Video Of The Week: From Agile To Immutable

Two weeks ago, my colleague Nick traveled to Hong Kong to attend a Blockstack event (Blockstack is a USV portfolio company) and deliver this talk, which covers some important questions/issues in the crypto sector.

This short tweetstorm sets up the video well, so I will start with that and follow with the video which is 17mins long.


Comments (Archived):

  1. William Mougayar

    I was there and presented on Decentralized Governance which ties in with Nick’s topic very well (video out next week).As Nick alludes to, the heart of the dilemma is that if you decentralize projects too early, what you gain in community support, you might lose in efficiency and speed to market. The trick is to progress properly, and not decentralize too early.The other factor is on-chain vs. off-chain governance, and that’s another topic in of itself, having pro’s and con’s on either side. Reality is that we are still learning about what the right balance is.

    1. awaldstein

      What does ‘properly’ mean in your context?If as you say it is too early to to know much of anything, your guidance is what then?This seems like the snake eating its tail from an advice perspective.Please clarify.

      1. William Mougayar

        Properly means there is a good balance between pace of progress and community/user/network effects support underneath it.What we don’t have enough experience in yet is the degree of balance between decentralization vs. centralization aspects. I cover some of it in my talk, and I’ve already pre-published the notes here: http://startupmanagement.or

  2. Kirsten Lambertsen

    What do all you cryptoheads think, then, about apps that are just clones of existing apps but are now decentralized? It doesn’t seem like that approach has been the answer to adoption (from my admittedly limited vantage point).

    1. William Mougayar

      That was one of the points in my talk. The incentives are not urgent enough yet to go to a decentralized Twitter, Facebook, messaging, storage, UBER, etc… The urge to move is not compelling enough yet except that your data is decentralized. But we are not valuing that feature too much yet.

      1. Kirsten Lambertsen

        I look forward to watching the video of your talk 🙂

      2. kenberger

        Yep, it’s like we are just at the release of the Ford model T and folks don’t see why they have to give up their horses yet 🙂

    2. Nick Grossman

      Ultimately, the big wins will be “something new”, I thinkHowever there will also be opportunities for apps that look like regular apps, but have better security properties because of some crypto/decentralized architecture. Apps built on blockstack are taking this approach:

      1. Kirsten Lambertsen

        I have BlockStack installed and check in on it once in a while. I need to go back and see what’s up 🙂 Given the events of the last two years, I’ve been thinking that political campaigns might be interested in using tools that are decentralized and on a blockchain…

        1. Nick Grossman

          there is a lot of activity on the blockstack platform right now

    3. Richard


  3. jason wright

    Do we need to start thinking like Vegans?… If the layered stack is the right architecture for the centralised network why would it also be the right architecture for the decentralised network?

    1. Michael Elling

      Because both are right! And both are wrong! What is needed is a framework and understanding that works across all physical and virtual networks that have existed since the dawn of mankind. Something that illustrates centralized-decentralized hierarchical networking (CDHNs) that we find in nature.

      1. jason wright

        is CDHN a thing?

  4. Andrew Cashion


  5. Muneeb Ali

    This was great talk! Made me think about two things:a) Decnetralization is not a Boolean (the current thinking of either you’re decentralized or you’re not) but rather a spectrum or a path. There are many trade offs at different points on the spectrum. Also, it might be reasonable to be in different states depending how far along are you on the path to adoption.b) Decoupling of user-interfaces from the underlying data/service. If I don’t like Evernote, I can delete it and start using a new notes app and my documents would just simply work. Users own the data (not the companies) and the apps come to you.

    1. sigmaalgebra

      For your second point, there is some old and relatively solid technology already worked out.So, from 20,000 feet up, what you want is a way to use a variety of software tools on your data. The tools would be developed independently, and there would both old tools and new ones, and all of them would be able to work with, create and/or process your data.So, what was done was some technology for object-oriented data definitions. They are not — IBM, sit down and listen up, I repeat NOT — object-oriented software.So, the definitions are object-oriented in the sense that there is an inheritance hierarchy. And to keep down name conflicts, there is a registration hierarchy.This work, done by the ISO or OSI or some such, and called CMIS/P for “common management information” and something or other. It’s close to ASN.1, abstract syntax notation #1. It’s old stuff, and I don’t remember all the details now.It’s related to what can be done with XML and related three letter acronyms.So, if your data is your collection of movie DVDs, then you pick some software and use it to enter your data. You or the software pick from the registration hierarchy some data object (format, structure) to use. For Blue-Ray DVDs, you use the object for that which inherits from the main movie DVD object.Then with your data entered, you can use lots of other programs to process your data. These other programs look up and download the definitions of the objects you are using, use standard syntax, etc. and do their work. If some piece of software will also help for CDs but won’t work with BluRay DVDs, then that software just does what is can with your data.I did something a little like that when I wrote software to schedule the fleet at FedEx: I wanted data on all the airports of interest, but realized that as the work went on I’d get more data, e.g,, time windows of operation, wanted to add such data to the data file, but wanted older programs still be able to run. So, I cooked up various tags, e.g., data types and data structures, and wrote the code to parse the file, look for the tags, and get the data.So, a lot of stuff like that has long been done. In particular, the intention of the CMIS/P stuff was to have all the people creating data, of some wide variety, growing over time, in agreement with all the people using the data for work of some wide variety, with functionality also growing over time.If that’s what you want, well, you should be able to get a lot already done for you.

  6. creative group

    CONTRIBUTORS:—CryptoKitties is a blockchain basedvirtual game developed by Axiom Zen that allows players to purchase,collect, breed and sell various types of virtual cats. It represents one of the earliest attempts to deploy blockchain technology forrecreational and leisurely purposes.SOURCE: WIKIPEDIA—-A virtual game for bored rich people. Just doesn’t make sense with a person thatvalues their time.Captain Obvious!#UNEQUIVOCALLYUNAPOLOGETICALLYINDEPENDENT

  7. creative group

    CONTRIBUTORS:it appears there is a push for the minds of pushing in one direction. Reminds us of the optical disc format standard war. (MMCD) (Became the DVD), backed by Philips and Sony, and the other was the Super Density (SD) disc, supported by Toshiba, Time Warner, Matsushita Electric, Hitachi, Mitsubishi Electric, Pioneer, Thomson, and JVC. The independent group Technical Working Group, or TWG was tasked with finding which format would best used for the consumer so the format war that took place between the VHS and Betamax wouldn’t be repeated.There needs to be an independent group to figure all this out to benefit the consumer that doesn’t include those with a monetary stake in the winner.Captain Obvious!#UNEQUIVOCALLYUNAPOLOGETICALLYINDEPENDENT

  8. Michael Elling

    The hidden elephant in the presentation is the hourglass framework of the internet; layers 3-4, not layers 5-7. It shouldn’t take 25 years from when IPv6 was first proposed to get to 40% penetration of the endpoints. In a world of rapid obsolescence we need a new model that is generative, sustainable and affords universal service.What is missing is a north-south (app to infrastucture) and east-west (network to network or core to edge, etc…) system of economic incentives and disincentives that better balances value with cost.The imagined conflict between centralized and decentralized eschews the fact that to clear marginal supply and demand within the above framework requires a constant tradeoff between layers and boundaries. For instance, layer 3 is nothing more than a tradeoff of layers 1 and 2. Think centralized-decentralized hierarchical networks (CDHNs).