The Libr a Bloc k c hain Zac har y Amsden, Ramnik Ar or a, Shehar B ano , Ma thieu B aude t , S am Blac k shear , A bhay B o thr a, G eor g e Cabr er a, Chris tian Ca t alini, K ons t antino s Chalkia s, E v an Cheng , Aver y Ching , Andr e y Chur sin, G eor g e Danezis, G er ar do Di Giacomo , David L. Dill, Hui Ding , Nic k D oudc henk o , Vic t or Gao , Zhenhuan Gao , F r anois Garillo t , Mic hael G or ven, Philip Hay e s, J. Mark Hou, Y uxuan Hu, K evin Hurle y , K evin L ewi, Chunqi Li, Zekun Li, Dahlia Malkhi, S onia Mar gulis, B en Maur er , P ay man Moha ssel, Ladi de Naur ois, V aleria Nik olaenk o , T odd Now ac ki, Olek s andr Orlov , Dmitri P er elman, A lis t air P o tt , Br e tt P r oc t or , Sha z Qadeer , Rain, Dario Russi, Br y an S c hw ab , S t ephane S ez er , A lber t o S onnino , Herman V ent er , L ei W ei, N ils W erner f elt , Br andon Williams, Qinf an Wu, Xif an Y an, Tim Zakian, Runtian Zhou * Abstract. The Libra Blo c k c hain is a decen tralized, programmable database designed to supp ort a lo w-v olatilit y crypto currency that will ha v e the abilit y to serv e as an efficien t medium of exc hange for billions of p eople around the w orld. W e presen t a prop osal for the Libra proto col, whic h implemen ts the Libra Blo c k c hain and aims to create a financial infrastructure that can foster inno v ation, lo w er barriers to en try , and impro v e access to financial services. T o v alidate the design of the Libra proto col, w e ha v e built an op en-source protot yp e implemen tation Libr a Cor e in an ticipation of a global collab orativ e effort to adv ance this new ecosystem. The Libra proto col allo ws a set of replicas referred to as v alidators from differen t authorities to join tly main tain a database of programmable resources. These resources are o wned b y differen t user accoun ts authen ticated b y public k ey cryptograph y and adhere to custom rules sp ecified b y the dev elop ers of these resources. V alidators pro cess transactions and in teract with eac h other to reac h consensus on the state of the database. T ransactions are based on predefined and, in future v ersions, user-defined smart con tracts in a new programming language called Move . W e use Mo v e to define the core me c hanisms of the blo c k c hain, suc h as the currency and v alidator mem b ership. These core mec hanisms enable the creation of a unique go v ernance mec hanism that builds on the stabilit y and reputation of existing institutions in the early da ys but transitions to a fully op en system o v er time. 1 Intr odu c tion The spread of the in ternet and resulting digitization of pro ducts and services ha v e increased efficiency , lo w ered barriers to en try , and reduced costs across most industries. This connectivit y has driv en economic emp o w ermen t b y enabling more p eople to access the financial ecosystem. Despite this progress, access to financial services is still limited for those who need it most impacted b y cost, reliabilit y , and the abilit y to seamlessly send money . This pap er presen ts a prop osal for the Libra proto col, whic h supp orts the newly ed Libra ecosys- tem that seeks to address these c hallenges, expand access to capital, and serv e as a plat for inno v ativ e financial services. This ecosystem will offer a new global currency the Libra coin whic h will b e fully bac k ed with a bask et of bank dep osits and treasuries from high-qualit y cen tral ∗ The authors w ork at Calibra, a subsidiary of F aceb o ok, Inc., and con tribute this pap er to the Libra Asso ciation under a Creativ e Commons A ttribution 4.0 In ternational License . F or more ination on the Libra ecosystem, please refer to the Libra white pap er [ 1 ]. 1banks. All of these currencies exp erience relativ ely lo w inflation, and th us the coin mec hanically inherits this prop ert y as w ell as the adv an tages of a geographically div ersified p ortfolio of assets. The Libra proto col m ust scale to supp ort the transaction v olume necessary for this curre ncy to gro w in to a global financial infrastructure and pro vide the flexibilit y to implemen t the economic and go v ernance p olicies that supp ort its op erations. The Libra proto col is designed from the ground up to holisti- cally address these requiremen ts and build on the learnings from existing pro jects and researc h a com bination of no v el approac hes and w ell-understo o d tec hniques. A k ey prerequisite for health y comp etition and inno v ation in financial services is the abilit y to rely on common infrastructure for pro cessing transactions, main taining accoun ts, and ensuring in terop er- abilit y across services and organizations. By lo w ering barriers to en try and switc hing costs, the Libra proto col will enable startups and incum b en ts to comp ete on a lev el pla ying field, and exp erimen t with new t yp es of business mo dels and financial applications. Blo c k c hain tec hnology lends itself w ell to address these issues b ecause it can b e used to ensure that no single en tit y has con trol o v er the ecosystem or can unilaterally shap e its ev olution to its adv an tage [ 2 ]. The Libra Blo c k c hain will b e decen tralized, consisting of a collection of v alidators that w ork together to pro cess transactions and main tain the state of the blo c k c hain. These v alidators also the mem b ership of the Libra Asso ciation, whic h will pro vide a framew ork for the go v ernance of the net w ork and the reserv e that bac ks the coin. Initially , the asso ciation and v alidators will consist of a geographically distributed and div erse set of F ounding Mem b ers. These mem b ers are organizations c hosen according to ob jectiv e participation criteria, including that they ha v e a stak e in b o otstrapping the Libra ecosystem and in v esting resources to w ard its success. Ov er time, mem b ership eligibilit y will shift to b ecome completely op en and based only on the mem b er’s holdings of Libra. The asso ciation has published rep orts outlining its vision [ 1 ], its prop osed structure [ 3 ], the coin’s economics [ 4 ], and the roadmap for the shift to w ard a p ermissionless system [ 5 ]. This pap er is the first step to w ard building a tec hnical infrastructure to supp ort the Libra ecosystem. W e are publishing this early rep ort to seek feedbac k from the comm unit y on the initial design, the plans for ev olving the system, and the curren tly unresolv ed researc h c hallenges discussed in the prop osal. Th us, the asso ciation has established an op en-source comm unit y [ 6 ] for the discussion and dev elopmen t of the pro ject. The Libra proto col. The Libra Blo c k c hain is a cryptographically authen ticated database [ 7 , 8 , 9 ] main tained using the Libra proto col. The database stores a ledger of programmable res ources, suc h as Libra coins. A resource adheres to custom rules sp ecified b y its declaring mo dule, whic h is also stored in the database. A resource is o wned b y an accoun t that is authen ticated using public k ey cryptograph y . An accoun t could represen t direct end users of the system as w ell as en tities, suc h as custo dial w allets, that act on b ehalf of their users. An accoun t’s o wner can sign transactions that op erate on the resources held within the accoun t. Figure 1 sho ws the t w o t yp es of en tities that in teract using the Libra proto col 1 v alidators, whic h main tain the database and 2 clien ts, whic h p er queries on the database and transactions to mo dify it. 1 validators client other validators 2 4 5 leader cution 3 cution 3 Figure 1 Ov erview of the Libra proto col. 2V alidators main tain the database and pro cess transactions ted b y clien ts for inclusion in the database 1 . The v alidators use a distributed consensus proto col to agree on an ev er-gro wing list of transactions that ha v e b een committed to the database as w ell as the results of cuting those transactions. This consensus proto col m ust b e reliable ev en in the presence of malicious or erroneous b eha vior b y a minorit y of v alidators. V alidators tak e turns driving the pro cess of accepting transac- tions. When a v alidator acts as a leader, it prop oses transactions, b oth those directly ted to it b y clien ts and those indirectly ted through other v alidators, to the other v alidators 2 . All v alidators cute the transactions 3 and an authen ticated data structure that con tains the new ledger history . The v alidators v ote on the authen ticator for this data structure as part of the consensus proto col 4 . As part of committing a transaction T i at v ersion i , the consensus proto col outputs a signature on the full state of the database at v ersion i including its en tire history to authen ticate resp onses to queries from clien ts 5 . Clien ts can issue queries to a v alidator to read data from the database. Since the database is authen- ticated, clien ts can b e assured of the accuracy of the resp onse to their query . As part of the resp onse to a read query , a v alidator returns a signed authen ticator for the latest v ersion i of the database kno wn to the v alidator. In addition, a clien t can optionally create a replica of the en tire database b y sync hronizing the transaction history from the v alidators. While creating a replica, a clien t can v erify that v alidators cuted transactions correctly , whic h increases accoun tabilit y and transparency in the system. Other clien ts can read from a clien t that holds a replica in the same w a y they w ould read from a v alidator to v erify the authen ticit y of the resp onse. F or the sak e of simplicit y , the rest of this pap er assumes that clien ts query a v alidator directly rather than a replica. Organization. This pap er discusses the comp onen ts of the Libra proto col Logical Data Mo del Section 2 describ es the logical data mo del that organizes the decen- tralized database visible to v alidators and clien ts. cuting T ransactions Section 3 describ es the use of Mo v e [ 10 ], a new programming language that is used to define and cute database trans actions. A uthen ticated Data Structures and Storage Section 4 describ es the mapping of the logical mo del in to authen ticated data structures based on Merkle t rees [ 11 ]. Byzan tine F ault T oleran t Consensus Section 5 describ es the LibraBFT [ 12 ] v arian t of the HotStuff proto col [ 13 ], whic h allo ws a net w ork with p oten tially malicious v alidators to main tain a single, consisten t database b y cuting transactions with Mo v e and coming to agreemen t on their cution using the authen ticated data structures. Net w orking Section 6 describ es the proto col that enables v alidators to comm unicate with eac h other securely , as required for consensus . Subsequen tly , w e presen t the op en-source Libra Core protot yp e [ 6 ]. Section 7 discusses ho w Libra Core com bines the comp onen ts of the Libra proto col to pro cess a transaction. Section 8 discusses p erance considerations. Finally , w e explain ho w the proto col is b eing used to supp ort the economic stabilit y and go v ernance p olicies of the Libra ecosystem. Section 9 sho ws ho w w e use the Mo v e language to implemen t the lo w- v olatilit y , reserv e-bac k ed Libra coin and a v alidator managemen t system that mirrors the go v ernance of the Libra Asso ciation. Section 10 concludes the pap er with a discussion of future plans and ongoing c hallenges for the Libra ecosystem. 3