0.

Hello everyone. Before I begin my first post, let me briefly introduce myself and my team.

I am a college student, majoring in mechanical engineering and computer science at Seoul National University of South Korea.

I missed out on a huge opportunity to stockpile on Bitcoin a few years ago. Luckily, early this year, I gained knowledge of and started investment in Ethereum. Ever since, the blockchain/crypto-currency market has fascinated me from both a financial and technological perspective.

This led me and some of my friends who share similar interests/backgrounds to prepare a cryptocurrency-specialized media outlet by the name of “CoinPresso”.

This first piece may lack in many areas. Therefore, I have labelled each subtopic, and any constructive criticism and/or discussions related to this piece will be duly noted and much appreciated.

Thank you.

Original author: Donggun Jung (CoinPresso [CEO] / [email protected])
Translation: Hyunbin Kim (CoinPresso [VP/Manager of American Affairs] / [email protected])

1. Background Knowledge

Golem is a token based on Ethereum.

Assuming that, since you are on Steemit after all, you have a basic knowledge of what Ethereum is and how cryptocurrencies work: What are tokens?

To provide an understandable analogy – let’s think about smartphones. Smartphones have various operating systems, i.e. platforms such as Android or iOS. On these platforms run various apps, which allow you to call friends, take photos, play games, and so forth.

Basically, cryptocurrencies such as Ethereum are platforms, while tokens such as Golem are special applications designed to work on those platforms.

Many tokens, like platform-cryptocurrencies, are listed and traded on international exchanges such as Poloniex or Bittrex.

Golem is currently priced in the 60-70₡ range, and has a market capitalization (price per unit multiplied by circulating supply – indicates size of market) of nearly $600mn, placing 13th among all existing cryptocurrencies and tokens.

So far, Golem has grown consistently with low volatility, which is why I have recommended this token multiple times to acquaintances who are newbies in the market and/or wish to start a long-term investment.

2. What Golem does

In layman’s terms: Golem lets users share their computers’ idle (computing) power.

A similar example – Airbnb. Utilizing a sharing economy, Airbnb lets users lend and borrow spaces when that space is unoccupied, that is, when it is not achieving its full potential.

The same thing happens with Golem, with the token acting as a medium for sharing computing power.

Numerous computers that run 24/7, e.g. computers in government institutions or supercomputers, retain idle power most of the time. This means that these computers, like unoccupied spaces, are not achieving their full potential.

Owners, therefore, have incentive to sell this idle power to max out upon their potential while buyers have incentive to gain extra computing power without the necessity to upgrade hardware.

Largely three doubts/concerns arise.

3. Doubt: Does Golem’s target market have enough potential growth?

For a sharing economy service to prosper, units must be priced at a level where the seller’s potential economic benefits are larger than the decrease in utility experienced by having to participate in a previously unnecessary action, i.e., nuisance in using the service.

Examples: Car/house sharing models Uber and Airbnb have both grown exponentially over the past couple of years.

On the other hand, services that have a smaller market pie, such as ones that help college students share their textbooks, are failing to gain tangible results.

Many Koreans think that the computing power-sharing market will be a minor one. This is due to the fact that PC rooms, where you can borrow computers at a rate of $1 per hour, are abundant in Korea.

Another problem that restrains consumers from participating in this market: Most individuals use computers for repetitive tasks. If such tasks require that individuals use more computing power than is available to them, the individual would be better off buying a better computer rather than paying a continuous stream of money every time he/she initiates the task.

This is why I believe large institutions will be the principal agents in sharing computing power.

That is, I believe that while in a B2C or C2C sector, the nuisance of sharing computing power might outweigh the benefits, in the B2B sector, it clearly has unique benefits.

So, in what fields are there high demands for computing power?

A. Entertainment

Blockbuster Hollywood movies require complex computer graphics, and this costs an astronomical sum of money.

For instance, the 2007 movie Spiderman 3 was reported to have used over $80mn on CG alone. Even excluding software and human resource costs, related hardware costs would still be somewhere in the millions of dollars.

To satisfy consumers’ needs without sharing computing power, the media industry would need to push and upgrade hardware to its limits, something that is very expensive and sometimes even impossible when computing power is already near its hardware-wise upper bound.

B. Artificial Intelligence

One of society’s focal points, and probably the one most debated upon, in the status quo.

This may seem counterintuitive to those uninformed on the topic. After all, Siri seems to provide quick responses even on your phone, so why would A.I. require so much computing power?

The truth A.I. service providers want you to be unaware of is that there are thousands of computers working at Apple’s servers.

These computers convert speech to text. They create part-of-speech tags for each word, and tries to comprehend the context of the speech. They calculate and create an acceptable response, converts it back to speech, and sends it back to your phone.

Most importantly, A.I. systems within these servers learn more and more about human languages and communication as data accumulates.

The same is true of AlphaGo. While it looked as if AlphaGo was effortlessly calculating its possible positions, there were close to 2,000 CPUs calculating each and every move back at Google’s servers.

The pre-match machine learning which took place in order to “teach” AlphaGo how to play Go also required tons of data and an unfathomable amount of computations.

Even though AlphaGo did win that match, considering the amount of capital and manpower devoted to creating AlphaGo, currently, raising one Lee Sedol or Ke Jie may be more cost efficient.

Thus, again, we see the importance of sharing computational power in B2B contexts.

A.I. is taking its seat in various fields such as speech-to-text, gaming, autonomous driving, and so forth.

The most important part in developing and implementing this new technology is computing power. Something that smaller companies may be unable to equip themselves with, since supercomputers and high-end processing units are extremely expensive.

As the A.I. industry grows in size and more and more companies dive into the field, shared computing power will be an industry with limitless growth potential.

From this perspective, Golem’s market capitalization of $600mn is far from overrated. Rather, it might even be underrated, unable to take into account its growth potential as of yet.

4. Doubt: Does Golem have competitiveness in its market?

Now that we have dismissed doubts on whether or not the market can grow, let us look at competitors within the market itself.

Those of you who are developers have probably thought of the Amazon Web Service (AWS) at least once up until now. You might be wondering why Golem needs to exist when AWS already does most of this for you.

Some of you, less informed on such topics, may simply be wondering if there are other tokens that do similar jobs better.

Let’s look at each possible competitor, and do a threat assessment on Golem.

A. Comparison with existing cloud-computing solutions

We know AWS is relatively cheap, but why is it cheap?

The only reason AWS can be cheap is because Amazon has been able to achieve economies-of-scale in the server market. That is, Amazon has reached a stage where larger outputs lead to lower costs per unit of output.

Due to this semi-monopolistic state, even including the managing fees Amazon imposes, AWS has been cheaper than most alternatives.

That is, the amount we pay Amazon is comprised of something more than just the capital Amazon invested into its servers to reach economies-of-scale. We are also paying managing fees, which aims to compensate the efforts Amazon, the central institution, has made to establish trust in its system.

Then, what if “trust” can be assumed without the involvement of a central institution?

This is the forte of decentralized smart contracts. Anyone can provide services, and the costs required for contracts to gain “trust” is minimal.

In short, the managing fees a centralized platform administrator normally takes are distributed to users of the blockchain network in a decentralized platform, making it that much more favorable due to lower costs.

Golem, with its significant reduction of managing costs, certainly has an edge over existing solutions.

B. Other blockchain competitors

Then, if existing solutions are worse than Golem, can’t new, more technologically competent competitors arise to expel Golem from its throne?

Personally, I consider market dominance to be one of the most important criteria in making investment decisions.

Simply put, I do not believe that a newly developed, better-tasting coke will be able to thwart Coca-Cola, which already dominates distribution channels, and, as a result, public perception.

Golem is currently the Coca-Cola of supercomputer/computational power-related tokens. This is evident in its vastly higher market capitalization compared to other, similar tokens.

(My decision not to invest in SONM was due to similar reasons)

I believe it is wise to only invest upon those dominators which have already gained the upper hand in their fields.

5. Doubt: Golem's technical integrity

Since Golem is still in alpha, there are bound to be concerns about whether or not Golem’s vision is actually attainable. For instance, some might question the feasibility of breaking down computations and distributing it among numerous supercomputers.

A. Development of parallel computing

Most modern computers handle computations using matrixes.

An example to help you understand:

Let’s think of a situation where you have to multiply two 1000*1000 matrixes.

For each of the 1,000,000 numbers in the first matrix, 1,000 multiplications and additions are carried out, adding up to a total of 1 billion calculations.

-Using series (non-parallel) computing: Each of the billion calculations are handled by the CPU. Runtime is determined by the number of calculations the CPU handles per second.

-Using parallel computing: Each of the 1 million numbers that we use are distributed to 1,000 smaller CPUs (Graphics Processing Units are mostly structured this way, utilizing groups of less competent CPUs). Each CPU is slower than the one above, but due to a much smaller load per CPU, runtime also becomes much shorter.

Of course, this is an extremely simplified example. In reality, distributing computations takes time in the execution stage and effort in the programming stage.

Still, many modern programs apply parallel computing to utilize the multiple cores that CPUs these days have.

Therefore, the technological difficulties are not as large as we might assume. Programs that are meant to be run by supercomputers have long embraced parallel computing as a principle.

B. Confidence in the Golem team

Numerous updates still remain. Golem is still in its infancy, leading many to question whether or not the developers are trustworthy.

https://goo.gl/OiIr3c

This blog post from the founder of Golem helped me solidify my trust.

Long story short, the post talks of how, rather than pressing Brass Golem to arrive before the deadline, they want quality upgrades, even if it means a certain amount of delay.

Conforming to the deadline would have raised Golem prices in the short-term. People rarely delve deep into the technological specifications of an update. Rather, the update itself acts as a trigger to induce price spikes.

If Golem continuously dispels such temptations and strives to realize their endgame, their big picture, and their technological goals, my investment will surely prevail.

[A word from the translator]

To those who have been kind and patient enough to read to the end. Bear in mind, this is not an exact translation of the original piece. Elements have been added and subtracted to suit non-Korean audiences (All of our writers are Korean, so some of our examples are unknown to foreigners). Sometimes, more explanation was given when elaboration was deemed necessary.

A short introduction of myself: I am currently double majoring in business and computer sciences at Seoul National University in Korea. These past few months, I have been pissed off at the lack of information on cryptocurrencies. It felt as though my investments were blind. They were substantiated only by mere gut feelings, none of which had anything to do with hard facts. Through CoinPresso, I wish to open my eyes. Not only that, I strive to help fellow investors open their eyes, so that this promising market is no longer swayed by speculative capital.

Opinion: Is Golem the future of super-computing?

0.