Not Before Time is a public broadcast facility built on existing technology.

Not Before Time guarantees that information:

  1. will not be readable before a certain future time (no early reading)
  2. was not sealed before a certain past date and time (no earlier sealing)
  3. was not electronically signed before a certain time (no late denials)

These guarantees support human rights, democracy and business.

Novel features of Not Before Time are:

  1. Time locked information is the opposite of the trend for all information to be available instantly, everywhere.
  2. The three guarantees are supplied to ordinary users using their everyday software on their existing devices.
  3. The mechanism and its novel new applications are easily explained to non-technical people, perhaps including you.

Much of NBT is already running in a demonstration called quicknet.

Picture of an old safe with a mechanical time lock

Introduction and overview

Not Before Time provides a new certainty to the age-old question “What did they know, and when did they know it?”

“New fundamental certainty” is a big claim. Some historical examples are:

Reliable timekeeping. The first portable clocks were developed to help ships navigate. In the centuries since, all of industrialised society has become defined by time. In the 21st century we assume every individual has a right to access accurate time. Legally it is included in other fundamental human rights.

GPS. Since the 1990s, always-on knowledge of latitude and longitude has similarly redefined society. Exact location knowledge is assumed in everyday life. Access to GPS is a fundamental individual right protected within the European Union (GPS means “any of the satellite-based location systems”.)

Radiocarbon Dating. The certainty of C14 dating made history jump into focus. For the first time we knew for certain when key events happened and what order they happened in. Objects and people could be placed in their correct temporal context. Myths and legends could emerge from the unknown, and humans gained certainty about their past story.

Evolution. With the certain knowledge that evolution drives all life, we have begun decoding the petabytes of instructions, historical records and physical potential encoded in DNA. The new certainty is that humanity can in principle hope to decode life itself. Although in practice it’s an overwhelmingly vast task, the idea of decoding life has changed all of biosciences and health, and expectations in art and literature and beyond.

💡
and now we add NBT NBT provides another new kind of certainty: anyone can now connect physical or electronic information with time in very reliable ways. This document is a conversational approach to explain how the human needs met by NBT can become a scalable solution for commercial and community use. See the last section for references and prior art, although this article is not a formal paper.

NBT is a clock that works for information. Like GPS, NBT is a public broadcast technology that enables new creative opportunities for an unlimited number of users. Like GPS, NBT cannot be limited to people of particular views, nationality, employer or any other difference. NBT removes the ultimate choice about the publishing of information from internet and media companies, and gives it back to individuals. NBT is the opposite of chat with disappearing messages, instead, it is a constant stream of messages from the past appearing in the present.

Technically, NBT contains no surprises. NBT needs no new cryptography. The components are published and analysed.

NBT resists obsolescence. NBT is intended to be effective for decades and longer despite the fact that decryption techniques get faster and better over time. The NBT design makes it no harder to use state of the art of quantum-resistant encryption methods than tried-and-true algorithms from a decade+ ago.

NBT has immediate and scalable applicability to finance, unlike cryptocurrencies. NBT has obvious application to financial transactions including cryptocurrencies. Guarantees of the contents of a time-encrypted finance document have many wins and require standalone analysis.

Example NBT use cases

  • Embargo. Secretive Ltd, wishes to make a massive announcement worldwide. An embargoed press release secured by NBT is circulated to press in all time zones, behind all manner of firewalls and on personal devices. Exposure of the announcement is maximised because it is already in everyone’s hands, and yet the embargo can’t be broken. At the time the embargo expires, the announcement becomes readable everywhere.

  • Sealed bids. Smart Ltd wishes to build a new office. Commercially sensitive bids are received from many construction companies, with NBT being a requirement for all bid submissions, and bids closing on 17th July 2027. The bids cannot be read before the closing date by anyone, not even the construction companies that sent them. Nobody can accuse Smart Ltd of giving secret pricing information from one bidder to another.

  • Email Decrypted After it is Received. Using Google Email with a free plugin, an encrypted message is received with the subject “This email will be readable from midnight, 25th December 2027”, in three days’ time. Three days later, the email turns to plain text.

  • Whistleblowers. A whistleblower wishes to publish information but have time to escape. She uses NBT to secure a file which she distributes as an anonymous bittorrent with the filename NBT-Locked-Until-2027-03-01-Interesting-Facts-For-Journalists-Worldwide.pdf.

  • Future payments. A series of payments can be scheduled in the future by encrypting digital cash with dates falling due at the first of every month, when the recipient will be able to decrypt and use the money. This need not be cryptocurrency, it could equally be an encrypted bank cheque, share certificate or other asset title.

  • Proof of earliest possible sealing. A video is encrypted for release in 2050. The sealed package includes an NBT private key first released at 10am on 11th April 2019. When the package opens, anyone can see it was not sealed before that hour. They cannot tell when the video itself was made.

  • Physical proof of earliest possible sealing. A controversial paper book is published with a QR code on the front cover. That QR code contains a unique number published under NBT on 18th September 2020. NBT certifies the earliest date the associated package could have been sealed. This is a form of protection for authors as well as assisting in historical chains of evidence.

  • Unencrypted Email With Proof of Earliest Possible Sealing. A standard Yahoo Email inbox receives a message with the subject “The NBT-signed attachment to email contains meeting notes sent after 14th October 2021”. Despite the fact that nothing is known about the sender’s email system, and that email is notoriously easy to fake, NBT proves the attachment was not sealed before 14th October 2021. For example, a court of law must reject a claim that the notes were sent on the 10th October 2021, because there can be no doubt about the attachment signature.

  • Combined Sealing Proof and Time Locking. Using Google Email, a plain text message is received with the subject “The attached encrypted PDF will be readable from midnight, 25th December 2027, and the first page contains a unique number created on 1st January 2021.” NBT thus makes it certain that the PDF was sealed after January 2021 and will not be read earlier than December 2027. This has many practical uses.

Email as the first everyday use

One early next step could be emails. Currently almost nobody has motivation to encrypt emails, despite the security and privacy benefits. The main barrier seems to be the adoption argument: why bother installing encryption if nobody else uses it? This is very understandable. Security and privacy are not major motivators for most people, even technical people. Laws such as the EU GDPR and ePrivacy have not been exciting motivators for improving security, and so encrypted email remains niche.

However, emails that are only visible after a certain time… that is something most people can understand instantly. Thus NBT could be a reason for encrypted mail to become common. With NBT the users have no key management problem, and enough use cases that maybe it could become widespread.

Building NBT

All of the components required to build NBT exist, and the cryptographic infrastructure has been running for some years now. Nobody appears to be applying the NBT features and principles at scale. Nevertheless, most of the technical work has been done.

This article discusses how NBT works, and the existing quicknet system.

Technical principle of operation

This section requires an understanding of Public Key Cryptography ↗.

There are two phases to establishing a Not Before Time network:

  1. Publish a static list of public keys far and wide, with one key for every time interval, for example, every half hour. This means there will be a public key corresponding to 0930 on the 3rd February in the year 2043, in the NBT timezone. This is not much data - the full list of keys for even just one century every half hour would require a few hundredths of a second to search on a small computing device.

  2. Broadcast each private key corresponding to the public key at the relevant time. This new private key adds to the growing list of all private keys as time passes, allowing everyone everywhere to unlock information. The guarantees are still kept regardless of any local propagation delays, because the system is Not Before Time rather than Not At Time.

The only significant technical problem left is how to keep the list of private keys secret in a trustable manner. There are several known solutions to this as described further down in this document.

Technical description of guarantees

In the language of Computer Science, the three NBT guarantees give strong cryptographic certainty to the users wishing to share information so that they can:

  • choose a date and time before which the information is unreadable, using accepted algorithms and implementations;
  • prove that the information was not electronically signed and sealed before a certain date/time. This is new, because electronic signatures do not currently incorporate a timestamp in a universally trusted manner. An RFC 3161 ↗ time-stamp proves only that a document existed no later than a given time;
  • prove conclusively that a sealed package was not assembled before a certain time. This is new, because electronic signatures do not make any statement about the original source of the material they are authenticating.

We can summarise these guarantees in this short-cut technical form:

  1. No early reading
  2. No late denials
  3. No earlier sealing (Backdating)

NBT is about a new, universal data primitive that has these properties.

Real-world implementation: the private key server

The simplest NBT would keep the list of private keys on a single server, which publishes each one when its time arrives. That would not be a good public because the NBT private key server is a single point of failure. A successful attacker has a secret copy of all private keys, and would be able to quietly monitor NBT-locked data and decrypt it for their own nefarious purposes.

This is the problem of human mistrust, studied by computer scientists in the field of Secure Multiparty Computation ↗, specifically Threshold Cryptography.

The human mistrust problem

The fundamental problem is that no one group of implementers can be trusted for NBT. For example suppose Russian computer scientists deploy the system; then Americans will claim that the FSA is in possession of all secret keys; if the Swiss government sponsors a neutral NBT server then the Chinese government will claim the CIA stole the secret list and so. How can we know what is true? Or suppose some large timekeeping company decide NBT is their commercial future and they wish to lock out other companies from the service, Apple or Tencent might issue legal threats to stop them, and so on. Nobody trusts anybody else.

The mistrust doesn’t just stop with initial key creation. Even if the NBT server is set up so that it automatically creates the two lists without human intervention, and even if the world can agree on who should do that work, someone else will claim that the code involved was compromised (or the computer doing the work was compromised before it was locked down) and so on.

And even if the notional NBT server was created securely, can we trust where it is housed? If enough money, intellect and political will is focussed then the NBT server can probably be cracked somehow.

The mathematical solution

Shamir’s Secret Sharing (SSS) ↗ is a secret sharing algorithm that has been tested and validated for decades. SSS is accepted to be resistant to even very well-funded modern attackers. All the threshold solutions analysed so far for managing the secret NBT keys are SSS variants.

Quoting the non-technical SSS explanation from Wikipedia:

The secret is split into multiple shares, which individually do not give any information about the secret.

To reconstruct a secret secured by SSS, a number of shares is needed, called the threshold. No information about the secret can be gained from any number of shares below the threshold (a property called perfect secrecy).

For NBT, the process is:

  • (say) 12 mutually mistrustful parties each prepare to host portions of the secret keys for publication.
  • The 12 parties run a distributed key generation ↗ protocol. Each party ends up holding its own portion of the secret keys, and no complete secret key exists anywhere before its release time. Michael Rabin and Christopher Thorpe proposed this approach in 2006, and drand uses it today.
  • Verifiable secret sharing lets each party check that the portions it received are consistent, so a party that cheats during setup is detected.
  • With identity-based encryption, the parties publish a single public key, and the public key for any future time interval can be computed from it. There is no list of public keys to distribute or search.

There are many implementation details and refinements. But in the end, whether the private key portion servers are co-located with existing time servers, or hosted by various militaries, or launched into space on cheap satellites, NBT always operates in the same way. Mutually mistrustful parties publish their key portions to each other, and anyone wanting to broadcast keys for reading time-locked information must be one of these parties and must collect enough of these key portions to meet the SSS threshold.

Zero-knowledge proofs

Secret sharing keeps each NBT key split among the (say) 12 parties until its release time, but it doesn’t show that each party’s key portion is correct. That needs zero-knowledge proofs ↗, which show someone has the right key without revealing what the key is. McFly (2023) ↗ uses zero-knowledge methods to prove someone has a message locked to a particular time without revealing anything else.

Many proof systems need to be initialised from a secret number, and anyone who knows the number could fake proofs. The Ethereum blockchain needed such numbers in 2023 and solved this problem by inviting the whole world to make numbers which Ethereum uses to check data attached to blocks, and then thrown away. In the KZG ceremony ↗, 141,416 contributors each mixed in random data from a browser which was then made into the secret. Since reconstructing the secret needs every contribution, any one of those contributors knows the system is secure without needing to trust anyone else. If NBT uses such a proof system, it needs such a ceremony. Some newer proof systems do not even need a ceremony, so this difficult problem has been solved in multiple ways.

quicknet

Since 2023 an organisation called the League of Entropy ↗, seventeen companies and universities, has run the quicknet network ↗. quicknet implements a large part of NBT.

Compared with the design above, quicknet has:

  • A key for every time interval. quicknet releases a key every three seconds, and one public key covers every future round through identity-based encryption.
  • Mutually mistrustful parties. The nodes ran distributed key generation, so no complete secret key exists before its round, and a threshold of nodes must cooperate to release one.
  • No early reading. Anyone can lock a file to a future round with the tlock ↗ library based on the 2023 Timelock Encryption Scheme paper ↗. An example sealed-bid auction ↗ has been built on it.
  • No late denials and no earlier sealing. These are implied by the quicknet design, although there is no client software support.

quicknet lacks the rest of NBT:

  • quicknet encryption is not quantum-resistant.
  • quicknet operators are a small group of specialists. Production scale and trust are a big further step.
  • there is no RFC series for quicknet.
  • there is no user software or integration with existing software stacks.

quicknet shows that the remaining problems are engineering, political and social.

Technical steps to demonstration projects

A prototype needs to build client software around tlock, which could be:

  • an NBT client as a web service, which receives unencrypted files and returns them locked to an NBT future date, and also receives locked files and either unlocks them or states when they can be unlocked.
  • a virtual printer device for various operating systems that functions in the same way as “Print to PDF”, except it asks for a date in the future in order to create an NBT-locked PDF.
  • a small modification to office suite software, where File/Export As PDF and File Save/As has a new field for “Not Readable Before This Date”.
  • a modified PDF reader OR a corresponding modification to File/Open in office suites such that if an NBT header is detected in a file then the key is fetched from a public quicknet node.
  • No earlier sealing, added to each of the above by including the most recently released quicknet key in every file locked or signed.

GPS analogy for NBT commerce

The NBT broadcast facility is base infrastructure without a direct commercial model, in the same way GPS (and GLONASS, Galileo etc) are not. It does not make sense to try to limit access to the NBT broadcast facility any more than GPS should be restricted.

For NBT, the immediate monetisable value is in applications on top of this base infrastructure. The human rights use case cannot be shut down without also shutting down commercial activity, so even from launch the facility gives its users a reason to protect it, by running more quicknet nodes and supporting more client software.

I have been working on the NBT concept for more than a decade, focussing mostly on the societal problems that it could address. In January 2023 a preprint by Ke Yuan et al caught my eye, called Multiple time servers timed-release encryption based on Shamir secret sharing for EHR cloud system ↗. This is a theoretical analysis of a practical real-world implementation of an NBT, with excellent literature research. quicknet is one implementation of an NBT, where other kinds of implementation are possible and indeed likely. This is where a carefully developed RFC suite will help coalesce implementation views.

Selected relevant publications: