Thing number 1 – Glitch!  It is like a mash up of geocities and heroku and git and animated gifs and unicorns.   I just learned about it. Apparently it used to be Fog Creek – which made debugging software.  I know about Fog Creek because I worked with a guy who was obsessed with Joel on Software – and rightly so – the guy has done some amazing things.

I then spent some time working in rust on cryptopals, which has replaced kundalini yoga as my meditative and spiritual practice and  then continued my research in borrowing and other things rustifarian.

Grin mining went live this week, or maybe last week. It’s a blur really, but I watch their github repo. Grin is written in rust and like monero is forked every 6 months to avoid ASIC mining.  It implements the MimbleWimble protocol.  Here is an article on how to mine grin, which I may implement if I spend more than an hour in the car this weekend.

Amy helped me with my fastai paperspace (a digital cloud for gpus)  issues namely

print(filename) cv2.imread(filename).astype(np.float32)/255

fastai claims filename is not an image. But it IS an image and it renders an image with other fastai tools. Amy thinks I have the wrong version of the fastai lib. So tonight I am trying to install this on the RC heap. But my ssh is not working because of IPv6 things.  I may try plaidml.

Amy and I reviewed how to use a model with a web app. I am really excited about this and plan to implement it once I finish training my model.  We then talked briefly about the decentralized web and web rings and scuttlebutt and mastodon . I mentioned blockstack. I have been working a bit with mastodon this week on my Haskell music transduction project so it was interesting to view it from the perspective of decentralization and open source (I was just using it because it was easy to create bots!)

Today Michelle and Terry demoed Kudos.  I think it is a really nice sample app for writing zulip bots. We deployed on now, which took a little bit of work but worked out fine.  In now 2.0 we can schedule tasks, but for now we use cron jobs (thanks Alan).  Michelle introduced me to the fanatic git tool – git flow,  and **/*.py (wildcard for this directory and children directories).

Alan walked me through some typescript, which I really appreciated, and it looks much less painful than using javascript.  Javascript offends my aesthetic sensibilities as I have mentioned on this blog before.

Finally,  here is some nyc housing data that someone shared on zulip. I am just going to stash it here because it may come in handy one day for a ML project. I am a digital hoarder.

View at Medium.com

Today was an internal RC Hackathon. Tenor and I were going to do some dev ops on the internal cluster, but that did not pan out. Instead I worked, sort of answering python questions, on a zulip chat bot spearheaded by Michelle and Terry.

At first we created a bot as part of the zulip bot architecture only to realize we did not have access to the zulip server. Then we moved over our app to use the rest api framework.  I wanted to deploy it with now, a sort of heroku for serverless apps, but the app was running as a statemachine.  Really what I need is iron.io. Do people still use that? I did some googling and could not figure out how to schedule now.  One bogus thing I could do would be to deploy the chat bot to now and then ping it with a cron job.

I made some head way building out the PrayerCoin contract and a contract for reifying non fungible tokens. These are up in github.  I am going to continue to clean them up.  I got integration working with haskell and mastodon. The more I work with Rust the less patience I have for Haskell, but something about the language seems very beautiful and I would like to stick with it. I had to do some pruning of my code base but that too is now up on github. I still need to write up my dapp ripple tutorial. But I feel like I am in a good groove with contract development in truffle and I have some ideas for creating automated dapp deployments.

There was a great discussion of consciousness in the consciousness reading group.  Why are we interested in the brain? What if consciousness is not located in the brain? DEEP THOUGHTS. Next week we are going to read about complex systems, mycelium, and the microbiome (my personal favorite for seat of consciousness) .

Now I am going to curl up with some green nettle tea and some James Hillman,

I had allocated 2 hours to write out cryptopals in Rust. Instead it took about 4 hours. I am including the hour I spent on a coffee chat and general mingling.  It was a long four hours – it seemed like 1000 hours. I struggled with Options and unwrapping and &Str and String.  I rewrote code to fix a bug to discover I still had the same bug and uglier code. But in the end it worked and I finally pushed it to my github repo.  I have a back log of problems to implement,  I hope they go faster.

Do you see my title –  that error stopped me for about an hour while I rewrote all my chars to be Strings, only to figure out I needed to make a variable mutable. OY!

In any case, I made the program interactive which was very helpful in debugging since I could pass in parts of a hex string rather than the entire string.  I was aided in my work by these trusty websites and Austin and Liz.

After I had a quick rust code review with some folks where I picked up a bunch of tips like using Chunks, and using bitwise operations for optimization.   Last week I was really impressed with a presentation on assembly and we looked at this fantastic tool call godbolt that translates code into assembly.  You can also do this on the command line: RUSTFLAGS=”–emit asm” cargo build and the cat (in my case) target/debug/deps/cryptopals-*.s. I think it would be interesting to see how the compiler handles multiplication vs bitwise operations.

Some links that fellow rc’ers put up today that I want to investigate later are: A Sol Lewitt drawing application that implements his ‘algorithms’, a conference for women and non-binary coders, rust debugging (I am still looking for a good debugger to integrate with neovim), and Curv -“a programming language for creating art using mathematics.”

Tomorrow is the RC hackathon. I will probably be hacking with Tenor on devops tools for a computer herre. I am sort of obsessed with getting to use an expensive gpu for machine learning.  Tenor wants to build an env for k8 deployment (I’m oversimplifying).  If we dont get access to said computer we will probably hack on a super secret zulip integration.

One first today! It was also the first time that I remember using git rm -rf –cached ., because I forgot to gitignore my target files.

View at Medium.com

I powered through the cryptozombies tutorials and it gave me a real sense for how to build NFTs and create interactions between NFTs.  But, the best part of the tutorial came from implementing ERC721x – NFTs that have limited quantities. I mentioned this in an earlier blog post but ERC721x are items that have multiples, like a sword in world of warcraft that has many copies in a game.

This is applicable to my prayer project since multiple people can have the same prayers -so there are multiple copies of the prayer.

It is also applicable to a supply chain idea I have for a while – refiedNFT- non fungeable tokens that you can turn into physical objects. This happen, if for example, I want to make my crypto kitty into 3D printed kitty, or crypto zombie into a real card, or my crypto tee shirt into a real tee shirt.  Looking at the ERC721x gave me a sense for what it would be to create my own EIP or ERC for this type of token and I have been going through the EIPs/ERCs submitted to the ethereum repo on github.

For a refiedNFT we would probably want to expand on ERC721x since we might want to create multiple editions, but to keep things simple we can start with an ERC721. To reify, we need a reification address (the manufacturers ethereum address), a physical shipping address – although this maybe we can encrypt on IPFS since it is a security issue, a description of what the physical item is, and confirmation of delivery (maybe a tracking address or the ethereum address of a shipper).

I imagine a function makePhysicalItem where the manufacturer offers to reify an NFT.  And then a reify function where someone orders the physical item.  I am still mulling this over.  Rather than only a _mint function there are 3 other funtions:  _make,  _order and _fulfilled.

In the meantime, I have settled on the basic structure of my prayer coin that I will post to github this week.  It will have a message (prayer), a uri (optional), strength, execution schedule, (execution place?), perhaps a notion of execution type, and then a mapping of users with number of prayers, mapping of prayers id to prayers, mapping of prayers to prayer instances (ERC721x), mapping of prayers to number of prayers.   My big issue with ERC721x is that it uses indexes to divide between types of NFT (multiples and singletons). Everything with an index under 1,000,0000 is a singleton, over is a multiple. I would rather keep separate mappings rather than use this convention.

I need logic for updating the strength of a prayer – this will depend on how many people share a prayer and the last time it was ‘prayed’.  I also need logic for updating the execution schedule.  This will depend upon the strength of the prayer.

Thinking about an execution type complicates things. What do you want your prayer to do?   Should I define it or should the prayer define it? Can the same prayer have different executions depending on who owns that particular prayer?  Maybe the prayer is written to the blockchain, or ethereum is donated to a particular charity (transferred to an address).  I will probably leave this out for the first iteration of the prayer coin  BUT  I can imagine an EIP for this to create a new token standard.  I will probably just write the prayer to the blockchain for this version or write the token id as Proof of Prayer!

But, what can you do with prayers like the eco system built around cryptokitties? Well, you could build out liturgies and rituals by chaining prayers.  Maybe build altars or memorials or intention experiments. I do not really know.   We can imagine that perhaps  many games started out as prayers or rituals – or a connection between games and ritual or prayer.  Is there a connection between pickup sticks and throwing the I-Ching?  I searched the internet and found this, perhaps bogus, but thought provoking article on St Teresa of Avila – the patron saint of chess. St Terese encouraged because the attributes necessary to become a good chess player are similar to the attributes necessary to become a spiritual person.  So maybe we can create a cryptokitties universe from cryptoprayers.

Sunday I spent most of the day in transit, so I used it to binge learn.  I powered through the last few tutorials of crypto zombies and through the first few lessons of fast.ai.

I highly recommend anyone interested in tokens and dapps to do the crypto zombies tutorial/game. It really filled in some gaps in my token knowledge – like when to create a modifier, how to implement the approve interface (if you saw my blog post a few days ago you will see my puzzlement around this interface method), and the secrets of Emit. Also there are some best practices for creating webdapps, that I had to figure out the hard way, so save yourself some heartache and go through the tutorial.

One of the main takeaways was how to create non fungible tokens that are not really unique.  This is under a specification called ERC732x (according to cryptozombies).  The example is something like game assets such as a sword or a magic rock. Many people might have the same game assets – such as the same sword.  But it is non fungible to a certain extent.  This is the same thing with prayers. We may have the same prayers, similar to the way we might all have the same sword in an RPG.

The biggest revelation came with my binge learning on fast.ai. It is blowing my mind. I did the Andrew Ng course on coursera a few years ago, but fast.ai is amazing. The teacher is very inspirational and it is all about learning tools and then coming up with new ideas to apply the tools to.   I finally understand what the different levels in a neural net do. Why you would want to freeze some and train others more.   The first example is to create a CNN for recognizing cats from dogs.  I replaced it to recognize mala beads from rosary beads.  I am hitting a few snags with the fast.ai library finding the path for my images but if I bite the bullet and use keras I think I could fix it.

But really I am interested in the relation between things not the differences! So step two – which I learned in lesson two of fast.ai is to separate out the different CNN layers to see what the interpretation is at each layer.  Then it is to use all manner of prayer tech, prayerwheels, tefillin, and maybe create an ai that creates new prayer tech – that would be kind of amazing.

But practically,I think it would be amazing to write an article around the tech behind the machine learning, and with the CNN visualizations this is really approachable.