mako
@mako.dreamshrine.org
philosophy of information systems (applied)
https://aboutmako.makopool.com
Currently (at least on this account) focused on protocol and UX. Radical extensibility agenda.
alt for foodposts: https://bsky.app/profile/makoconstruct-food.bsky.social
https://aboutmako.makopool.com
Currently (at least on this account) focused on protocol and UX. Radical extensibility agenda.
alt for foodposts: https://bsky.app/profile/makoconstruct-food.bsky.social
github.com/bluesky-soci...
Passed comment on a potential `$extensions` field for atproto objects. I think it's necessary (though not sufficient) to escape the need for the mutation of lexicon types and actually decentralize the evolution of the protocol.
Passed comment on a potential `$extensions` field for atproto objects. I think it's necessary (though not sufficient) to escape the need for the mutation of lexicon types and actually decentralize the evolution of the protocol.
Extensions field in records for third-party usage · bluesky-social atproto · Discussion #1889
Is your feature request related to a problem? Please describe. There is friction in third-party clients wanting to add quality-of-life functionality that requires adding additional/custom public me...
github.com
July 14, 2025 at 12:54 AM
github.com/bluesky-soci...
Passed comment on a potential `$extensions` field for atproto objects. I think it's necessary (though not sufficient) to escape the need for the mutation of lexicon types and actually decentralize the evolution of the protocol.
Passed comment on a potential `$extensions` field for atproto objects. I think it's necessary (though not sufficient) to escape the need for the mutation of lexicon types and actually decentralize the evolution of the protocol.
Mutable globals are good actually as long as you can confine mutations to scopes (but we can't just yet because it's hard to reconcile that with async)
github.com/makoConstruc...
github.com/makoConstruc...
GitHub - makoConstruct/global_scoping: describes and to some extent implements the scoped override pattern
describes and to some extent implements the scoped override pattern - makoConstruct/global_scoping
github.com
June 22, 2025 at 12:53 AM
Mutable globals are good actually as long as you can confine mutations to scopes (but we can't just yet because it's hard to reconcile that with async)
github.com/makoConstruc...
github.com/makoConstruc...
Discovered a datastructure for upcastable/downcastable objects. Basically just a sorted array of (Type, k) pairs where k is the pre-order index of each segment/component of the inheritance heirarchy. Given this, upcasts and downcasts run in logarithmic time. github.com/makoConstruc...
github.com
June 20, 2025 at 5:13 AM
Discovered a datastructure for upcastable/downcastable objects. Basically just a sorted array of (Type, k) pairs where k is the pre-order index of each segment/component of the inheritance heirarchy. Given this, upcasts and downcasts run in logarithmic time. github.com/makoConstruc...
I don't believe in pure functional programming languages, and some pure functional programming languages seem to be coming around to agreeing with me: post: hackmd.io/@xZDxp0NgTQe...
June 8, 2025 at 9:22 PM
I don't believe in pure functional programming languages, and some pure functional programming languages seem to be coming around to agreeing with me: post: hackmd.io/@xZDxp0NgTQe...
(blog post:
hackmd.io/@xZDxp0NgTQe...)
Privacy-preserving real identity systems are totally possible, and it's disturbing that no one is talking about that given what's currently going on (age verification discourse, llm spam apocalypse, generally languishing democracies)
hackmd.io/@xZDxp0NgTQe...)
Privacy-preserving real identity systems are totally possible, and it's disturbing that no one is talking about that given what's currently going on (age verification discourse, llm spam apocalypse, generally languishing democracies)
June 6, 2025 at 2:05 AM
(blog post:
hackmd.io/@xZDxp0NgTQe...)
Privacy-preserving real identity systems are totally possible, and it's disturbing that no one is talking about that given what's currently going on (age verification discourse, llm spam apocalypse, generally languishing democracies)
hackmd.io/@xZDxp0NgTQe...)
Privacy-preserving real identity systems are totally possible, and it's disturbing that no one is talking about that given what's currently going on (age verification discourse, llm spam apocalypse, generally languishing democracies)
Wow, important? imo that most valuable cognitive work is *like language*, or language processing is deeply involved in most work, so I think this isn't just about language learning
www.youtube.com/watch?v=984r...
www.youtube.com/watch?v=984r...
The Language School That Teaches Adults like Babies
YouTube video by Matt vs Japan
www.youtube.com
May 7, 2025 at 11:45 PM
Wow, important? imo that most valuable cognitive work is *like language*, or language processing is deeply involved in most work, so I think this isn't just about language learning
www.youtube.com/watch?v=984r...
www.youtube.com/watch?v=984r...
Tim Sweeny (Unreal engine, Verse (logic programming language for a new web)) essentially advocating for the distributed type system I'm currently working on (that he could not have seen because the reference deserializer/interpreter isn't out yet)
May 3, 2025 at 7:11 PM
Tim Sweeny (Unreal engine, Verse (logic programming language for a new web)) essentially advocating for the distributed type system I'm currently working on (that he could not have seen because the reference deserializer/interpreter isn't out yet)
Taking a week or two to make a timer app with an extreme focus ergonomics (eg, support one-handed use, other things)
Galled that flutter/android never established a standard way of getting the screen size in mms,
Galled that flutter/android never established a standard way of getting the screen size in mms,
April 3, 2025 at 3:19 PM
Taking a week or two to make a timer app with an extreme focus ergonomics (eg, support one-handed use, other things)
Galled that flutter/android never established a standard way of getting the screen size in mms,
Galled that flutter/android never established a standard way of getting the screen size in mms,
#obsidian still doesn't fully get trees (see video comparison), which means it's not really good for reprioritizing todolist items (that have nesting), and reprioritizing todolists is a really important activity!
March 24, 2025 at 12:41 AM
#obsidian still doesn't fully get trees (see video comparison), which means it's not really good for reprioritizing todolist items (that have nesting), and reprioritizing todolists is a really important activity!
"Internet Computer" seems to be the current best in class? internetcomputer.org/how-it-works
infinite TPS globally (while any individual actor is 900tps), so it has that crucial but currently very rare quality that it wouldn't just become prohibitively expensive and choke if it saw global adoption.
infinite TPS globally (while any individual actor is 900tps), so it has that crucial but currently very rare quality that it wouldn't just become prohibitively expensive and choke if it saw global adoption.
How it works | Internet Computer
Learn how the Internet Computer blockchain realizes the vision of the World Computer. Dive into its technology, open-source repositories, in-depth video academy sessions, white papers, publications, a...
internetcomputer.org
February 25, 2025 at 5:52 AM
"Internet Computer" seems to be the current best in class? internetcomputer.org/how-it-works
infinite TPS globally (while any individual actor is 900tps), so it has that crucial but currently very rare quality that it wouldn't just become prohibitively expensive and choke if it saw global adoption.
infinite TPS globally (while any individual actor is 900tps), so it has that crucial but currently very rare quality that it wouldn't just become prohibitively expensive and choke if it saw global adoption.
Iroh's explanation of advanced bloom filters for set reconciliation was great www.youtube.com/watch?v=BIN2... www.youtube.com/watch?v=B943...
Apparently 40x faster than merkle stuff. I can now see why. But I look forward to forgetting why and just using someone else's implementation cause it's a lot
Apparently 40x faster than merkle stuff. I can now see why. But I look forward to forgetting why and just using someone else's implementation cause it's a lot
Invertible Bloom Filters - Part 2 of 3
YouTube video by number 0
www.youtube.com
January 25, 2025 at 9:42 PM
Iroh's explanation of advanced bloom filters for set reconciliation was great www.youtube.com/watch?v=BIN2... www.youtube.com/watch?v=B943...
Apparently 40x faster than merkle stuff. I can now see why. But I look forward to forgetting why and just using someone else's implementation cause it's a lot
Apparently 40x faster than merkle stuff. I can now see why. But I look forward to forgetting why and just using someone else's implementation cause it's a lot
Reposted by mako
New: Last Week in #bluesky
- freeourfeeds.com announces campaign for an independent governance organisation for ATProto
- Mark Cuban looking to fund a TikTok alternative on ATProto
- A deep dive explainer on Lexicons, and why they matter
Read at […]
- freeourfeeds.com announces campaign for an independent governance organisation for ATProto
- Mark Cuban looking to fund a TikTok alternative on ATProto
- A deep dive explainer on Lexicons, and why they matter
Read at […]
Original post on mastodon.social
mastodon.social
January 16, 2025 at 7:40 PM
New: Last Week in #bluesky
- freeourfeeds.com announces campaign for an independent governance organisation for ATProto
- Mark Cuban looking to fund a TikTok alternative on ATProto
- A deep dive explainer on Lexicons, and why they matter
Read at […]
- freeourfeeds.com announces campaign for an independent governance organisation for ATProto
- Mark Cuban looking to fund a TikTok alternative on ATProto
- A deep dive explainer on Lexicons, and why they matter
Read at […]
Every RPC protocol surface with promise pipelining is essentially a cloud vm for a weird programming language.
January 17, 2025 at 12:57 AM
Every RPC protocol surface with promise pipelining is essentially a cloud vm for a weird programming language.
Reposted by mako
Would you like to eat the other half of this giant pie thing I made in one sitting with me
December 31, 2024 at 4:51 AM
Would you like to eat the other half of this giant pie thing I made in one sitting with me
I've noticed a way of representing cyclic structures in content-addressed datastructures. Please talk me out of making it a load-bearing feature of the APC protocol. whtwnd.com/makoconstruc...
Self-referential/cyclic structure is both necessary and possible in content-addressed protocols, but... | mako
Content-addressing is good. It's a way of giving objects a unique global ID that anyone can verify on their own is the correct ID for that object. This can cut down on a lot of network requests and re...
whtwnd.com
December 7, 2024 at 7:10 AM
I've noticed a way of representing cyclic structures in content-addressed datastructures. Please talk me out of making it a load-bearing feature of the APC protocol. whtwnd.com/makoconstruc...
Reposted by mako
How Decentralized Is Bluesky Really? dustycloud.org/blog/how-dec...
A technical deep-dive, since people have been asking me for my thoughts. I'll expand a bit on some of the key points here in a thread. 🧵
A technical deep-dive, since people have been asking me for my thoughts. I'll expand a bit on some of the key points here in a thread. 🧵
How decentralized is Bluesky really? -- Dustycloud Brainstorms
dustycloud.org
November 22, 2024 at 4:06 PM
How Decentralized Is Bluesky Really? dustycloud.org/blog/how-dec...
A technical deep-dive, since people have been asking me for my thoughts. I'll expand a bit on some of the key points here in a thread. 🧵
A technical deep-dive, since people have been asking me for my thoughts. I'll expand a bit on some of the key points here in a thread. 🧵
Dear neuroscientists: whtwnd.com/makoconstruc...
Neuroscientists should definitely get into interpretability research! | mako
For a while we've been saying that neuroscientists should consider getting into artificial neural network (ANN) interpretability research; it's kind of important right now (it lets us understand why p...
whtwnd.com
November 24, 2024 at 10:37 PM
Dear neuroscientists: whtwnd.com/makoconstruc...
Hey @jimrutt.bsky.social ! I loved Network Wars, but I'm never sure whether to recommend it to people, because it's so addictive. Has there been there evidence of any skill transfer to other tasks? Like any game, it certainly feels like I'm learning something from it, but it's hard to nail it down.
November 22, 2024 at 7:44 PM
Hey @jimrutt.bsky.social ! I loved Network Wars, but I'm never sure whether to recommend it to people, because it's so addictive. Has there been there evidence of any skill transfer to other tasks? Like any game, it certainly feels like I'm learning something from it, but it's hard to nail it down.
Tell me something you think we should be able to do with social media systems that isn't really happening in any of the systems we currently have, and I will struggle to figure out why it isn't happening and then weave a solution into the stuff I'm building.
November 22, 2024 at 1:55 AM
Tell me something you think we should be able to do with social media systems that isn't really happening in any of the systems we currently have, and I will struggle to figure out why it isn't happening and then weave a solution into the stuff I'm building.
@shazow.net Why room-scale VR specifically? :]
November 12, 2024 at 3:48 AM
@shazow.net Why room-scale VR specifically? :]
Reposted by mako
very excited to share two new blog posts on peer-to-peer database primitives!
the first is a high-level introduction to "causal logs" - how and why replicating operation-based CRDTs means storing them in a partially-ordered dependency graph.
the first is a high-level introduction to "causal logs" - how and why replicating operation-based CRDTs means storing them in a partially-ordered dependency graph.
Introduction to Causal Logs | Joel Gustafson
joelgustafson.com
October 1, 2024 at 5:14 PM
very excited to share two new blog posts on peer-to-peer database primitives!
the first is a high-level introduction to "causal logs" - how and why replicating operation-based CRDTs means storing them in a partially-ordered dependency graph.
the first is a high-level introduction to "causal logs" - how and why replicating operation-based CRDTs means storing them in a partially-ordered dependency graph.
@faircompanies.bsky.social Hey, the contact form at faircompanies.com/contact/ is broken, and you also can't be DM'd here.
contact
Feel free to contact us:
faircompanies.com
October 9, 2024 at 9:36 PM
@faircompanies.bsky.social Hey, the contact form at faircompanies.com/contact/ is broken, and you also can't be DM'd here.
Are graph databases anything more than relational databases that include a table where two of the columns are external ids of the same type?
I feel like they *shouldn't be*? Like from what I can gather, "specialist" graph dbs never needed to be a thing,
I feel like they *shouldn't be*? Like from what I can gather, "specialist" graph dbs never needed to be a thing,
September 15, 2024 at 11:35 PM
Are graph databases anything more than relational databases that include a table where two of the columns are external ids of the same type?
I feel like they *shouldn't be*? Like from what I can gather, "specialist" graph dbs never needed to be a thing,
I feel like they *shouldn't be*? Like from what I can gather, "specialist" graph dbs never needed to be a thing,
*doing protocol research for the UX demo*
*has a bit of a cry about IPLD::dag-cbor only allowing keys to be strings*
... You know what? I don't accept this.
If you're going to introduce a special type for binary-encoding CIDs in values, you should also allow that type when CIDs are used as keys,
*has a bit of a cry about IPLD::dag-cbor only allowing keys to be strings*
... You know what? I don't accept this.
If you're going to introduce a special type for binary-encoding CIDs in values, you should also allow that type when CIDs are used as keys,
September 6, 2024 at 7:17 AM
*doing protocol research for the UX demo*
*has a bit of a cry about IPLD::dag-cbor only allowing keys to be strings*
... You know what? I don't accept this.
If you're going to introduce a special type for binary-encoding CIDs in values, you should also allow that type when CIDs are used as keys,
*has a bit of a cry about IPLD::dag-cbor only allowing keys to be strings*
... You know what? I don't accept this.
If you're going to introduce a special type for binary-encoding CIDs in values, you should also allow that type when CIDs are used as keys,
Never put a book or an article down without skim-reading it. If you're willing to put it down then you're willing to do a quick half-assed reading. The optimal amount of attention to pay to anything that made it to you is never zero.
August 15, 2024 at 3:18 AM
Never put a book or an article down without skim-reading it. If you're willing to put it down then you're willing to do a quick half-assed reading. The optimal amount of attention to pay to anything that made it to you is never zero.