The Tech Policy Greenhouse is an online symposium where experts tackle the most difficult policy challenges facing innovation and technology today. These are problems that don't have easy solutions, where every decision involves tradeoffs and unintended consequences, so we've gathered a wide variety of voices to help dissect existing policy proposals and better inform new ones.

Does An Internet Infrastructure Taxonomy Help Or Hurt?

from the might-create-more-problems dept

We've been running our Greenhouse discussion on content moderation at the infrastructure level for a bit now, and normally all of the posts for these discussions come from expert guest commentators. However, I'm going to add my voice to the collection here because there's one topic that I haven't seen covered, and which is important, because it comes up whenever I'm talking to people about content moderation at the infrastructure level: do we need a new taxonomy for internet infrastructure to better have this discussion?

The thinking here is that the traditional OSI model of the internet layers is somewhat outdated and not particularly relevant to discussions such as this one. Also, it's hellishly confusing as is easily demonstrated by this fun Google box of "people also ask" on a search on "internet layers."

Clearly, lots of people are confused.

Even just thinking about what counts as infrastructure can be confusing. One of my regular examples is Zoom, the video conferencing app that has become standard and required during the COVID pandemic: is that infrastructure? Is that edge? It has elements of both.

But the underlying concern in this entire discussion is that most of the debate around content moderation is about clear edge providers: the services that definitely touch the end users: Facebook, Twitter, YouTube, etc. And, as I noted in my opening piece, there is a real concern that because the debate focuses on those companies, and there appears to be tremendous appetite for policy making and regulating those edge providers, that any new regulations may not realize how they will also impact infrastructure providers, where the impact could be much more seismic.

Given all that, many people have suggested that a "new taxonomy" might be useful, to help "carve out" infrastructure services from any new regulations regarding moderation. It's not hard to understand a concept like "maybe this rule should apply to social media sites, but not to domain registrars" for example.

However, the dangers in building up such a taxonomy greatly outweigh any such benefits. First, as noted earlier, any new taxonomy is going to be fraught with difficult questions. It's not always clear what really is infrastructure these days. We've already discussed how financial intermediaries are, in effect infrastructure for the internet these days -- and that's a very different participant than the traditional OSI model of internet layers. Same with advertising firms. And I've already mentioned Zoom as a company that clearly has an edge component, but feels more like it should be considered infrastructure. Part of that is just the nature of how the internet works, in which some of the layers are merged. Marc Andreessen famously noted that software eats the world, but the internet itself is subsuming more and more traditional infrastructure as well -- and that creates complications.

On top of that, this is an extremely dynamic world. Part of the reason why the OSI model feels obsolete is because it is. Things change, and they can change fairly rapidly on the internet. So any taxonomy might be obsolete by the time it's created, and that's extremely dangerous if the plan is to use it for classifying services for the purpose of regulation.

The final concern with such a taxonomy is simply that it seems likely to encourage regulatory approaches in places where it's not clear if it's actually needed. If the intent of such a taxonomy is to help lawmakers write a law that only puts its focus on the edge players, that's unlikely how it will remain. Once such a mapping is in place, the temptation (instead) will simply be to create new rules for each layer of the new stack.

A new taxonomy may sound good as a first pass, but it will inevitably create more problems than it solves.

Techdirt and EFF are collaborating on this Techdirt Greenhouse discussion. On October 6th from 9am to noon PT, we'll have many of this series' authors discussing and debating their pieces in front of a live virtual audience (register to attend here).

Hide this

Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.

Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.

While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.

–The Techdirt Team

Filed Under: content moderation, infrastructure, regulations, taxonomy


Reader Comments

Subscribe: RSS

View by: Thread


  • icon
    Upstream (profile), 6 Oct 2021 @ 1:15pm

    Good discussion

    This is an issue that really needs to get more attention, particularly amidst all the calls for and against regulation, censorship (or the opposite: requiring content be allowed), anti-trust action, etc.

    Anywhere you attempt to draw a line in the Internet, the blurring and exceptions quickly become apparent. The heavy hand of government regulation has never been good at dealing with such nuances.

    And, as Mike correctly points out, with technology and the Internet, change happens very quickly. Government and it's regulations are famous for lagging far behind, even when it comes things that change much more slowly.

    We can never have an intelligent discussion about anything unless the terms involved are clearly defined, and those definitions are well-understood by everyone involved in the discussion.

    While a new taxonomy (which tends to focus on categorizing things into groups or layers) might not be in order, maybe some new terminology, clearly defined and broadly understood, might be what is needed.

    link to this | view in chronology ]

  • icon
    Ian Smith (profile), 6 Oct 2021 @ 4:50pm

    I build the Internet for a living and the main thing that helps me sort out the concepts is to separate who from where because they aren't fixed, and as technology and society move, the actor can and will start to show up in different places.

    • Application -- Applications are the software facilities supplying particular a function for people
    • Application Services --Services are the software facilities supplying a particular function for applications
    • Infrastructure Services -- Services are the software facilities supplying a particular function for infrastructure
    • Infrastructure -- Infrastructure is the basic facilities for compute, connectivity, and storage

    The actor that a piece of software does something and who it does that action for is (relatively) clear cut. In your example, Zoom is an application because it supplies a function to people as it's primary purpose, while DNS and BGP are infrastructure services because they help machines that connect together to form the Internet make those connections. But where an actor exists is less clear.

    • Customer Edge -- The final device, typically owned by the customer. Examples include: IoT device, smartphone, laptop, router, or smartTV.
    • Access/Network/Mobile Edge -- Cellular Radio towers and warehouses filled with computers operated by ISPs; the limit of the ISP's control, a location nearest the user. Examples include: 4G & 5G; ISP's firewalls, policy enforcement points; ISP's caches & video optimization servers, Network Address Translation; Netflix Open Connect Appliances; Industrial IoT gateways.
    • Service Edge -- The mid-point between the user and the application Examples include: CDNs; Cloudflare workers; WAFaaS; Auth-aaS; ad networks.
    • Data Center -- Warehouses filled with computers Examples include: The cloud; FAANGs & other hyper-scalers; DNS root servers; Equinix and other data hotel providers.
    • Core -- big cables and data centers that connect ISPs, Data centers, Countries, and Corporations together. Examples include: Internet Exchanges, Internet backhaul providers (Level3, etc.), BGP routing.

    Zoom the application, like most applications, has a part on the customer device and a part in one or more data centers. When the customer device part is a Web browser we tend to ignore that.

    What is important about the physical layout of the Internet, with respect to content moderation, is to separate Infrastructure and Infrastructure Services -- which start to take on utility-like characteristics as they mature and commoditize -- from Applications and Application Services. Facebook is an application because, as we just saw, you can turn it off and the Internet is fine; you go to Facebook (and any other Application). If you turn off Verizon, on the other hand, 100+ million people are off the Internet; you go though Verizon (or any other Infrastructure provider).

    That holds even when you start to talk about Edge Computing because putting an application at the Customer Edge or in a cell tower, or in a ISP data center doesn't change the fact that you still go to it, even when it is more local.

    link to this | view in chronology ]

  • identicon
    Anonymous Coward, 6 Oct 2021 @ 7:17pm

    I don't know if you need additional taxonomy, but that's a thought. Taxonomy is not limited to a singke stack of hierarchical layers.

    Or, on the other hand, just be specific. Humans can stop trying to categorize everything by cramming it into boxes in which it does not belong.

    Also: It doesn't matter what the consumer / end user thinks of as infrastructure for shorthand. (Just look at all the content moderation, public square, firat amendment bs). That they may feel like Zoom is just like POTS is irrelevant, and maybe that should be dissuaded instead of investing in taxonomy or additional layers.

    link to this | view in chronology ]

  • icon
    Upstream (profile), 11 Oct 2021 @ 7:04am

    Categories / Definitions

    At the risk of clouding or distracting from the point of the article, here is an analogy.

    There is a fundamental difference in companies that use the roads, like:
    UPS
    Fedex
    J. B. Hunt

    and companies that build and maintain the roads, like:
    APAC Inc
    Bob’s Barricades
    State DOTs

    even though all of them drive lots of trucks on the roads as an integral part of their operations.

    The distinctions that are fairly obvious between these two types of road-related organizations may not be quite so obvious when it comes to Internet companies and their operations, but they are the kinds of distinctions that need to be made in order to have fruitful discussions of the topic.

    link to this | view in chronology ]


Follow Techdirt
Essential Reading
Techdirt Insider Discord

The latest chatter on the Techdirt Insider Discord channel...

Loading...
Recent Stories

This site, like most other sites on the web, uses cookies. For more information, see our privacy policy. Got it
Close

Email This

This feature is only available to registered users. Register or sign in to use it.