Wisdom: Centralized Apps vs dApps Long-Term

dapps
chrisburniske

#1

2 great threads from Chris Burniske:


1/ Centralized apps building atop decentralized protocols may be the winning combination, each furthering the adoption of the other.

2/ As many have pointed out, “decentralize all the things” is a mistake. Decentralize the things that benefit from decentralization.

3/ Starting with centralized apps (think @coinbase @circlepay etc), if you’ve been paying attention, they are utilizing an increasing amount of decentralized protocols & open-source crypto infrastructure, to the benefit of their end users.

4/ The realization this week that @coinbase and @circlepay built their #stablecoin on top of @zeppelin_os was another tick in this trend.

5/ Centralized #crypto apps that build upon decentralized protocols have:

  • Global capabilities from day one
  • Lower costs due to the on-demand and perfect-competition nature of protocol services
  • More rapid innovation as they build upon others’ open source building blocks.

6/ Long term, centralized #crypto apps that build atop decentralized infrastructure will find it the key to what allows them to scale beyond, and outcompete, incumbents that fear public networks & cling to owning their infrastructure.

7/ Having covered cloud companies for 3 years & seeing how long it took big incumbents to embrace the public cloud (and some still haven’t), I have a hard time seeing these incumbents embrace decentralized protocols before it’s too late. Their loss.

8/ All of this leaves decentralized apps (#dapps) in hot water. As many have been pointing out, pure-play dapps are not getting much use.

9/ Connection points to end users need to be hyper-responsive, and due to the laws of physics, #dapps are fighting an uphill battle against centralized apps in response times.

10/ Projects like @graphprotocol or Picolo Network ((link: https://picolo.network) picolo.network) are aiming to help #dapps with their response times, but it remains TBD if it will ever be good enough for a human to not discern the lag.

11/ I also think customer service via #dapps is going to be a hard problem to solve, as customer service is high-touch and doesn’t come for free.

12/ Centralized apps in #crypto also currently have the vast majority of distribution to the end user, which so long as they continue to remain technologically relevant I have a hard time seeing them cede to #dapps.

13/ For these reasons, if you’re focusing on investing in [#cryptoassets]https://twitter.com/hashtag/cryptoassets?src=hashtag_click) (as @placeholdervc is), I recommend focusing on layer 1 protocols, and important middleware protocols atop the layer 1’s.

14/ Look at what protocols the centralized apps are building atop, and getting real utility out of.

Those will have fundamental value, regardless of how much some obsess over “economic abstraction” and “infinite velocity.”

15/ As always, I’m open to being proved wrong, but I feel reasonably confident that in the short-to-medium term it will be the centralized apps that continue to capture most all of the users — longer term, as people work to make dapps more responsive, we’ll see.


And then he expands on the above…:

1/ Going to expand on “centralized apps atop decentralized protocols” more, as the idea has incited healthy debate

  • I want to emphasize that the short-to-medium term architecture (as crypto bootstraps to the mainstream), will be very different from the end-game.

2/ To begin, I’m going to stop using the term “app” or “dapp” and instead use “interface,” as suggested by
@jmonegro
. The term “dapp” has been muddied by broader use of the term “app” within tech, incurring much confusion.

3/ In the protocol & interface world, there are an arbitrary number of layered protocols, which then have third-party interfaces that draw from these protocols to aggregate & candy-coat functionality for the end-user.

4/ When I say “centralized interface,” I’m referring to a service that hosts user-data, sometimes in a proprietary form, thereby conceding privacy for the sake of performance.

5/ A “decentralized interface” does not host the data directly, but is instead granted access to requisite user data from underyling protocols, always with user permission.

Self-sovereignty :slight_smile:

6/ While “self-sovereignty” is the end-game, we have to acknowledge that what we want as the early builders of crypto is bound to differ from what the majority will want — and the majority has yet to arrive.

7/ So we have to be ready for some of our strongly held ideologies to by diluted by the mainstream’s preference for convenience above all else.

The mainstream will compromise on nothing.

8/ The little bit of the mainstream that has trickled into crypto has already exhibited the tendency to follow convenience above all else.

9/ For example, how many warnings have people been given about the import of custodying their own private keys?

And yet, the vast majority of people have chosen a “centralized interface” to do it, because it’s easier.

10/ Given the immature state of decentralized protocols, it is hard (if not impossible) for any interface, be it centralized or decentralized, to rely wholly on these protocols to provide a reliable and responsive user-experience.

11/ Centralized interfaces (most often in the form of a company), get around this by only partially relying on decentralized protocols, and then filling-in the infrastructure gaps with centralized resources, to provide a cohesive UX.

12/ Decentralized interfaces have a hard time signing SLA’s with centralized infrastructure providers, not to mention their on-demand nature matches best with on-demand resource provision, making most of them rely on decentralized protocols entirely – that’s hard work right now.

13/ For decentralized interfaces, the Q becomes: When will decentralized protocols & tooling get so good that devs can quickly create and evolve these interfaces, and end-users sacrifice zero-convenience in choosing the “decentralized interface” over the centralized? 10+ years?

14/ In a recent conversation with @stevejang, I was reminded of how long it took for all the infrastructure of the web to be built, and how web applications meant to capture the end-user were serious endeavors to build in the 90s.

15/ “Decentralized interfaces” in crypto are in a similar stage as web app builders in the 90s. Some infrastructure and tools are there, but it requires a lot of filling in the gaps, often building missing infra components to get the interface off the ground if no-one else will.

16/ On the bright-side, experimentation around “decentralized interfaces” will become more and more prolific as the time costs to build them drop, and so we could see a relatively quick inflection (years from now) in “dapps that struggle” to “dapps that go viral.”

17/ In closing, I’m not trying to dissuade dapp developers.

You are doing important work that respects user’s rights more reliably than centralized interfaces, a big part of why many of us got into crypto in the first place.

18/ Long term, respecting user’s will be something that society demands, and citizens viscerally understand the importance of, but chances are they will still want soveignty + continued convenience.

That’s the end-game, but it’s not a 2-3 year thing, it’s a 2-3 decade thing.

via: https://twitter.com/cburniske/status/1058382995751690240


#2

this is pretty much our playbook for #yen.

and we’re in it for the long-haul… however long it takes.