https://youtu.be/xv84BxQRn1MTau Chain Monthly Video Update for December is out!
Text Summary below:
Tomas
- IRC TML bot created, allowing multiple users to use TML in the chat, create databases, find intersections with other databases (consensus detection). It understands commands in TML itself, acting as a REPL which can simply be extended by UI or integrated into other software.
- Being the first real program written in TML it's a prototype so it delegates some activities to system scripts. Dragan Juan and Tomas will be moving functionality from system scripts to TML.
- Wish to make the bot updatable by itself users. This involves putting the source code into the database, so when the latter is altered the program changes. This process of collaborative code update allows us to create Alpha together.
Andrei
- Created prototype UI for TML/Alpha
https://idni.multilink.cloud/- Allows for TML only discussions, natural language English discussions, friend lists, etc.
- Automatically translates English statements in TML logic and uploads relevant subclasses. This functionality should improve in accuracy with time.
- This UI is tentative of course, a UI expert will eventually be hired.
- Opinion over opinions will also be supported at some point in the future.
- Also there's a sentiment analysis and entity detection bot in telegram/IRC that will help us understand the logic further.
Juan
- Worked on TML arithmetic support. Arithmetic function not only over universal symbols but specifically over subsets, which will make it more efficient. For example if we have two subsets {0 2} {1 3} and wish to apply the addition operator, (0+1, 0+3, 2+1, 2+3), it ends in a subset C {3 5 1}. This was implemented over BDDs in bitwise operations for addition and are very close to completing it for multiplication (over the next few days).
Fola
- Significant progress made on the legal side. Awaiting a public statement from lawyers regarding this, which will hopefully arrive pretty soon. This will reveal what's been done on the legal end over the past few months.
- Currently reading drafts of the whitepaper which will hopefully be ready for public release pretty soon
Ohad
- We can already imagine the power of Tomas' IRC bot being implemented in a UI such as what Andrei showed us. This would go a long way to making our technology more accessible to the masses. Of course, allowing natural languages will also make it much easier as users will not have to express themselves in formal logic.
- We're proud to have reached the milestone of working consensus detection in 2019 =)
- After Umar created a parser for first and second order logic, implemented transformation of formulas to Linux in conjunctive normal form. Umar's next task will be to working on context free parsing ability in TML, a field in which he's an expert.
- Dragan enhanced more of the builtin support we already made and is working on a fundamental change in TML: flexibility in changing bit order within BDDs. With dynamic bits, we can support types and everything will be more intuitive eg. When you say not something, you can associate it with something else. This feature will also help with performance, as without it right now, we use more bits than we really need, which means exponentially larger BDDs.
- Ohad himself continued researching second order logic and is considering the ability to use reflection. Reflection is an important concept in programming languages in general, but in TML to reflect the interpreter internal data would be to reflect the BDDs.
- Turing machines and many other models of computation such as Random Access Machine are infinite by definition and therefore not realizable as our physical computers don't have infinite memory.
Now if we take such a Turing or Random Access machine and limit it to a fixed amount of memory, then firstly intrigued listeners can think about why the halting problem for such machines is decidable. Secondly, running such a machine is a task in PSPACE and thus double directly in TML. Because Turing machine or Random Access machine or Automaton are basically only transition relations which are nothing more than 4 arity or 5 arity relations.
This means simulating a machine with constant memory can be done in TML right now. Once we have second order logic we can synthesize programs from requirements. But to simulate even an assembly language, as long as it has fixed memory, can be done in TML right now.
This has a huge implication: this means we can take any program in any language and compile it to TML, while it won't be equivalent to the original program, the point of equivalence will never be practically visible because it will be exactly when we run out of memory. By that we can compile OSes or even TML itself in this automatic way with which we can do something things by converting arbitrary machine code to TML.
- We've also been thinking about incentives or gaming mechanisms to encourage users to participate in the alpha and/or formalize knowledge so Ohad was playing with the idea of a point system incentivized debating platform, which is just a generalization of discussion platform as a discussion is just a debate with winners and losers.
