By Ben Munster
5 min read
How do you build a blockchain that is decentralized, secure, and fast? Is it possible? Inevitably, the successful implementation of one of these variables comes at the expense of another.
Nevertheless, the developers behind Aptos think they have the answer in an experimental transaction-ordering algorithm that is the successor to the Diem experiment. But it's not without some hiccups.
Aptos is very new on the scene. The recipient of a $150 million funding round led by FTX and investors including Parafi in late July, the speedy blockchain landed another $200 million strategic round from big players like Andreessen Horowitz, Multicoin Capital, and Haun Ventures in March.
Founded by Avery Ching and Mo Shaik, who worked on Diem’s Novi wallet, Aptos makes use of a technique outlined in the Diem days called “parallel execution” (of which more later) that purports to whip up transaction speeds while keeping them cheap.
Like another project, Sui, it brands itself as the inheritor of Diem’s legacy, and uses that blockchain’s proprietary programming language, Move.
The blockchain’s tokenomics drew a lot of criticism for being unveiled the day after its mainnet launch in October. A common refrain on Twitter at the time was that after raising $350 million at a valuation of more than $4 billion, Aptos should have been quicker to share details about how its native APT token had been distributed.
There remain questions over how different Aptos is from other blockchains, whether “parallel execution” works as promised, and how the new blockchain can muster a large enough ecosystem of programmers willing to learn its relatively unknown programming language.
As mentioned, Aptos’s key feature is parallel execution. Most blockchains use a method of transaction ordering called sequential, or serial execution, in which a single chronology of transactions is continually updated: every time you make a trade or buy something, that transaction is added to a single long ledger containing every transaction ever executed on the network and updated via thousands of nodes.
Because they are all added one at a time, it’s necessary to wait for each new transaction to be verified: This takes a long time and is the reason for most blockchains’ glacial pace of settlement.
Parallel execution, on the other hand, runs multiple simultaneous chains in, well, parallel, allowing—theoretically—for more to be processed at once. Indeed, Aptos claims its testnets have already hit 130,000 transactions a second; compare that to Ethereum’s more leisurely 30 a second.
We don’t know yet: Parallel execution is controversial. Whatever benefits it brings to efficiency may come at a cost to security. “Sequential” blockchains are the way they are for a reason: Relying on a single chronology of transactions makes it easier to check the chain’s contents against itself; there is only one place a transaction can feasibly go.
Introducing parallel chains complicates things. With more chains, it is harder to get a full view of the transaction history, and transactions may attempt to process through multiple chains simultaneously, causing them to fail (as well as potentially enticing opportunists to attempt a fraudulent “double spend”).
Aptos’s solution is to wait until the transactional dust has settled and verify all chains at once: adequately placed transactions will go through, while failed ones will be “re-executed.”
It’s an execute first, ask questions later policy—and it is, supposedly, working.
Since its launch, Aptos’s critics have wondered whether its developers can overcome the stigma of having worked at the multinational formerly known as Facebook.
The record suggests yes: Aptos says it has incorporated many of the projects involved with Libra into its ecosystem, and it seems they are happy to build exclusively in Aptos’s comparatively untested programming language.
The prestige and experience of Aptos’s founders appear to have attracted new developers, too: Aptos’s Discord already has over 63,000 members, around 8,000 of which are developers, and the protocol is supported by over 20,000 nodes.
Recently, Aptos also completed a testnet update, and a mainnet launch is due later this year. However, the token won’t be available in the U.S. for regulatory reasons.
Decrypt-a-cookie
This website or its third-party tools use cookies. Cookie policy By clicking the accept button, you agree to the use of cookies.