Disqus - Latest Comments for windley

Re: Identity, Sovrin, and the Internet of Things

Phil Windley — Fri, 28 Jul 2017 17:24:33 -0000

Creating non-hierarchical certificate/trust structures doesn't preclude hierarchies of devices where that's desirable. Although I prefer random grouping ability as spelled out in this post on Communities of Things (called "societies" in the post): http://www.windley.com/arch...

Re: Identity, Sovrin, and the Internet of Things

Phil Windley — Fri, 28 Jul 2017 15:56:30 -0000

Wouldn't its response be an identifier? Not sure what you're driving at.

Re: Sovrin Use Cases: Authentication

Phil Windley — Wed, 12 Jul 2017 18:08:57 -0000

Yes, ironic indeed.

Re: Phil Windley's Technometria | Capture Mode and Emacs

Phil Windley — Sat, 17 Jun 2017 13:18:45 -0000

Its been a long time since I've done this. But you need to have the Quicksilver Terminal plugin installed. Then from Quicksilver, you can enter the text and select "Run in Terminal" from Quicksilver. Setting up a trigger for that is pretty straightforward after that.

Re: Using Picos for BYU's Priority Registration

Phil Windley — Fri, 31 Mar 2017 12:50:51 -0000

A couple of things. First, the numbers I report for the legacy system are for peak load, not maximum throughput.

Second, contrary to what you seem to believe, the system isn't run by a bunch of incompetent hacks. We have looked at it and it is contention at the persistence layer that's can't be easily solved. But like any other system, it could be fixed, but no one has deemed that the fix is worth the cost. Obviously, if this were the core system for a bank, those economies would be different and it would be changed.

Re: The New Pico Engine Is Ready for Use

Phil Windley — Fri, 31 Mar 2017 12:45:42 -0000

Your comment on the other page makes me think you're perhaps misunderstanding what this is. It's not an event bus, it's an actor-model programming system with built-in persistence.

At its current performance level there are obviously use cases for which it might not be applicable, but there are lots of use cases, even outside the classroom, where it has utility. For example, for most IoT applications of picos that I've explored these numbers wouldn't even be close to a problem. What's your use case?

There is still a lot of low hanging fruit for optimizing the engine. For example, the new pico engine isn't yet using a real persistence store and I expect that to help a lot.

Also, to be fair, the registration demo was a proof of concept, not a performance test. No one took any pains to find bottlenecks. Could be something as simple as how the HTTP calls were made and a bottleneck there.

Re: The New Pico Engine Is Ready for Use

Phil Windley — Thu, 30 Mar 2017 11:57:34 -0000

You can be impressed or not, I don't care. Given everything it was trying to do, it was a useful milestone for us.

Comparing it to a Node app is apples and oranges given what the pico engine is intended to do. Furthermore, these figures are with no optimization other than removing logging and there's considerable room for improving the persistence layer to increase performance.

So, whether you were impressed or not, I was happy to hit the milestone.

Re: Using Picos for BYU's Priority Registration

Phil Windley — Thu, 30 Mar 2017 11:51:03 -0000

The legacy system is over 20 years old with all that implies. The performance issue on registration basically comes down to disk contention, which might sound easy to solve, but is not as simple as it sounds for lots of reasons.

The pico system has not been optimized in any way, so yeah, considering what it's doing here, I was happy with the performance. The problem with the current implementation basically comes down to persistence layer contention as well, although for different reasons. This architecture allows that problem to be solved through the straightforward application of some basic scaling concepts.

Re: The CloudOS: Connecting Hue Lights and Insteon Controllers

Phil Windley — Mon, 20 Mar 2017 11:02:58 -0000

I agree this is an important point. We've been reworking the pico engine to run on Raspberry Pi's and similar processors in order to push as much processing to the edge as possible.

Re: Every Computer is Distributed

Phil Windley — Thu, 26 Jan 2017 22:22:16 -0000

Apparently, there's a thread on Reddit about the problems I've experienced: https://www.reddit.com/r/ap...

My fix (rebooting everything, including the monitor) only brings temporary relief.

Re: Every Computer is Distributed

Phil Windley — Thu, 26 Jan 2017 22:21:06 -0000

Yeah, an electricians are rarely trained to be computer techs.

Re: Book Review: Seeing Like a State

Phil Windley — Thu, 26 Jan 2017 18:48:22 -0000

I read this book several years ago and similarly enjoyed it. I was reading primarily from the perspective of how identity systems impact and nurture a centralized, administrative regime. A more identity-oriented treatment of these issues can be found in Documenting Individual Identity by Caplan and Torpey (eds). I trad Anti-Fragile some time after reading this, and had not connected them the way you did. Thanks.

Re: On Sovereignty

Phil Windley — Sun, 30 Oct 2016 18:53:41 -0000

Thanks Darrel! It hit me that way too when I finally got the dual nature of sovereignty and that its duality makes it even more powerful.

Re: Self-Sovereign Identity and the Legitimacy of Permissioned Ledgers

Phil Windley — Wed, 05 Oct 2016 11:22:38 -0000

While I can't predict how they'll react, our hope is that they will react just as you've outlined. That by giving people control, we lessen the regulatory burden on businesses and other organizations.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:20:46 -0000

In part by solving the "cheap pseudonym problem". The actual computation of the reputation is outside the scope of Sovrin. Sovrin merely records claims and proofs. Other systems will use them to generate reputation scores.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:19:11 -0000

It's not clear if stewards make money fro Sovrin or because or it. Stewards are likely going to be stewards because of the benefits it provides, not because there's a business model that pays them (as stewards). We're working on an economy of claims that includes premium claims that cost something to use with the proceeds being split among many parties.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:16:42 -0000

The details are yet to be defined, the Trust Framework working group is currently dealing with those questions.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:15:52 -0000

Yes, there is a need for semantic interchange. Lots of work to do in making that all work.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:14:28 -0000

I believe so but that probably depends on what part of the IAM infrastructure we're talking about. Sovrin doesn't do away with the need for organizations to have internal accounts and keep their own data. So to the extent existing IAM systems do that, they will co-exist and make use of each other.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:12:41 -0000

It can't since all that data is in the bank's systems. Sovrin prevents correlation between domains, it can't control what happens within a single domain.

Re: How Sovrin Works

Phil Windley — Wed, 05 Oct 2016 11:10:13 -0000

Out of scope. For now they simply memorialize an agreement. Sovrin, in general, makes relies appeals processes and legal arrangements that exist outside the ledger.

Re: An Internet for Identity

Phil Windley — Tue, 06 Sep 2016 21:41:11 -0000

OpenID failed because it was impossible to know what identities you could trust. There was no trust framework. An SIS needs to solve the trust problem to be viable. If it does that, there are plenty of businesses who will use it and sponsor users on the network.

Re: Why Companies Need Self-Sovereign Identity

Phil Windley — Wed, 18 May 2016 11:30:02 -0000

Yes. See Properties of Permissioned and Permissionless Blockchains (http://www.windley.com/arch.... Short answer: persmissioned blockchains. Quoting: "Permissioned blockchains have a governance process that selects validators and thus don't have to survive Sybil attacks since the validators are known--there are no pseudonymous validators."

Re: Properties of Permissioned and Permissionless Blockchains

Phil Windley — Thu, 21 Apr 2016 09:19:12 -0000

Thanks Daniel. I'll take a look.

Re: Properties of Permissioned and Permissionless Blockchains

Phil Windley — Sun, 17 Apr 2016 18:32:21 -0000

One additional feature of permissioned blockchains is they're faster (transactions per second) because they don't need to spend time on proof of work. Note that I didn't say they're less resource intensive. Most centralized transaction clearing systems (ledgers) are pretty expensive to run (think SWIFT, Visa, etc.). I don't know of evidence that proof-of-work schemes are necessarily resource hogs compared to other systems we have to clearing transactions in a ledger.