Now that I’ve gotten that rote disclaimer out of the way, let’s dive in.
Anastasios Angelopoulos, engineer, Stanford alum B.S. EE
It has been a year since I first invested in Golem and in my opinion the project has little to show for its effort. That’s not necessarily disappointing given the team’s goal of creating something that has never been done before at scale. Needless to say, it appears Golem’s approach is reaching major hurdles. With distributed architectures, there are two primary unavoidable consequences.
The first is the high amount of network communication and data transfer overhead. The second is the heterogenous nature of distributed networks: hardware is differentiated, capacity per node fluctuates, and node availability comes and goes as computers and servers go offline. This can be solved for in centralized data centers since engineers are familiar with the overall architecture, state, and availability of each machine. With distributed networks, the known state and availability of each machine is unknown. This dynamism can be solved by increasing communication overhead but that will make computations over a distributed network expensive. Additionally, it forces projects such as Golem to focus on specific tasks (rendering) to reduce another layer of overhead. Might as well stick with data centers or Amazon AWS.
I mentioned CPUs up above to bring up a point of GPU miners. Mining rigs typically utilize their full capacity to solve one specific problem. The type of computing Hypernet aims to achieve is broader than GPU mining and Golem’s focus on rendering. Hypernet’s goal of enabling broader computing is much more ambitious. Golem and SONM focused the tasks they solved, so Hypernet's ambitions may be its largest risk.
Hypernet will provide an API suite to perform parallel data operations. Bandwidth and availability will be managed by the Distributed Average Consensus (DAC). The DAC would presumably dynamically adjust supply even after nodes drop out or experience bandwidth declines to meet the buyer’s needs. Buyers will post a job to the market and that job will be assigned to a Virtual Machine (VM) or a series of VMs. In practice, the buyer would be able to run multiple jobs in parallel on a machine. Once a particular VM completes its task, it can move to the next job even if the job is completely different. Recruiting CPU and GPU capacity at sufficient specifications to match buyer's needs is another risk factor to Hypernetwork. If the incentives do not align properly to bring high end computing resources online, it leaves Hypernetwork to solve lower level computational needs. The billion dollar question lies in how valuable are low level tasks and what is the overall opportunity costs of missing out on high computational tasks.
One of the first tests the team did was fun HN on six different Raspberry Pis to perform a parallel Matrix Multiplication (MMUL). On Raspberry Pis! I am still waiting on the team to reply to me to see their full stack demo. Their GitHub is also currently not public so I will definitely add to this review as they announce when it will be public. Since the project is potentially groundbreaking, I would hope they open source their code base for the general community to innovate further on.
Hypernet is in its early stages. I see this project garnering a ton of hype just based on the bare bones team assembled. With a reasonable pre-raise the team could afford to bring on professionals focused on Go-To-Market. ICOs are no longer merely whitepaper projects but require the support of sales, marketing, PR, and legal teams. Publishing a protocol is just not enough anymore.
With that said, I invested in Golem because I wanted to fund the next era of space exploration… I mean distributed computing. I wanted to be a part of the future. I get that same “feeling” with Hypernet. As a former finance professional it pains to use the word “feeling” when it comes to investing. But this is what it is. My numbers based argument should be one that you’re thinking about right now. You might be thinking Golem is a competitor. Were Tesla and Edison competitors in enabling electricity? They were, but they also complemented and pushed each other to bring a product ready for mass consumption. Hypernet and Golem are taking different approaches, as they should. It’s unclear which architecture or if any of these will prevail. Golem has already reached a market cap well over half a billion dollars. That may be a fluffy number so let’s mark this down into the bargain bin at 50%. With Hypernet’s stated hard cap of $15M it’s possible Hypernet will richly reward early speculators. As far as long term technology goes, I sure hope Hypernet figuratively and literally lands us on the moon. The question remains... will Hypernet's computing be faster, cheaper, and more secure than centralized computing. I'm not sure, but I will be personally looking to participate in this project. With such a small cap, an intelligent team, and a speculators market I will be rolling the dice.
Sources
Thank you for coming to the site. Quantalysus publishes blockchain research and analysis for the crypto community. Please follow on Twitter, Steem (please follow and upvote if you can – thanks!), Telegram channel (New!), and Medium to stay up to date.
If you want to earn Aelf (ELF) tokens for just using Twitter and Reddit, sign up for their candy / bounty program.
If you learned something:
