IPFS and dat have similar aims but differ in many respects:
|Team||Portland, Denmark, Oakland, Taipei, Berlin…||Concentrated in Palo Alto|
|Funding||$1,370,000 from the Knight Foundation & Alfred P. Sloan Foundation||$52 million from Union Square Ventures and others|
|Principal aim||Powerful data sharing from your desktop||the Distributed Web|
|Default mutability||mutable (you can modify datasets)||immutable (addresses always refer to the same entity)|
|Reference implementation license||BSD-3-Clause||MIT|
|Plan for keeping information online when your laptop turns off||hashbase.io, PublicBits.org||FileCoin|
|Cross-referencing a website to a decentralized website||.well-known dat file||dnslink|
A potential point of confusion is that the ‘decentralized web’ is low on dat’s stated priorities, but it’s the top priority for IPFS. But dat has a working, easily-installable web browser, and IPFS doesn’t. So, ironically, it’s currently much easier to create a website with dat than with IPFS.
As I write this, the buzz surrounding an IPFS-adjacent project, FileCoin, is heating up. It’s IPFS’s plan for distributed storage, and is launching in the flashy 2017 style of an Initial Coin Offering. If you’re very wealthy and brave, you can invest in it. This is not investment advice.
To compare the end goals of the FileCoin & dat approaches:
Anyway: this is not meant to be a buyers guide or technical deep-dive. I’m going to use IPFS and see how it works.
Okay: so, there’s no Beaker Browser-like downloadable desktop application for
IPFS. There’s a command-line tool called
ipfs. Because the reference implementation
of IPFS is written in Go, they can distribute this command-line tool as one file
that you can download and that doesn’t have any dependencies (you don’t have to install Go).
The user experience of downloading IPFS is acceptable but not great. From the homepage, you click ‘Try it’, ‘Install IPFS now’, ‘Install IPFS for your Platform’, and then ‘download go-ipfs’. The flow has some kinks:
macOSever since 2016.
go-ipfs, but the path until that point talks about installing IPFS, not go-ipfs.
Okay, so I’ve got
go-ipfs_v0.4.10_darwin-amd64.tar.gz downloaded, so I unpack it and move
/usr/local/bin so that typing
ipfs on the command line works. I run through
IPFS’s first run experience:
~/Downloads/go-ipfs〉ipfs init initializing IPFS node at /Users/tmcw/.ipfs generating 2048-bit RSA keypair...done peer identity: QmVf5UwKt2E5foja4y9gsXXCPhb3bnsR9wTR6PejbtYdFT to get started, enter: ipfs cat /ipfs/QmVLDAhCY3X9P2uRudKAryuQFPM5zqA3Yij1dY8FpGbL7T/readme ~/Downloads/go-ipfs〉ipfs cat /ipfs/QmVLDAhCY3X9P2uRudKAryuQFPM5zqA3Yij1dY8FpGbL7T/readme Hello and Welcome to IPFS! ██╗██████╗ ███████╗███████╗ ██║██╔══██╗██╔════╝██╔════╝ ██║██████╔╝█████╗ ███████╗ ██║██╔═══╝ ██╔══╝ ╚════██║ ██║██║ ██║ ███████║ ╚═╝╚═╝ ╚═╝ ╚══════╝ If you're seeing this, you have successfully installed IPFS and are now interfacing with the ipfs merkledag! ------------------------------------------------------- | Warning: | | This is alpha software. Use at your own discretion! | | Much is missing or lacking polish. There are bugs. | | Not yet secure. Read the security notes for more. | ------------------------------------------------------- Check out some of the other files in this directory: ./about ./help ./quick-start <-- usage examples ./readme <-- this file ./security-notes
Quick notes here:
ipfs init, it refers to
initializing IPFS node at /Users/tmcw/.ipfs. In Getting Started, this is referred to as a ‘global local object repository’. Is it a node or a repository? Also, how can
global localbe a compound adjective? 🤔
ipfssays ‘Global p2p merkle-dag filesystem.’ and the readme says ‘interfacing with the ipfs merkledag!’. Is it merkle-dag or merkledag? The IPLD spec also refers to ‘merkle dag’.
Apology for the pedantry of this post. That said, IPFS and the decentralized web is introducing a lot of new lingo, as they replace an ever-increasing portion of web technology. Technical writing is difficult. Words are work. If we’re going to have new words, clean and consistent usage will make things a lot easier for everyone involved.
Okay, so I’m running
ipfs daemon, which has the following output:
Initializing daemon... Swarm listening on /ip4/10.6.117.62/tcp/4001 Swarm listening on /ip4/127.0.0.1/tcp/4001 Swarm listening on /ip4/184.108.40.206/tcp/4001 Swarm listening on /ip6/::1/tcp/4001 API server listening on /ip4/127.0.0.1/tcp/5001 Gateway (readonly) server listening on /ip4/127.0.0.1/tcp/8080 Daemon is ready
There’s a pattern in IPFS that differs from dat: where dat pretty early on
dat:// URLs for its own, IPFS has been slower on the uptake and instead
refers to IPFS resources like a filesystem, like
This is fine, but this output seems needlessly confusing:
why refer to
/ip4/127.0.0.1/tcp/8080 instead of
added QmZKTxWvJZADMXTRfYJKKgci1oGq38f6Tyb1K3onv4AQxL _site/projects added QmU6REJYTGSXogDYzB6k2MmVzSDC5fdtrEhcdNSLTzENFV _site/resources added Qmc4LtYiBWRsQnanVTHHSbSq6jm6c1hkCqrbRhfB2jZAeD _site/resume added QmSFpHibaTzYUNjZjQ5qh1WJrWrR2r28Uhsoftayef1DMy _site/simple-statistics added QmbvADibAvBgusnaaNTVsqrhwteZuRc2XngyrK7KC2tBjm _site/swift added QmXHHgNSQVf2KjdAsxxXeVG9fN582CLFhDxdwvJyjMPVDP _site/talks added QmZmEkw6srCD2mCLQV4SpPbqFpKES4kX3d8wW2tRBHeuHf _site/topics added QmR96iiRBEroYAe955my8cdHkK4kAu6SApGC2eS5kSzHwt _site
I have the IPFS daemon running and the way to access one of these files is to put
it in the $HASH part of
http://localhost:8080/ipfs/$HASH/, so I open
Anyone who maintained a website in the old bad days of subdirectories can see what comes next: the homepage works, but clicking a link like
<li><a title='08-01-2017' href='/2017/08/01/recently.html'>Recently</a></li>
http://localhost:8080/2017/08/01/recently.html, and 404s.
And even if I used the base tag
to re-route that link so that it points
It wouldn’t work: the correct URL for that Recently post is, instead,
So IPFS content links are fully content-addressed. I suppose to make my site fully IPFS, I’d have to build each individual page and then construct a home page that linked to the generated hashes for those pages. That leaves an open question: how could two pages link to each other? Adding a link from one page to the other would change its hash, so wouldn’t it be impossible for pages to reference each other? This might be a lack-of-coffee problem on my part.
So, I published my site to IPFS and was able to visit it, but couldn’t figure out how links could work. Now I’d like to visit another website with IPFS.
For that, there’s NeoCities, Kyle Drake’s indie effort to create a GeoCities-like website host in the modern world. Kyle enabled IPFS for NeoCities, instantly enlarging the IPFS web by 140,000 websites.
From a NeoCities page about an awesome 8-bit Dragon Quest game, I go to archives, and then hack that into a URL that asks my local IPFS daemon for it: http://localhost:8080/ipfs/QmSjtF5z9XfGTjdGAM6e6dFR3LE8WH7Tnwwncs8xPou8Jb/
That works. But note that this way of visiting IPFS websites doesn’t follow the paranoid security standards of Beaker Browser - that request to that Dragon Quest game required jQuery over HTTP. If there’s a way for IPFS websites to be as secure than normal websites, I haven’t found it.
This is all alpha and beta software. Even the most optimistic observer would say that decentralized web tech has another 2-5 years until it might be mainstream. Protocol Labs is, like the dat team, focusing on low-level fundamentals first, rather than user experience testing.
From an end-user perspective, IPFS is significantly more intimidating and challenging than dat right now. You need to be comfortable with reading URLs, using the command-line, and digesting lots of new technical language. There’s little encapsulation, little abstraction on top of the protocol.
I expect a lot of that will change, but still wonder which terms and concepts will be exposed to eventual users. Will a Jane Everywoman say merkledag, and will immutable & content-addressable hashes become a user-facing feature?
The IPFS is useful here and now section of the homepage has me a little wary: is it really useful right now, with this flow? Trying to host a website with IPFS seems like a hack right now - the web security model isn’t there, the technical requirements for someone to browse an IPFS website are outside the reach of non-programmers.
Of course, all of this is going to change. FileCoin will likely make persistent web hosting much, much easier. Browser support for IPFS could be on the horizon. Documentation will hopefully improve. Protocol Labs’s papers and presentations are impressive, and they’ve garnered a lot of respect from adjacent communities. Using the tech as it stands today is fun, but not easy. It’s promising, but it’s hard to figure out what it’ll become.
Also, some of my experiences here may be simply mistakes or misunderstandings on my part. If so, absolutely let me know: my email’s on the /about page.