount of peers on their own zite, said zite's availability is restricted by one's ability to get a certain port open on the network, which is often not possible with free VPNs or on mobile networks.

Despite these challenges, the state of hosting one's content on darknets via Android is much more stable and accessible than it was several years ago. But many of these darknets still depend on being able to access the wider Internet to reach the servers responsible for controlling the darknet: Tor is heavily centralized in the sense that the network depends on the existence of a few consensus nodes, and the app used in the ZeroNet section of this post depends on Github to retrieve the Python 3 runtime and other dependencies, and Yggdrasil requires at least one public peer configured in order to talk to devices outside of the local network. As much as I dislike the state of ZeroNet's development, it would probably be the most functional in a situation where Internet access is limited and one only has an Android device and needs to share information, followed by Freenet.

As far as creating that information, whether webpages or books or other media? Well, that is a problem for a later post.

I have successfully managed to microwave my brain. Spending forty-plus hours squeezed into a single month on a project you wanted to make for several years and then getting into it and finding out it's just tedious data entry all the way down will do that to you. Doing not much of anything else, writing or making art or even going outside, has made me go animalistic and crazy to the tune of an oral fixation of gnawing on chunks of wood stolen from the organics waste bin at the side of the house. Not the greatest idea I've ever had. Hopefully "start working on a new book of poetry" will be a better one once the mental fog starts to lift. I hope this isn't what remote work will feel like once I do get a job.

Speaking of excessive technology use driving people batshit insane, I had a dream last night where a bunch of techbros had forced me to stand in front of a tribunal and explain why Let's Decentralize doesn't cover web3 projects. Neither my argument nor theirs made any sense, being a dream and all, but I remember "your shit is obtuse and doesn't make any sense to a layperson" not being taken for an answer. But since at least two people have emailed me asking if I could shill their crypto projects and the only response I've pulled out so far is "I don't do free labor", I decided that I would spend today (yesterday, if you're reading this on the day of publication) actually researching web3 so I can have an actual concrete argument when the tribunal does come.

Filecoin

Filecoin was the first web3 project that came to mind. I mean, it has the name "file" in it! So one would think that the whole point would be to host files online. And what is a website other than a pile of files?

Since I was already familiar with IPFS, which Filecoin is apparently built off of, I thought that setting up a Filecoin node would be as simple as compiling and installing the IPFS node is. Apparently not. The main website felt janky and broken with a "loading experience" spinner than just spun forever, and the homepage displayed little to no relevant information other than a glorified version of "we store files", which one could already guess from the name. Going to "Build" and then "View Filecoin storage tutorials" gave an IPFS resolving error. Digging on an external search engine turned up the GitHub for the Lotus daemon, which allows one to interact with the Filecoin blockchain, but the "getting started" documentation at the end of the page also 404ed with an IPFS error. Because what you truly want to see when researching if a new technology is trustworthy or not is to see it crashing and burning at every turn.

Strangely, a different link to the documentation was on the sidebar. "Surely we're getting somewhere," I wondered. "Not even figuring out IPNS took this long."

If the destination was "imagining myself slamming my fists into my keyboard", then sure, we got somewhere.

The Lotus daemon has absolutely ridiculous hardware requirements: it would cost more to acquire 128 GB of RAM and the energy to power that than I'd ever earn with Filecoin, assuming I was ever in it to earn anything and didn't just want another mirror for my website. Lite nodes are slightly better, but still way higher than anything I have: my VPS has only 8 GB, and my desktop has 16, and there's no way I'm running a bandwidth-intensive application like this from my home connection or paying more for a beefy VPS from Contabo when it would all just get sucked up by a single application. Besides, it wouldn't even work for long: the chain grows at approximately 38 GB every day. With the storage my current VPS has, I'd be out of disk space in four days!

What seems to be common for web3 projects is that, if one doesn't want to run a node or doesn't have the resources to do so, the only way to interact with the network is by interfacing with a third-party API, which kind of defeats the point of "decentralization" if my only method of access is forever mediated by someone else.

So Filecoin is out.

Siacoin / Skynet

I did not go in with high hopes. My first introduction to Siacoin and its associated projects was a huge callout post warning about how shoddy and, as we zoomers say, "sus" the development team is. I strongly urge anyone considering going into web3 to read through that post, even if not going to Siacoin per se, since many of the arguments about monetization and faux-decentralization will carry over into anything web3-related.

Despite this, I was still willing to give Skynet a try. I mean, it bills itself as "a collection of cryptographic protocols for storing and retrieving data over the Internet." What is a website if not a collection of retrievable files over the Internet? Nothing in that definition explicitly excludes websites, even if web3 seems to be mainly comprised of a bunch of shoddy Node.js webapps.

This time, mainly due to the influence of the callout post, I went on a little tour of the community first before getting into the documentation. I wanted to test the waters, to see the typical userbase of Skynet, to judge the competence of those who I would be trusting with my data. After all, web3 has quite the reputation for scams and money laundering and general financial incompetence. Given that I am currently unemployed and dependent on the goodwill of my parents to have a roof over my head and food in my stomach, I don't exactly have the "disposable" income to be burning it for the purposes of playing with "magical Internet money".

I should have expected less than nothing.

One of the example apps is clearly written by competent developers I can trust with my private files. "Always make soure, you're logged in!" Yeah, I'll make "soure"... to stay away from cloud storage that's effectively a black box concerning where my sensitive data goes.

One of the top posts in the subreddit is a "meme" about a USB stick that uses Siacoin for umlimited storage. The "meme" (in quotes because the only humor value is in how half-baked the ideas within are) claims it would require "no installing" and "no need to understand crypto". Only about one or two commenters stopped sucking the OP off for long enough to realize that such a device would require a persistent Internet connection and enough crypto knowledge to get a Sia wallet set up to actually buy whatever storage space ends up getting used and also for drivers to be preinstalled on every operating system in existence in order to require no installation of additional software and to appear just like any other flash drive.

The rest of the subreddit is just the standard crypto whining about Binance/Coinbase trading and praising the main devs for the "new Internet" they have supposedly created.

So what if I want to actually access and host content on Skynet? You know, the reason we're here?

With IPFS, I just have a node on my computer running on localhost port 8082 (yours may differ) that exposes a web UI I can use to view and download content. My website is set up to use DNSLink, so you can just throw the regular domain into any IPFS portal, whether local or hosted by someone else. For people without domains, they can still use IPNS (which allows for updating content at the same hash), just by distributing the hash itself instead of linking it to a domain name. All of this can be done from a computer I already own that is powerful enough to run an IPFS node, which I've been able to successfully do from a toaster with 2 GB of RAM and a 100 GB hard drive shared among the whole operating system, far lower than the requirements for a Skynet portal.

Skynet also has portals, which allow for free viewing of content... but unlike IPFS, where I can just run ipfs add -r /path/to/content and get a hash I can then throw into ipfs name publish MYEXTREMELYLONGHASHHERE, the design of Siacoin mandates that one pays for the storage space one uses. Prices are set by each individual portal, but portals often offer an extremely restricted free tier, which just throws those without money back into the sharecropping paradigm. And if a portal goes down, both free and paid users lose control of all their data on said portal. One could run their own portal to avoid shutdowns, but that costs money beyond the initial expense of the hardware and bandwidth and electricity to get a server in the first place.

Oopsies, looks like you need money to participate after all!

And also you apparently need a LastPass account. A centralized proprietary service to set up a decentralized Internet? (No, a CLI client does not count.) Sia is clearly run by clowns.

Arweave

By this point I was ready for the post to be done. But still I persisted. Surely there must be something out there other than IPFS that won't demand access to my money?

Arweave seemed promising at first, but the second tab on their subreddit's header says "Get free tokens now", implying that tokens actually cost money. Under the tab next to it, "Technical", there is supposedly information about hosting content on the Arweave network, but it 404s.

After retrieving a saved copy of the deploy page from the Wayback Machine, I installed the NPM package used to deploy files... and failed, since the tool required a keyfile and, for whatever reason, that command wasn't listed first. After generating said keyfile, I tried again... and got an error about insufficient balance. As expected. But hey, at least I tried. And didn't have to buy a supercomputer to do so.

sudo npm uninstall -g arweave-deploy, and into the trash bin it goes.

If the whole point of web3 is to decentralize everything, then I don't want to use third parties to host my website! I want to host my own stuff without needing a supercomputer or a persistent connection or static IP address!

As far as serverless hosting goes, we already have Freenet and the gazillion ZeroNet forks and whatever Beaker Browser is using nowadays and IPFS, the latter of which works just fine without the gazillion "blockchain domain" scams. In fact, since most of these web3 projects are built off IPFS anyway, why go through a middleman? Because these projects incentivize people to host your shit? If the content is popular enough, there doesn't need to be a profit motive for your content to stay alive and be well-propagated; it'll just happen naturally.

How are low-income and other disenfranchised people supposed to participate in web3 without just becoming sharecroppers all over again? With the profit motive, and since storage space is limited, nodes have an incentive to seed anything a wealthy person wants and give the leftover scraps of storage and bandwidth, if any are left, to the "free" users. If microtransactions are required to view anything, then how is a person with little to no "disposable" income supposed to discover new content they may like? (Of course, a counter-economy of non-commercial content sans paywalls may very well rise up in response.)

I say: let them. Let them lock down their works so tightly that they become utterly inaccessible. Let them miss out on the money they would have earned from now-disgruntled customers. Let the corporations destroy themselves in building a dam to maximize every dollar flowing to them only to find their river is drying up.

I wish that web3 evangelists would realize that the average layperson doesn't give a shit about decentralization in the computing sense. They don't care about making their files as accessible as possible, just to themselves. To them, Google Drive or OneDrive or whatever other cloud storage services are in vogue are plenty good enough for them. They would rather see ads and sharecrop on centralized social media platforms than open their wallets to pay for an inferior service. And of those with the technological expertise to operate one of these nodes if we really wanted to, well... I know that I wouldn't want to live in a world where every interaction with my fellow (physically) human beings is monetized. And I doubt that many others would either. If improving the Internet is the goal, especially on this increasingly fragile planet, replacing the current system with something functionally worse but multitudes more energy-intensive is not the solution.