
California sunlight shines on decades of discovery. Gold glistens in our history: a magical rock that summoned pioneers from around the world.

This is the land where pioneers dug for nuggets, with shovels full of dirt and hearts full of hope. This is the land where Native Americas sustained vibrant cultures, and immigrants created homes. This is where empty pockets can become rich, and dreams can become neighbors with reality.

And now I’m here - halfway between the Hollywood Hills and Silicon Valley - staking a claim on the shores of self-discovery. At 4 AM, I rise and say hello to the darkness of hard things: a 1.2-mile swim, 56-mile bike, and 13.1-mile run.

| **First Swim in Morro Bay**                                     |
| :-------------------------------------------------------------- |
| ![Time Space Resoltuion Tradeoff](/blog/morroBayFirstSwim.webp) |
| _Moments after feeling the fifty four degree water._            |

This journey began in September, high above a rainy Pittsburgh river. My shaky hands grab a rope, and I inch across the branches of a great oak tree. I'm afraid to jump. Afraid to fall. I’m not ready, but I begin to count: three, two, one: and then I let go. I let go of the tree and _the false idea that we must be comfortable to do extraordinary things._

It would be months before I decided to race: Morro Bay was always a destination, not a challenge. I ran the Pittsburgh streets between a triangle of rivers: never comfortable, but never satisfied. I needed a challenge: something big, scary, hairy, and audacious.

This is it—the Morro Bay half ironman. There are 2500 contestants, but my only competitor is myself. I've never swum a mile; I've never biked 56 miles. My ankle is sprained. The water is freezing, 54 degrees. The race is long and I will be exhausted. But that's why I’m here: to discover gold beneath the dirt of difficult things.


Jett Hays

California Gold


![I'm laying on the grass. My image is split by light and split glass.](/blog/multiJettOnGrass.jpg)

> The Only Thing I Do Know Is That We Have To Be Kind. Please, Be Kind. Especially When We Don't Know What's Going On.

— Everything Everywhere All at Once

Bad choices can lead to good outcomes, and good choices can lead to bad outcomes. There is still uncertainty even when we are confident in our choices.

We are answered by chance when we knock on the door of choice. So what to do in the face of uncertainty? What to do in a world where intentions do not guarantee outcomes?

Act with kindness. Act with courage. There will be doubt, and there will be failures, but from the dark of night comes the light of day. After every winter, there is a spring.


Choice and Chance


Talking is not doing. Words are not work. Promises can be shared, but ultimately they must be kept.

Do more. Say less.


Do More, Say Less


![Denver dreams. Towl around my shoulders. Packed bag in the background.](/blog/denverDreams.jpeg)

I hear the whisper of mysterious things in the quiet hour before I dream.

I’m in a hotel in Denver. The Rockies are just beyond the city lights: cold and sharp, great big triangles that pierce the low-hanging western sky. Inside, I sit with a towel wrapped around my shoulders and my mind wrapped around a thought. The thought of Belize. The thought of leaving the country for the first time. The thought of sleeping on the beach and kayaking in the Caribbean.

These thoughts were once dreams, and tomorrow these thoughts will be a reality. I bought the ticket, and now I will take the ride: from America to Belize, from the comfort of home to a land of mystery.

Soon I will sleep, and then I will wake and live the life I have dreamed of. Goodnight from Denver.


Denver Dreams


Anyone is capable of running a marathon. 26.2 miles is a long way, but the human body is powerful beyond belief.

The hard part isn’t finishing a marathon; the hard part is starting a marathon. Getting to the starting line requires belief; not belief in God or the race organizers, but belief in yourself.

Religions around the world have perfected the art of belief.

Christians read the Bible.

Muslims practice Salah.

Buddhists meditate.

**Runners run the hard miles.**

The hard miles are unlike the easy miles.

Easy miles have been done before and can be done again. Easy miles feel good. Hard miles hurt.

Hard miles are an article of faith; when you run the hard miles, you increase your belief.

When you believe, anything is possible.

**Run the hard miles.**


Run the Hard Miles


**This article is a research paper, not a regular post. If you're interested in cryptography keep reading. Otherwise, I'll see you on the next edition of Jett Blu!**

Digital wallets enable essential applications like encrypted messaging and online banking. However, digital wallets are vulnerable to loss and theft. In this paper, we investigate current wallet standards. We show that multi-signature wallets are an insufficient replacement for classical hierarchical deterministic keychains. We address the usability concern of multi-signature wallets with threshold cryptography. SWORD improves the security of digital wallets by distributing account management across a group of shareholders. In addition to removing a single point of failure, SWORD simplifies the wallet experience by incorporating passwordless authentication. SWORD has been adopted and is used in the open-source Kryptik wallet.

## 1. Introduction

The internet relies on a peer-to-peer connection between computers, but the majority of services we use daily are centralized. While corporations like Chase and Facebook are easy to use, centralized convenience often comes at the cost of privacy and ownership.

Since the development of Bitcoin in 2009, there has been a significant push to make the internet more decentralized. Blockchains like Bitcoin and Ethereum promise to shift power from the few to the many, opening a new frontier of digital exploration.

A few examples of popular blockchain apps are shown below.

- **Augur.** A prediction market for betting on real world events like sports games and election outcomes.
- **Opensea.** A marketplace for digital collectibles.
- **Lens.** A user controlled social media platform. Developers can use a standard interface.
- **AAVE.** Lending and borrowing with over four billion dollars in assets.
- **SIWE.** One click login using an Ethereum account.
- **Uniswap.** A decentralized exchange for digital currencies.
- **IPFS.** Store files on a global network of computers.

Every blockchain application requires a digital wallet that can send and sign transactions. Since 2009, over one hundred million wallets have been created worldwide.

## 2. Wallets Are Hard to protect

One of the best parts of cryptocurrency wallets is the ability to control your own money and data. Unlike traditional banks or third-party services, no third party holds assets on your behalf. However, with great power comes great responsibility.

> Over $85,000,000,000 worth of cryptocurrency has been permanently lost after users have lost access to their wallets.[^1]

Today, holding your own cryptographic keys requires memorizing long mnemonic phrases, trusting third-party custodians, or purchasing complicated hardware wallets. These solutions are insufficient for most internet users looking for a simple way to manage private keys.

## 3. Standard Solutions

Noncustodial wallet management is an open problem, with an inherent tradeoff between security and ease of use. Standard solutions are shown below:

**Multisig Wallets.** A multi-signature wallet requires multiple private keys to authorize transactions.

**Smart Contract Wallets.** Smart contract wallets are just a multisig with extra bells and whistles for recovery and permissions. For example, Argent— a smart contract wallet with over four million accounts— allows ‘guardians’ to replace the account owner and set approvals for value-subtracting transactions.[^2]

**Account Abstraction.** Account abstraction turns every blockchain account into a smart contract wallet.[^3] Account abstraction has not been implemented on major blockchains like Ethereum despite discussion since 2014.

Smart contract wallets have been described as a breakthrough in the security and usability of crypto wallets; However, smart contracts have significant drawbacks, explained below:

- **Network Specific.** Smart contract wallets are implemented as code that runs on a single blockchain. Chain-specific contracts require users to create a separate wallet for each blockchain which is a hassle and a security risk.
- **Time Lag.** Adding guardians or transferring ownership requires a ‘security period’ that can last multiple days.
  No UTXO support. Smart contract wallets require Turing-complete blockchains, so there is no support for transaction-based networks like Bitcoin.
- **Variable fees.** Processing payments requires executing smart contract code, which can vary in cost for each wallet.
- **Upfront cost.** Smart contract wallets require an initial setup fee to pay for the creation of a new contract on the blockchain.

Large companies may benefit from the policies provided by smart contract wallets. However, smart contracts restrict wallets to a single blockchain and prevent users from accessing the benefits provided by a unique public-private key pair.

**Multi-party computation (MPC).** MPC wallets require multiple participants to authorize a transaction. Unlike multi-signature wallets, MPC wallets submit a single signature to the blockchain, reducing transaction fees and enabling interoperability with multiple blockchains.
However, multi-party computation protocols have high computation and communication requirements. The industry standard requires four rounds of online computation or seven rounds of pre-signing computation.[^4] In addition, most user-facing implementations like Coinbase dApp Wallet and Zengo allow a single computer to initiate a signing round.[^5]

> MPC wallets are best suited for institutions that can maintain multiple independent signers.

Multi-party computation also encourages walled gardens. Users are limited to the provider’s ecosystem without access to the wallet seed.

## 4. Wallet Basics

**Random Seeds.** At the core of every wallet is a single seed: a string of randomness. As a seed grows in length, the less likely it is to be guessed by an attacker.

Most wallets use seeds with 256 bits of entropy. In practice, this looks like a long string of ones and zeros, with each character having two distinct possibilities. The number of possible seeds grows exponentially. For example, if a seed has 8 bits of entropy, there are $2^8$ (256) possibilities of what it could be. With 256 bits of entropy, there are 2^256 possible seeds, which is more than the number of atoms in the observable universe.

| **Entropy As an Image**                                                                                                  |
| :----------------------------------------------------------------------------------------------------------------------- |
| ![Entropy as an Image](/blog/entropyAsAnImage.jpg)                                                                       |
| _Entropy is a measure of randomness. The more entropy increases, the less likely a piece of information can be guessed._ |

Seeds are hard to guess and are an ideal source for generating public-private key pairs.

**Deterministic Wallets.** Early Bitcoin clients used a random stash of keys once per transaction. Once the initial supply of keys was exhausted, wallets would invalidate the backup; this made key management complex and insecure.[^6]

Deterministic wallets simplify key management. A single seed can generate an unlimited number of keys. However, most deterministic wallets store keypairs on a single chain. Sharing a deterministic wallet is “all or nothing”: sharing a single key requires sharing all keys.

**Hierarchical Deterministic Wallets.** Hierarchical deterministic wallets are structured as a tree of key pairs. A single master key is stored at the root, and new child keys are derived from a unique derivation path. Each path of the HD tree can be shared without exposing the complete tree.

| **HD Tree**                                                               |
| :------------------------------------------------------------------------ |
| ![Entropy as an Image](/blog/hdtree.jpg)                                  |
| _A tree of HD keypairs. Each branch represents a unique derivation path._ |

> For cryptocurrency wallets, each blockchain has a unique derivation path. The derivation path contains directions to a key’s location within the tree structure.

HD wallets provide several benefits. For example, HD Wallets permit an online shop to accept payments at a new address for each order without giving the server access to the corresponding private keys (required for spending the received funds).

## 5. Secure Wallet Origin Distribution (SWORD)

To secure ownership and privacy across the web, we need a better way to protect asymmetric keys. Fortunately, threshold cryptography provides a neat solution.

> As a quick reminder, threshold cryptography protects information by splitting pieces of a secret between groups.[^7]

Instead of relying on a smart contract with trusted guardians, SWORD distributes shares of an encryption key to a group of shareholders; k of n shares are required to reconstruct the original secret.[^8] The encryption key creates an encrypted version of the wallet. Whenever a user wants to regain access to their wallet, the shares are reassembled, and the wallet is decrypted.

| **SWORD Shareholders**                                                                                                                                                               |
| :----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| ![KSWORD Shareholders](/blog/swordShareholderRing.jpg)                                                                                                                               |
| _Kryptik key management splits a secret encryption key among n shareholders. Each share is represented as a coordinate point. K shares are required to recover the original secret._ |

The Kryptik key management has a number of positive properties:

- **Network Agnostic.** Kryptik is network agnostic. Use any blockchain you want.
- **Quick recovery.** Account recovery takes seconds instead of days. Just log in like you would for any other app. Creating a new shareholder is also fast.
- **Multi-factor Authentication.** Verify ownership of your account with a combination of any standard authentication scheme like 2fa.
- **Predictable Fees.** Transaction verification logic is unmodified and the cost can be hard coded by applications.
- **No setup cost.** No smart contracts need to be created or paid for.

In short, Kryptik key management addresses the complexity, fees, and time delay of smart contract wallets. Furthermore, shareholders can be used for wallet recovery and authentication.

## 6. Challenges

While secret sharing improves the security of online wallets, there are several challenges. Each challenge presents tradeoffs between security and ease of use that must be addressed by Kryptik.

**Ciphertext Storage.** Ciphertext storage presents a tradeoff between security and convenience. By keeping ciphertext on the original device, shareholders function as gatekeepers, releasing shares when specific criteria have been met: for example, when a login link has been clicked, or security questions have been answered. Since the wallet owner is the only one with access to the ciphertext, shareholder collusion will not compromise the wallet. The secret encryption key may be recovered, but decrypting the wallet is impossible without the corresponding ciphertext (held in local storage).

> _Local Storage Drawback_
>
> Wallet recovery is no longer an option if the user's device is lost or stolen.

Cloud storage removes the user's device as a single point of failure. Anyone who has access to the remote database has access to the ciphertext. If the database is a blockchain like Ethereum, then the wallet ciphertext is public and viewable by anyone. In this case, shareholder collusion would compromise the wallet.

> _Cloud Storage Drawback_
>
> K of n shareholders can collude and decipher the public ciphertext.

Permissioned databases like AWS provide an alternative to public storage on a blockchain. Users can upload a copy of the ciphertext to a cloud provider like AWS and benefit from server-side access control. However, each shareholder would need access to the database to retain the benefits of cloud recovery. At this point, the security assumptions become identical to the public model.

| **Local Storage**                                          | **Cloud Storage**                                           |
| ---------------------------------------------------------- | ----------------------------------------------------------- |
| Keeps ciphertext on the same device that generates shares. | Keeps ciphertext on a remote database like AWS or Ethereum. |
| A single copy. Easy to lose.                               | Multiple copies. Hard to lose.                              |
| Recovery requires possession of the device.                | Recovery requires access to the database.                   |

**Denial of shares.** Shareholders may refuse to respond. K of N shares are required to reconstruct the original secret. If more than M-K shareholders refuse to respond, there is no way to reconstruct the original secret.

The wallet owner can keep k shares to avoid the ‘denial of shares’ attack, but this creates a single point of failure. Shares can be stored on separate devices, but there is still the underlying concern of a single person owning k shares.

**False shares.** Malicious shareholders may cheat and submit incorrect shares. Cheating does not compromise wallet security, but it does impair recovery. A single incorrect share is enough to invalidate reconstruction.

In the case of a cheater, the recovery protocol can cycle through k of m submitted shares to obtain a valid subset. Recovery is possible as long as k of m submitted shares are honest.

## Kryptik Implementation Details

Distributed key management is a general tool for sharing secrets online. While several secret sharing implementations— such as SLIP 044— have been proposed, each implementation suffers from various drawbacks that limit usability. SWORD uses encryption and authentication to achieve a better balance of security and usability. Specific design decisions made by SWORD are discussed below.

**Optimal threshold.** SWORD requires two-of-three shares for wallet recovery. An external database manages one share; separate user devices store the remaining shares. The two-of-three threshold provides fault tolerance in case one of the shareholders becomes unresponsive.

| **Shareholder**       | **Purpose**                                                               | **Holds Ciphertext?** |
| --------------------- | ------------------------------------------------------------------------- | --------------------- |
| Primary User Device   | Distributes shares and initiates transactions.                            | Yes                   |
| Secondary User Device | Used for wallet recovery if the external database refuses to respond.     | Yes                   |
| External Database     | Authenticates share requests. Release share if the request is authorized. | No                    |

**Wallet Storage.** Local wallets are held in a software vault. Vaults are locked by default and can only be unlocked with a valid set of shares. The standard vault interface is shown below.

```typescript
interface VaultContents {
  // ciphertext of encrypted seedloop
  seedloopSerlializedCipher: string;
  // version: default is 0
  vaultVersion: number;
  // share of the encryption key
  localShare: string;
  // timestamps stored as...
  // number of milliseconds since the epoch
  lastUnlockTime: number;
  createdTime: number;
  // user id generated by wallet app
  uid: string;
  // if unlock correct, should read "valid"
  check: string;
}
```

**Share Generation.** New vaults are encrypted with a 256-bit encryption key, which is then split into m shares. Shares are created using shamir secret sharing configured with a finite field in $Gf(2^8)$ and 128 bits of padding. Each shareholder receives a single share.

**External Authentication.** An external authentication provider generates a one-time code, which is shared via email. After validation, the code is exchanged with the database for an access token, and the user session begins. No passwords are required.

| **Kryptik Authentication Flow**                                                                                                             |
| :------------------------------------------------------------------------------------------------------------------------------------------ |
| ![Kryptik Authentication Flow](/blog/authenticationFlowPlain.png)                                                                           |
| _The passwordless authentication flow. A one-time code is sent via email. After validation, the user can access the server’s secret share._ |

**Recovery.** When a new user session begins, SWORD retrieves the database share and combines it locally to recover the original encryption secret. The wallet is then decrypted and made available for use. This arrangement results in a strong security scheme, where an attacker must compromise two of three systems to obtain a user’s private seed.

> If the database is unresponsive, a share can be retrieved from the user’s secondary device via a QR code.

**Synchronization.** Initial share distribution requires an exchange of credentials between the user’s primary and secondary devices. The user’s primary device displays two QR codes: one code for the local shamir share and one code for the wallet ciphertext. Both codes are protected from prying eyes by a temporary encryption key only authenticated users can access.

After the QR codes have been scanned and decrypted, an identical vault (with a unique shamir share) is created on the secondary device.

## Conclusion

SWORD uses shamir secret sharing and symmetric encryption to improve wallet security. When implemented in software, SWORD provides users with a simple wallet experience that does not require passwords or smart contracts.

[^1]: Charoenwong, Ben and Bernardi, Mario, A Decade of Cryptocurrency ‘Hacks’: 2011 – 2021 (October 1, 2021). Available at SSRN: https://ssrn.com/abstract=3944435 or http://dx.doi.org/10.2139/ssrn.3944435
[^2]: Argent (2020, October 8). Guardian Approvals. Retrieved January 4, 2023, from https://support.argent.xyz/hc/en-us/articles/360008828238-Guardian-approvals-needed-
[^3]: Ethereum (2021, September 29). EIP-4337: Account Abstraction Using Alt Mempool. Ethereum.org. Retrieved October 7, 2022, from https://eips.ethereum.org/EIPS/eip-4337
[^4]: Canetti, R., Gennaro, R., Goldfeder, S., Makriyannis, N., & Peled, U. (2020). UC Non-Interactive, Proactive, Threshold ECDSA with Identifiable Aborts. In CCS 2020 - Proceedings of the 2020 ACM SIGSAC Conference on Computer and Communications Security (pp. 1769-1787). (Proceedings of the ACM Conference on Computer and Communications Security). Association for Computing Machinery. https://doi.org/10.1145/3372297.3423367
[^5]: ZenGo (2018, June 28). Multi-party ECDSA. Github. Retrieved January 11, 2023, from https://github.com/ZenGo-X/multi-party-ecdsa
[^6]: BitcoinWiki (2019, May 23). Deterministic Walet. Retrieved February 15, 2022, from https://en.bitcoinwiki.org/wiki/Deterministic_wallet
[^7]: Alfredo De Santis, Yvo Desmedt, Yair Frankel, Moti Yung: How to share a function securely. STOC 1994: 522-533
[^8]: Adi Shamir. 1979. How to share a secret. Commun. ACM 22, 11 (Nov. 1979), 612–613. https://doi.org/10.1145/359168.359176


SWORD


There are many paths through life: some easy and some hard. We all end up in the same place eventually, but what we do before we die is up to us and the habits we practice every day. Building strong habits is hard; change inertia. I overcame the resistance of getting started by breaking down habits and goals into 30-day challenges: a journey of a thousand miles begins with a single step.
 
 
### Memorize a Shuffled Deck of Playing Cards 🃏
**January**
 
Inspired by the book *Moonwalking with Einstein* I developed a Person-Action-Object (PAO) memorization system. For example, I represent the King of Hearts as Kevin Hart clapping his hands. By the end of January, I could memorize an entire deck of shuffled cards.
 
 
### Zero Added Sugar 🍜
**February**
 
For February, I walked through the sugar desert: 28 days, no added sugar. Instead of munching on processed foods, I was more intentional with my eating habits, relying on complete meals and whole foods. I also discovered that sugar is added to many foods, including peanut butter, pasta sauce, and pita chips.
 
![Zero Sugar Veggie Bowl](/blog/veggieBowl.jpg)
 
### Meditate Ten Minutes a Day 🧘🏼‍♂️
**March**
 
I meditated for ten days out of 31: with a 68% failure rate. If I were to do this challenge again, I would arrange a designated meditation space and gradually increase the time spent meditating.
 
### No Social Media ⚠️
**April**
 
Removing social media was relatively easy. By tuning out 'social noise', I could focus on the relationships that truly matter: the relationship with myself and those around me.
 
### Run a Marathon 🏃🏼
**May**
 
On May 1st, I ran 26.2 miles through the rainy streets of Pittsburgh. [Read this article](/blog/marathon) for additional insight into my marathon experience.
 
![Marathon Run](/blog/marathonBefore.jpg)
 
### Freestyle Rap For 90 Seconds Every Day 🎤
**June**
 
In the spring, I auditioned for Carnegie Mellon's premiere improv troupe. After being selected, I flopped during rehearsals. The only way to improve was to practice. Freestyle rapping was a fun and challenging way to improve my improv skills.
 
### Write a Joke Every Day 😜
**July**
 
I love to laugh. Writing a joke every day allowed me to find the funny threads woven throughout the fabric of life.
 
>The nondairy aisle and girls with nose rings go together like hummus and pita chips.
 
### Cap Phone Usage at 25 Minutes Per Day 📱
**August**
 
Phones are like a leaky faucet: before you know it, your attention has been drained and you're left standing in a flood of trivial pursuits. I was inconsistent with limiting my phone usage, but this challenge forced me to be more considerate when interacting with the magic rectangle.  
 
### Stretch For Ten Minutes Every Day ✨
**September**
 
Stretching was my favorite challenge; I didn't miss a day. My body felt loose at the end of every stretching session, and my mind relaxed.
 
### Write an Explainer Doc Every Day 💻
**October**
 
For most of 2022, I programmed a mountain of code. Writing helpful explainer docs helped others scale the technical mountain and contribute to my project (a digital wallet).
 
### Plank For Two Minutes Every Day 📏
**November**
 
For November, I planked for a minimum of two minutes every day. Finally, on November 31st, I completed the challenge with a tough but rewarding five-minute plank: my first ever!
 
### Write a Quote Every Day 🖋️
**December**
 
Most thoughts are worth a penny a piece. But there are some thoughts and quotes that are priceless. For December, I collected thirty-one quotes. Below each quote, I wrote a matching story from earlier in the day.
 
>"We make a living by what we get, but we make a life by what we give." - Churchill




Jett's 12 Challenges of 2022


The internet relies on a peer-to-peer connection between computers, but the majority of services we use daily are centralized. While corporations like Chase and Facebook are easy to use, centralized convenience often comes at the cost of privacy and ownership.

Since the development of Bitcoin in 2009, there has been a significant push to make the internet more decentralized. Blockchains like Bitcoin and Ethereum promise to shift power from the few to the many, opening a new frontier of digital exploration.

A few examples of popular blockchain apps are shown below.

- **Augur.** A prediction market for betting on real world events like sports games and election outcomes.
- **Opensea.** A marketplace for digital collectibles.
- **Lens.** A user controlled social media platform. Developers can use a standard interface.
- **AAVE.** Lending and borrowing with over four billion dollars in assets.
- **SIWE.** One click login using an Ethereum account.
- **Uniswap.** A decentralized exchange for digital currencies.
- **IPFS.** Store files on a global network of computers.

Every blockchain application requires a digital wallet that can send and sign transactions. Since 2009, over one hundred million wallets have been created worldwide.

> Digital wallets enable essential applications like encrypted messaging and online banking.

## Random Seeds

At the core of every wallet is a single seed: a string of randomness. As a seed grows in length, the less likely it is to be guessed by an attacker.

Most wallets use seeds with 256 bits of entropy. In practice, this looks like a long string of ones and zeros, with each character having two distinct possibilities. The number of possible seeds grows exponentially. For example, if a seed has 8 bits of entropy, there are $2^8$ (256) possibilities of what it could be. With 256 bits of entropy, there are 2^256 possible seeds, which is more than the number of atoms in the observable universe.

| **Entropy As an Image**                                                                                                  |
| :----------------------------------------------------------------------------------------------------------------------- |
| ![Entropy as an Image](/blog/entropyAsAnImage.jpg)                                                                       |
| _Entropy is a measure of randomness. The more entropy increases, the less likely a piece of information can be guessed._ |

Seeds are hard to guess and are an ideal source for generating public-private key pairs.

## Deterministic Wallets

Early Bitcoin clients used a random stash of keys once per transaction. Once the initial supply of keys was exhausted, the backup would be invalidated; this made key management complex and insecure.

Deterministic wallets simplify key management. A single seed can generate an unlimited number of keys. However, most deterministic wallets store keypairs on a single chain. Sharing a deterministic wallet is “all or nothing”: sharing a single key requires sharing all keys.

## Hierarchical Deterministic Wallets

Hierarchical deterministic wallets are structured as a tree of key pairs. A single master key is stored at the root, and fresh child keys are derived from a unique derivation path. Each path of the HD tree can be shared without exposing the complete tree.

| **HD Tree**                                                               |
| :------------------------------------------------------------------------ |
| ![Entropy as an Image](/blog/hdtree.jpg)                                  |
| _A tree of HD keypairs. Each branch represents a unique derivation path._ |

> For cryptocurrency wallets, each blockchain has a unique derivation path. The derivation path contains directions to a key’s location within the tree structure.

HD wallets provide several benefits. For example, HD Wallets permit an online shop to accept payments at a new address for each order without giving the server access to the corresponding private keys (required for spending the received funds).

## Wallets Are Hard to protect

One of the best parts of HD wallets is the ability to control your own money and data. Unlike traditional banks or third-party services, no third party holds assets on your behalf. However, with great power comes great responsibility.

> Over $100,000,000,000 worth of cryptocurrency has been permanently lost after users have lost access to their wallet.

Today, holding your own cryptographic keys requires memorizing long mnemonic phrases, trusting third-party custodians, or purchasing complicated hardware wallets. These solutions are insufficient for most internet users looking for a simple way to manage private keys.

## Standard Solutions

Noncustodial wallet management is an open problem, with an inherent tradeoff between security and ease of use. Standard solutions are shown below.

**Multisig Wallets.** A multi signature wallet requires multiple private keys to authorize transactions.

**Smart Contract Wallets.** Smart contract wallets are just a multisig with extra bells and whistles for recovery and permissions. For example, Argent— a smart contract wallet with over four million accounts— allows ‘guardians’ to replace the account owner and set permissions for value-subtracting transactions.

**Account Abstraction.** Account abstraction turns every blockchain account into a smart contract wallet. Account abstraction has not been implemented on major blockchains like Ethereum despite discussion since 2014.

Account abstraction and smart contract wallets have been described as a breakthrough in the security and usability of crypto wallets; However, smart contracts have significant drawbacks, explained below.

- **Network specific.** Smart contract wallets are implemented as code that runs on a single blockchain. This requires users to create a separate wallet for each blockchain which is a hassle and a security risk.
- **Time lag.** Adding guardians or transferring ownership requires a ‘security period’ that can last multiple days.
- **No UTXO support.** Smart contract wallets require turing-complete blockchains, so there is no support for transduction based networks like Bitcoin.
- **Variable fees.** Processing payments requires executing smart contract code, which can vary in cost for each wallet.
- **Upfront cost.** Smart contract wallets require an initial setup fee to pay for the creation of a new contract on the blockchain.

Large companies may benefit from the policies provided by smart contract wallets. However, smart contracts restrict wallets to a single blockchain and prevent users from accessing the full range of benefits provided by a personal public-private key pair.

## Distributed Key Management

To secure ownership and privacy across the web, we need a different way to back up asymmetric keys. Fortunately, threshold cryptography provides a neat solution. As a quick reminder, threshold cryptography protects information by splitting pieces of a secret between groups.

For the past year, I have been developing an open-source wallet (Kryptik) that uses encryption and secret sharing for authentication and account recovery.

**Kryptik Key Management.** Instead of relying on a smart contract with trusted guardians, Kryptik distributes shares of an encryption key to a group of shareholders; k of n shares are required to reconstruct the original secret. The encryption key creates an encrypted version of the wallet. Whenever a user wants to regain access to their wallet, the shares are reassembled, and the wallet is decrypted.

| **Kryptik Shareholders**                                                                                                                                                             |
| :----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| ![Kryptik Shareholders](/blog/kryptikShareholderRing.jpg)                                                                                                                            |
| _Kryptik key management splits a secret encryption key among n shareholders. Each share is represented as a coordinate point. K shares are required to recover the original secret._ |

The Kryptik key management has a number of positive properties:

- **Network Agnostic.** Kryptik is network agnostic. Use any blockchain you want.
- **Quick recovery.** Account recovery takes seconds instead of days. Just log in like you would for any other app. Creating a new shareholder is also fast.
- **Multi-factor Authentication.** Verify ownership of your account with a combination of any standard authentication scheme like 2fa.
- **Predictable Fees.** Transaction verification logic is unmodified and the cost can be hard coded by applications.
- **No setup cost.** No smart contracts need to be created or paid for.

In short, Kryptik key management addresses the complexity, fees, and time delay of smart contract wallets. Furthermore, shareholders can be used for wallet recovery and authentication.


Digital Wallets


As a statistics and machine learning major at Carnegie Mellon, I have taken many math classes ranging from linear algebra and 3D calculus to probability and number theory. However, one of the most powerful ideas I have learned at CMU requires nothing more than 5th grade math.

## The Problem

Consider this problem. Eleven scientists have access to a nuclear bomb. They wish to lock the bomb in a vault so that the vault can only be opened if six of the scientists are present. How many locks are required?

Each combination of six scientists requires a unique padlock. It can be shown that the minimal solution requires 462 locks and 252 keys.These numbers are impractical and grow exponentially as the number of scientists increases.

Instead of physical padlocks the scientists can use a secret code to lock the vault. But how should the secret code be protected?

The most secure solution would keep the code in a single well guarded location: a human brain, a computer, or a safe. However, this scheme is unreliable: a single misfortune (an untimely death, malfunction, or sabotage) would permanently lock the vault.

Copying the code and stashing it in multiple locations adds redundancy, but increases the risk of a security breach. Many copies offer many opportunities for the secret code to be stolen by attackers.

At its core the problem is this: how can we provide the redundancy of multiple backups without increasing the risk of a security breach? Our first attempts at a solution have all had significant drawbacks:

1. Giving each combination of scientists a padlock is too complex and doesn’t scale beyond a few stakeholders.
2. A single secret code is vulnerable to misfortune. If the code is forgotten or destroyed, the bomb becomes unaccessible.
3. Storing copies of the secret code increases the risk of a security breach.

## Threshold Schemes

Security— from nuclear vaults to online encryption— requires accessing and transferring data without revealing the original secret. Threshold schemes help protect information by splitting pieces of a secret— like the code to a nuclear vault— between a group. Recovering the original secret requires k of the n group members to recombine their shares of the secret.

> Threshold schemes have the following properties:
>
> 1. K or more shares can always always be combined to recover the original secret.
> 2. Knowledge of K-1 shares leaves the secret undetermined— all possible values are equally likely.

It turns out that polynomials are all we need to implement a threshold scheme.

## Polynomials 101

Polynomials are just lines that can be represented as an equation. The degree of a polynomial is the largest exponent. For example, $y=2x^2+10x+4$ is a polynomial of degree two.

| **Example Polynomial**                                                         |
| :----------------------------------------------------------------------------- |
| ![Example Polynomial](/blog/polynomial.png)                                    |
| _A third degree polynomial. Four points are required to reconstruct the line._ |

> In general, each term in a polynomial can be written as $cx^k$ where c is a constant, x is a variable, and k is the exponent.

The following polynomial properties are relevant for threshold schemes:

- Every polynomial line crosses the y-axis exactly once.
- Every x-value corresponds to exactly one point on a line.
- A polynomial of degree k can be reconstructed from k+1 points.

As a final example, consider the polynomial y= x+10. This equation represents a straight line that extends from negative infinity to positive infinity and crosses the y-axis at a height of ten. Since the largest exponent is one, we can recreate the polynomial from any two points along the line. The process of recreating a polynomial from a collection of points is called polynomial interpolation and is used as the foundation for shamir secret sharing.

## Shamir Secret Sharing

In 1979, Adi Shamir proposed a threshold scheme for sharing secrets.

Shamir secret sharing divides data D into n shares, so at least k shares are required to reconstruct D.

At a very high level, shamir secret sharing hides data in a crowd— where the crowd is the infinite amount of numbers along the y-axis. The location of the secret data is given by a polynomial of degree k-1. The polynomial is broken into n shares and distributed among a group. K shares can be used to recover the secret’s location along the y-axis.

To create a k of n threshold scheme, shamir secret sharing uses the following steps.

_Share Creation_

1. Convert data D into a number.
2. Create an equation for an arbitrary polynomial of degree k-1.
3. Randomly select n points (xi, yi) from the polynomial.
4. Distribute each share to members of the group.

_Secret Recovery_

1. Collect at least k points from the group.
2. Use lagrange interpolation to derive a k-1 polynomial.
3. Convert the y-intercept of the polynomial back into the original secret.

## Applications

Threshold schemes like shamir secret sharing are very useful when individuals with conflicting interests— like eleven scientists from different governments— must cooperate. In addition, shamir secret sharing gives the redundancy of copying a secret, while minimizing the risk of a security breach.

There are many uses of shamir secret sharing ranging from passwordless authentication to collaborative signatures. After learning about threshold schemes my sophomore year at CMU, I built a digital wallet that uses shamir secret sharing for authentication and account recovery. Over the past year, the wallet has received support from many foundations showing that sharing secrets online is a critical problem made easier by polynomials.


Secret Sharing: The Power of Polynomials


For the past fifty years, many of the smartest humans on Earth have been building the internet. So far, change has been rapid. Million dollar mainframes have become front-pocket mainstays, iphones are as common as toasters, and conversations that once took weeks and a postage stamp now take milliseconds and an internet connection. Distances are shrinking, opportunities are opening and none of it would be possible without asymmetric cryptography.

## History of a Secret

Humans have been keeping secrets long before the internet. In Rome, Julius Caesar shared secrets by shifting the alphabet three spaces. The ‘Caesar Cipher’ uses the same key to encrypt and decrypt messages and is a form of symmetric cryptography.

In the 1940's, the Germans used symmetric cryptography to share battle plans and war strategy. The [Enigma](https://bletchleypark.org.uk/our-story/enigma/) cipher was eventually broken by a global team of scientists and supercomputers that included Alan Turing and Polish ‘Bombas’. While the Nazi cipher system kept secrets for over two deades, it required daily key lists and top-secret code books. In addition, the Enigma relied on custom machinery that cost millions of dollars to manufacture. The Enigma may have been useful for a national bureaucracy run by the Nazis, but it was completely unusable for regular people.

**Enigma Requirements:**

- Custom Machinery
- Top Secret Codebooks
- Daily Key Lists
- National coordination
- Multiple secure channels

In 1974, Whitfield Diffie and Martin Hellman changed everything. Their new system— [asymmetric cryptography](https://ee.stanford.edu/~hellman/publications/24.pdf)— allowed public keys to be shared with anyone, eliminating the need for custom hardware and code books. With a bit of math and a dash of insight, Diffie and Hillman created a way for peers to share secrets without the burden of bureaucracy. Public key cryptography is now embedded within the fabric of the internet and used by billions people around the world.

## Public Crypto 101

Asymmetric cryptography uses public-private key pairs. The two primary functions of key pairs are encryption and digital signatures.

#### Encryption

Public keys can be shared and are used to encrypt messages, but only those who know the private key can decrypt the incoming message. Encryption helps preserve privacy and draws a strong boundary between what is personal and what is public.

#### Digital Signatures

Private keys can be used to create unique signatures that verify the authenticity of messages. Signatures prove ownership and can be validated by anyone who has access to the sender’s public key.

| Public Key        | Private Key      |
| ----------------- | ---------------- |
| Share with anyone | Keep to yourself |
| Encrypt Message   | Decrypt Message  |
| Verify Signature  | Create Signature |

When combined, public-private key pairs provide ownership and privacy: both of which are essential for interacting online.

## Applications

Public-private key pairs are woven into the fabric of the internet and for good reason: privacy and ownership make the internet a better place to protect value and share ideas. Here are a few examples of applications that use asymmetric cryptography.

- **HTTPS:** HTTPS uses public key encryption to send messages between websites and your browser. HTTPS is essential for sensitive activities like logging into email or your bank account.

- **Signal:** Messaging apps like [Signal](https://signal.org/) use public key encryption to ensure messages stay between you and your intended recipient.

- **Bitcoin:** Blockchains like [Bitcoin](https://bitcoin.org/bitcoin.pdf) use private keys to sign transactions. Public keys are used as payment ‘addresses’.

**Bottom Line:** If you use the internet, you use asymmetric cryptography.

## The Promise

With public-private keys, we can unlock a world where ownership and privacy are the default. Whether you’re nineteen in New York or ninety in New Delhi, you have cryptographic superpowers that were once exclusive to national bureaucracies.
Privacy and ownership in your hands. No third party handcuffs. No misplaced trust. Just verifiable ownership that you control.

## The Problem

Most of us have surrendered our right to online privacy and ownership. We’re back to a permissioned society that relies on third party gatekeepers:

- Google can read your messages, but it's still the most popular email system.

- Facebook has a horrendous privacy track record, but it’s still the most popular social network.

- Blockchains offer a peer-to-peer cash system, but the majority of transfers are between custodial exchanges.

The internet relies on a peer to peer connection between computers. But the majority of services we use everyday are centralized. We surrender our privacy and ownership to corporations, because it’s easy. Because companies like Google, Facebook, and Coinbase offer a tremendous amount of value for free.

As consumers, we have surrendered our keys and the gatekeepers became kings. Our content, our messages, and our value are controlled by someone else with different incentives. When we rely on gatekeepers to manage our online experience, we trust they won’t close the gate. We trust they won’t read our messages. We trust they won’t share our data. We trust, we trust, we trust.

## What if….

But what if we had…

- Private messages by default
- Portable social profiles that we control
- Global money transfers in seconds, for pennies

Asymmetric cryptography has given us the keys to an online kingdom: a world where we control what we share and what we own. But it’s up to us as builders to make this dream a reality. The status quo won’t change until everyone has a public-private key pair that can be used with ease.

If we don’t take action soon, then third party gates will become a cage and we won’t be able to get out, because we let someone else hold our asymmetric keys.


Asymmetric Cryptography


Some people like birthday cake and some people like birthday balloons. Most like both. I like neither. Birthday cake appeals to my mind, but the heavy cake and rich cream put my mouth on a treadmill that doesn't stop until the entire delicious slice is swallowed and sitting in my stomach. Inevitably, that slice gets lonely and wants a friend to visit. So naturally, I abide and allow another slice to slide down my throat for a party of two. And then… those guys get bored and want something more exciting like a corner piece to knock on the door of self-restraint, like a crumbly pile of bad intentions. And as for balloons? Ok, I lied. Birthday balloons are awesome. But more than balloons and much more than cake, I prefer birthday adventures.

When I turned eighteen, my friend and I camped in the forest and drove his 2002 RAV-4 [100 miles per hour](https://www.youtube.com/watch?v=qc7rLAHVfTY) down a lonely desert road while listening to Adele with our shirts off. When I turned nineteen, I [jumped out of a plane](https://www.youtube.com/watch?v=XlC0fNytLeI).

The thing about skydiving though, is you don't dive or fly. You fall. But I didn't know that when I pulled the hatch open and let my feet dangle over the green farms of Snohomish, Washington. I only knew I was thrilled. And then when I fell from the plane I was terrified. And then, as I soared past Mount Olympus and fell through the clouds, I was alive. My soul stirred and rattled against the cage of death, screaming:

“I don't want to die. I don't want to die. I went to wander and live, ponder big questions and struggle for answers. I want the best and I want the worst. I want all of my time. I want to live! I want to throw the conditions of humanity out and live free and on my own terms.
I want to explore who I am and who I am becoming and I want balloons and candles all the way to the bitter end. And when I get there— when you think you have me wrapped up in your sweet swaddle of death— I want to keep going. I want to trace the edges of infinity with the grooves of my mind. I want to laugh, smile and cry as I journey through time. I don't want to merely exist and be extinguished like a flickering candle. I want to live like a great fire burning through eternity. My passion for life prevents me from being boxed in by our society of right angles. Somebody else can be boxed up and sent to live a life of spiritual death. Not me.”

In the 340ish days since my last birthday adventure, the light of my existence has been split into a spectrum of experience. I've been to many places and felt many feelings. I snorkeled above a school of fish in the Pacific and raced friends into the Atlantic. I moved across the country and wandered the halls of higher education. I've hiked through the Rockies and backpacked along the cliffs of Hawaii. I’ve hamocked in the forest and hammocked on the beach. I’ve swam in the ocean at sunset and I’ve swam on rooftops in the winter. I ran through Central park at 9 AM and biked through New Mexico at midnight. I’ve slept on cold Manhattan floors, and shivered in windy tents. I've run through snow covered forests and rejoiced with my loving family. I’ve lived the life of dreams: a charmed life, a fortunate life.

Amid the fortunes, there have been failures— I've failed tests, and botched speeches, missed many shots and made many mistakes— but the joy of life remains. I have the body of a teenager, the freedom of an adult, and the soul of a child.

But as I have traveled from here to there— twelve Flights, twelve states— there is still distance between who I am and who I am becoming. In a loud world filled with noise and expectations, I have struggled to hear myself: to hear my own heart beat though the body of time.

So For my 'turning twenty' birthday adventure, I am running a marathon to better understand who I am. The race begins in a few hours and wiggles across bridges and through the cobbled streets of Pittsburgh. I have two big finals the next day, but this is the test I truly care about. When exhaustion removes ego and pain requires honesty, who will I become? What will I hear as I listen to my soul?

Very excited,

Jett Hays (4-30-22)


The First 26.2


> ###### Jett Hays
>
> ###### Human Computer Interaction Department
>
> ###### Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA
>
> ###### jehays@andrew.cmu.edu

## What

Here are a few metaphors to help you understand what Brainstorm is:

Brainstorm is a map that tells you how to get from neural signals to language.

Brainstorm is a set of instructions that tells you how to assemble language from thought.

Brainstorm is a brain to English translator.

Brainstorm abstracts what it does with a neat user interface that can run on any personal computer. The result of a Brainstorm session is a chunk of text (created with your brainwaves) that can be shared with the world through email. Brainstorm is also a development platform that allows researchers to rapidly test their own mind controlled keyboard ideas.

## Why

Effective communication systems— from Twitter to the telephone— reduce the friction between what we think and what we share. Existing systems like keyboards do a good job of this, but they are still rudimentary. Even the most advanced interfaces like touchscreens are equivalent to punching a bunch of different buttons. By interacting directly with the brain, mind controlled keyboards can expand the intellectual throughput of our society. Brainstorm represents a significant step in the long journey to typing with our minds.

## Definitions

For those unfamiliar with the technical terms, here are a few definitions:

EEG: A noninvasive tool to measure brain activity. A series of sensors measure changes in voltage along the brain’s surface. The image below shows Brainstorm’s sensor arrangement.

| **Brainstorm Sensor Positions**                                                               |
| :-------------------------------------------------------------------------------------------- |
| ![Brainstorm Sensor Positions](/brainstorm/ElectrodeLocations.png)                            |
| _An overhead view of the Brainstorm’s sensor positions. Each sensor produces a unique value._ |

Natural Language Processing: A branch of machine learning that makes predictions about text.

Frequency Band: A range of frequencies (how often something occurs), often used to describe complex time series data.

Features: Information that can be used to make a prediction.

Logistic Regression: An algorithm used by Brainstorm to classify EEG data. Each classification includes a ‘confidence’ score that indicates the likelihood of a given class.

## Challenge

EEG Data is notoriously noisy. Determining what someone is thinking [the signal] requires filtering out irrelevant information [noise]. This process follows a long and intricate sequence of steps that uses complex algorithms. However, as you’ll see, understanding how Brainstorm gets from no signal to noisy signal to a decision about what you’re thinking is easy once you consider the inputs, outputs, and purpose of each component.

| **Example Brainwave**                                                                                   |
| :------------------------------------------------------------------------------------------------------ |
| ![Example Brainwave](/brainstorm/signalBounds.png)                                                      |
| _An example brainwave. As you can see, the eeg reading goes up and down a lot with no obvious pattern._ |

## Solution

_“Any fool can know. The point is to understand.” — Albert Einstein_

Brainstorm uses a combination of techniques— including signal transformations, natural language processing, and machine learning— as it runs through the five step process shown below.

| **Brainstorm System Flow**                                                   |
| :--------------------------------------------------------------------------- |
| ![Brainstorm System Flow](/brainstorm/bstormFlow.png)                        |
| _The five main events Brainstorm initiates to translate brainwaves to text._ |

While you read descriptions of the following steps and their purpose, keep in mind the ultimate goal of Brainstorm: to bridge thought and language with a noninvasive brain computer interface. Every line of code and each facet of Brainstorm’s design accomplishes a part of this goal. When combined into one system, these parts become a complicated whole. But if you can hold the purpose of Brainstorm in one hand, while you hold technical details in the other, you can grasp how Brainstorm translates brainwaves to language.

### Trigger

> Key Point: Brainstorm initiates the SSVEP phenomenon with boxes flashing at specific frequencies.

If you were to guess what someone was thinking at a random point in time, how would you do it? There are a number of different tactics you could adopt, but each approach relies on gathering additional information. The information gathering process could use a number of different methods— analyzing prior writing for thought patterns, reading body language, etc.— but every method is dependent on behavior. For decades, this was how scientists gathered information about the brain: by using behavior as a proxy for thought. But the development of brain imaging technology has allowed us to study what happens between thinking and doing. However, the brain is incredibly complex and the tools we have developed to study it are still rudimentary.

Unless you’re willing to cut open the skull to insert a bundle of expensive sensors, there is a tradeoff between spatial and temporal resolution. This may sound complicated, but it really just means that the more certain we are of where something happened in the brain, the less certain we are of when something happened, and vice versa.

| **Time-Space Resolution Tradeoff**                                                                    |
| :---------------------------------------------------------------------------------------------------- |
| ![Time Space Resoltuion Tradeoff](/brainstorm/tradeoff.png)                                           |
| _A simplified illustration of the tradeoff between our knowledge of where and when something occurs._ |

### Response

> Key Point: The SSVEP phenomenon is when a flashing object induces neural activity at the same frequency. Brainstorm receives 8960 new pieces of information about this interaction between trigger and response every second through its 14 electrodes.

The SSVEP phenomenon is very simple: frequencies between 3.5 Hertz and 75 Hertz induce similar frequencies in the brain. Scientists like to use this phenomenon when studying neural activity, as SSVEP is consistent across people and resistant to artifacts.

An easy way to understand the SSVEP phenomenon is to think about a hall of mirrors. With the hall of mirrors analogy, the trigger [flashing boxes] is in the center of a room surrounded by mirrors. Every time the trigger fires, the mirrors bounce the signal back and forth through the room. As the signal bounces through the room, it becomes slightly distorted. Instead of a clear picture of the trigger, we are left with dozens of mirrors each containing a distinct portion of the original signal.

Instead of mirrors, Brainstorm uses 14 electrodes to ‘reflect’ the brain waves traveling through the room [a brain]. Each electrode produces five frequency bands that, when put together with machine learning, produce a clear picture of the trigger. This picture is assembled across time, with 8960 new data points produced every second. The filtered output from one electrode is shown below.

| **Sample Electrode Output**                                        |
| :----------------------------------------------------------------- |
| ![Sample Electrode Output](/brainstorm/output.png)                 |
| _The data received from one electrode across a recording session._ |

### Processing

> Key Point: Brainstorm uses logistic regression for SSVEP classification. Feature selection, binning, and majority voting increase the accuracy of each classification.

To begin, consider what Brainstorm’s prediction system is being asked to do: given 70 numbers, return the most likely frequency. Satisfying this request, requires Brainstorm to balance long term patterns with short term noise. Brainstorm uses the following techniques to achieve this balance.

Feature Selection
The brain is like a room, and neurons are like people. Groups of people chat and the room fills with noise. Joining a worthwhile conversation is hard, but if you visit enough groups you can gain a better understanding of which conversations are relevant and which are not.

With Brainstorm, groups of neurons are represented by electrodes. Each electrode has a different ‘conversation’, but only some of these conversations provide meaningful information about the SSVEP paradigm. The chart below shows the association between the stimulus and the strength of an electrode’s response, which provides an indication of which electrodes provide meaningful SSVEP information.

| **Electrode Response To Frequency Range**                                                                                                                       |
| :-------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| ![Response Features](/brainstorm/responseFeatures.png)                                                                                                          |
| _On screen frequency charted against theta frequency band strength. The red line is a simple regression, indicating correlation between stimulus and response._ |

As can be seen above, the original 14 electrodes can be reduced to six: AF3, F7, F3, F4, F8, and AF4. A quick glance at the electrode location chart verifies this reduction, as these electrodes are clustered near the visual cortex. By using experimental data to filter meaningful ‘conversations’ from noise, Brainstorm is able to feed logistic regression higher quality features which results in higher quality predictions.

### Binning

Just because these six electrodes normally provide meaningful SSVEP information, doesn’t mean they always do. The data stream from an electrode can be corrupted by a number of external factors including movement, sensor impedance, and signal transmission. Brainstorm accounts for this noise, by calculating the following frequency power bands over two second bins of neural activity.

Brainstorm Frequency Bands (Hz) — Calculated over 2 second bins and updated twice per second

Theta: 4-8

Alpha: 8-12

Beta Low: 12-18

Beta High: 18-25

Gamma: Greater than 25

By using binned data to calculate frequency bands, Brainstorm boosts each electrode’s signal-to-noise ratio; this improves SSVEP classification performance.

### Majority Voting

Brainstorm collects predictions over time. For a final frequency prediction to be made, the candidate frequency must have received a simple majority of votes. By using majority voting, Brainstorm minimizes the influence of any single prediction which may be corrupted by noise (despite the noise controls detailed above).

### Output

> Key Point: Brainstorm increases typing speed by providing users with autocomplete suggestions.

After Brainstorm processes EEG data, the predicted frequency is matched to flashing text on the screen. This process is repeated, prediction after prediction, with the final result being a chunk of text. For most virtual keyboard systems, this would be enough. Having added evidence to their thesis, the researchers behind these systems would close their program and call it a day. However, Brainstorm is not an intellectual exercise; it’s a practical prototype: a significant step in the long journey to redefine the relationship between humans and computers.

Any system that redefines this relationship needs to be accurate, consistent, and fast. Signal processing and classification take care of accuracy and consistency, while Brainstorm leverages the NLP technology behind Google to boost words per minute, a proxy for typing speed.

| **Autocomplete Example**                                                    |
| :-------------------------------------------------------------------------- |
| ![Autocomplete Example](/brainstorm/nlp.png)                                |
| _From top to bottom: likely fragments, likely letters and general letters._ |

Everytime a character is added to the output, Brainstorm queries Google for autocomplete suggestions which are then parsed and displayed to the user. By allowing users to select from high quality suggestions alongside characters, Brainstorm reduces user burden which in turn reduces system burden. The more information Brainstorm can gain from natural language processing, the less information Brainstorm has to extract from brain waves. This leads to an increase in “typing” speed, with Brainstorm achieving an average of 40 characters per minute.

| **Trigger to Output**                                   |
| :------------------------------------------------------ |
| ![Trigger to Output](/brainstorm/flow.png)              |
| _The flow from frequency box to brainwave to language._ |

### Action

When the user is done typing, the text can be sent to anyone anywhere via email. By connecting the user’s mind with an established communication platform like Gmail, Brainstorm represents a viable prototype that minimizes the gap between what we think and what we share.

## Key Stats

**Speed:** 40 characters per minute

**Code:** One open source codebase

**Data:** An open source data set containing over 50,000,000 labeled data points

**Interaction:** The world’s first recorded email sent with a mind controlled keyboard

## Future

_“A journey of a thousand miles must begin with a single step.” — Lao Tzu_

The brain is the final frontier. Having conquered Earth, we now turn to conquering ourselves. But as scientists have come to realize, understanding how we think is incredibly difficult. Brainstorm represents a significant step in the long journey to typing with our minds, but there is a lot of room for improvement. Here are a few opportunities for developers looking to extend Brainstorm’s capabilities:

- Increase classification accuracy by swapping logistic regression for neural networks

- Adopt a non-SSVEP paradigm that reduces Brainstorm’s dependence on external stimuli

- Stress test the software and provide additional error handling where needed

- Add support for EEG systems outside the EMOTIV ecosystem


Brainstorm System Overview


> ###### Jett Hays
>
> ###### Human Computer Interaction Department
>
> ###### Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA
>
> ###### jehays@andrew.cmu.edu

**Languages:** C# and Python

**Application:** Windows form

**Repository:** https://github.com/jettblu/brainStorm

**Architecture:** Model View Controller (Event Driven Data Feed)

## Is this for you?

This paper is for developers trying to customize Brainstorm or for anyone trying to understand how a complex system can be constructed from simple principles.

## Philosophy

Note: Software is the general architecture of Brainstorm’s programs, while code is the manifestation of that architecture.

With a prototype like Brainstorm, it is impossible to separate great research from great software design. While Brainstorm is still a work-in-progress, it works and represents progress, because it follows the design principles below.

## Design Principles

1. **User first, Code Second.** Brainstorm is a machine whose value stems from replacing user burden with technical solutions. The mechanics of how Brainstorm transforms brainwaves into language are unimportant for the user. All the user cares about is if Brainstorm works and if it works well. The software behind Brainstorm ensures both ifs are met with consistent performance and clear error handling

2. **Make Assumptions Few and Weak.** Strong assumptions = weak programs. Weak assumptions = strong programs. As a platform for both users and developers, Brainstorm tries to minimize the scope and strength of its assumptions. This allows for a better user experience and enables developers to customize algorithms with ease.

3. **Easy to Understand Code > Slightly More Efficient Code.** Don’t make simple things complex. Variable names are descriptive and reusable computations are relegated to helper functions. Breaking a complex chain of logic into understandable steps makes Brainstorm easier to update and maintain.

4. **Leave a Trail.** Each section of code has in line documentation that details what and how it’s doing. By leaving a trail through complex code structures, Brainstorm allows developers to locate and understand relevant algorithms.

## Implementation

Brainstorm is woven together by the robust seams of Model View Controller architecture. MVC architecture allows for an iterative testing process in which various aspects of the software can be evaluated and subsequently modified based on performance without demolishing the architecture as a whole. By separating concerns, Brainstorm provides developers with a versatile system that can be adapted to fit future virtual keyboard research. An event driven data feed supplies the model with a continuous stream of information that is used to update the view and change system state.


Brainstorm Design Principles


We’re all on loan. Our time is borrowed against fate. The chances of us existing are slim, the chances of us continuing to exist are slimmer still. It is against this fortunately miserable backdrop that we plot the course of our lives, our hand guided by time and expectation. The needle of existence dives in and out of experience, weaving the story of us. Of whether this story will end, there is no question; of whether this story can be enjoyed, there are entire philosophies.

We’re told what to do and when, but the reasoning is always circular, the houses are always square, and the shooting is always straight. This is suburbia—a land of right-angled duplicates, where cookie cutter families fill cookie cutter homes, living master planned lives in master planned communities. In theory, we can do whatever we want; in practice, we do what is convenient, constructing our
future with respect to the suburban blueprint. We hide our personal beliefs behind gates and hedges and our dreams behind tuition and diplomas. We do so because it’s safe, expected, and all part of the master plan.

My life, as a white American male living fifteen minutes from the beach, is not a hard one. The way forward is easy: tough decisions have been replaced by schedules and dreams have been replaced by goals. College was never negotiable; my dreams of skipping freshman year to bike across the country took a right turn at orthodoxy and never came back.

So now I stand at the corner of my youth, soon to be swept away by the marching mass of suits and 401ks. This corner, though, is expensive. Carnegie Mellon charges $72,000 for it and my aspirations are constantly taxed by the weight of practicality. Why stay? Because it’s easy. But I don’t think I will stay. Screw convenience and screw the plans. Life is too beautiful and our time too short to live cookie cutter lives in cookie cutter homes. The suburban dream may be America’s, but it’s not mine.

![Suburbia](/blog/suburbia.png)


Suburban Blueprint


**∞ - -9 months:** We float around in space, forever 5.5 inches away from being a thing.

**2:** Tis’ a simple, befuddling existence: we throw more tantrums than we throw parties and cry more than we think.

**13:** We’ll never know more than we think we do when we’re 13.

**18:** We’re free to do as we choose, so long as we choose to go to college. We have a diploma, but no driver’s license—or maybe, that’s just me. 58.1% of us vote.

**21:** We marinate our minds in college and alcohol.

**25:** The majority have sex; the minority have kids. A lot of people do a lot of things because a lot of people expect them to.

**30:** We put our heads down and work towards a day when we don’t have to.

**44:** We’re a long way from the womb. Hopefully, we’re just as far from the tomb.

**50:** It’s our 32nd class reunion: we wonder where the time went and are shocked by how some of our classmates are already chillin’ in coffins.

**67:** We think about when we were 30 and yearning to be 67 so we could reap the benefits of social security and think about when we were 30.

**72:** We’re probably grandparents now.

**78:** There’s a lot of staring at our grandchildren and misplacing our dentures.

**86:** We’re officially out of date, but we don’t care: they love us anyways. Hopefully.

**90:** Just when we think we’ve overstayed our welcome, we’re thrown a raging birthday party. Or maybe it’s a ‘you can go now’ party, but we don’t care: we’re just having a good time.

**97:** If we haven’t died by now, there’s a good chance that this is the year.

**Death - ∞:** We float around in space no longer 5.5 inches away from being a thing. We already got lucky once; who’s to say we get lucky twice?


What We Do and When We Do It


Let’s face it, some people are farther from the tree than others.

![The Knowledge Tree](/blog/knowledgeTree.png)

**FAR FROM THE TREE**

_Adjective_

1. Highly evolved

2. Original/ innovative

_Origin_

The mind of Jett Hays while in A.P. Gov.
