Topic: HADRON - WEB BROWSER AI MINING (Read 189 times)

Un post interesant de pe reddit referitor la faza de alpha testing:
"I'm seeing lots of posts on the telegram channel asking when we can trade, what the tokenomics are, why alpha testing isn't yet open to everyone. And whilst there's a pinned FAQ explaining the process, not everyone seems to get it. Hopefully, this will explain what alpha testing is, and what it isn't ...

TLDR ("Too Long, Didn't Read" - although I hope you do) or as they used to call it when I was at school ... 'Summary':

This is about IT. But it might help you understand what alpha mining is trying to achieve.

Websites are very complicated applications. They need to run on stuff. They need to be scalable. They need logging and monitoring and a whole host of other things to support them. When a change is made, it can be horrific to roll out. And ideally no-one should notice! This web application needs to run on every device in the world (including smart mirrors - and who knew they even existed? Not me!). Oh and real users do stuff developers never expect them to do (this often results in swearing, and sometimes ... crying)

There is a team of five people doing all of this. They are doing a shedload of stuff. They want it to run on a shedload of devices. Like billions. They need people to test it. We (the eventual end users of some use cases) are the best testers. By seeing how we actually use the website, and by breaking it, the team can learn and build out the product.

This is what alpha mining is all about.

First of all, my background:

I have a background in IT development, more specifically in designing websites which sit on front of horrifically complicated, old pensions systems, with horrifically complicated and fragmented IT systems connecting the printing system (actually the five different print systems because the old systems NEVER go away ..), the processing, the call centre, the sales team. You get the idea.

I don't have a background in crypto, or ICOs. I farmed ripple xrp back in the 'computing for good' days. That's it.

I don't know anything about neural nets.

So I come at this purely from an IT perspective. I know first hand a lot of the issues surrounding 'web development'. And the first issue, really, is that 99.999% of websites are far from 'just' websites ...

It's not 'just' a website

The website is almost exactly like the tip of an iceberg. Under the water is a complicated web infrastructure providing resilience, security, data caching, routing to name a few things. Then there's the service layer - the things that process the requests from the website. Then we get to the data layer. Databases, and backups.

For all of this, we need infrastructure (computers) to run it all on. Enter the cloud. It's wonderful. You may know it in such forms as Microsoft 360, or Google drive. But ... The cloud is really made up of a lot of machines which are 'just big enough' to do what they need to do. Bigger or faster machines cost more. Why pay more than you need to?

But, what if you want to add, say, 3000 users to your system? You need more capacity in everything. Almost at once! That doesn't happen by magic. Usually there will be some kind of automated scaling process which increases capacity as required (for instance by upgrading a server to the next model). When I say automatic, obviously I don't really mean automatic. What I mean is that actual people (the dev team) have to craft code to allow the system to do this. And test that it works. Oh ... And make sure that the 'website' stays up and running.

What if the team want to change a bit of code? Well depending on what the code does, it could impact just the website (really, in this case, I do mean just the website), or maybe the service layer, or the data layer. In a lot of cases it's going to hit them all. And the team need to be able to test it then roll it out. All that infrastructure is going to be replicated (at a very much smaller scale) for testing. Likely the team will also be putting in place automated testing - the idea behind that being you can run a suite of tests 'automatically' that will give you the confidence you haven't broken anything. These tools are going to cover a lot of things - like code syntax, and basic functional tests, and will need to be able to stop a build if it fails - and report back to the team. Once again - all of that needs built, and developed (this is a very basic summary of a 'devops pipeline').

And then there's the monitoring. You need more things to 'automatically' restart other things when they fail. You need things to monitor the website usage. Who is using it? How are they using it? Are they getting errors? What about logging? All of these things log. When systems are 'live' (I mean really out there, like Amazon. Not in alpha testing) they probably just log errors. In early testing you'll log everything. And there will be a lot of errors at the start. So you have lots of logs. You need something to search those haystacks for needles, 'automatically'. So someone needs to work out (ie write) the rules for what's hot and what's not in the logs.

Security. Did I mention that? All our personal data is encrypted 'at rest', for instance. And then the hackers. If you are able to be reached on the internet, they will try to hack you. So threat detection etc needs built in. And if you get hacked, well, the reputational damage will be huge. You might never recover.

This is what a 'website' is.

You don't have to get it right from day 1, but you need the backbone of everything in there so you can build out features as you develop the application.

None of this is easy. But this is a flavour of what this very small team of FIVE have been working on (Apologies Hadron. I will have missed a significant amount of stuff!). Oh yes and at the same time as all of that, they're very actively supporting a telegram channel with 8.6K people on it (and rising). Setting up welcome bots, and spam bots and responding to queries. Applaud them. They deserve it.

So let's talk about testing now!

Remember this is a small team of five. They want us to be able to perform AI computations on billions of devices. How on earth would you test that? Most likely you'll take advantage of software like Browser Stack which allows you to 'virtually test' on real devices (I never had any idea how that magic worked ...!). Or you might be able to hire a room from a company who have devices for testing. It's highly unlikely they'll have a smart mirror though ...

At some point the team will have done all they can. They need to start rolling out an early version of the product to testers - us!

So the first wave of testers was small. I'll speculate (perhaps someone can confirm) they were asked to test very specific scenarios. I think I heard of someone running 96 tabs? I imagine there was a lot of structure and communication sitting around this first wave. And debugging and retesting. Then the numbers increased gradually. I think I was part of the second wave. I saw a lot of people joining the mining thread at the same time. We broke it. I could barely mine at all for the first day. I think I got 4 coins in total! But we're not mining - we're testing. And that was an extremely successful test. I saw a lot of server gateway timeouts (that's one of the things sitting behind the website) so I imagine the infrastructure scaling didn't go quite as planned.

This is what we are doing - helping get the product stable. Once that happens they team can turn their focus more to other aspects, like tokenomics. And start to answer some more questions.

Conclusions

So for everyone who is in the queue for alpha mining, I hope this gives you an idea of what we are testing and the work that is going on behind the scenes. The team are doing a great job of fixing problems, and we're seeing ever-increasing stability. Rest assured your time will come to test, hopefully soon. This is a great opportunity for the community to help build a valuable product, and from what I hear, the opportunity to contribute to a product this early on is scarce. Grasp that opportunity, and remember, when you do, contribute to the mining thread, raise bugs when you find them and contribute to the project. Think about more than just getting to your echelon cap!

I hope this helps you understand why not everyone can mine at once right now, and what we are helping to build. It's definitely not 'just a website'."

Un mic update despre Hadron. Aseara s-a facut si un update si s-au rezolvat niste erori care apareau in timpul minatului. Probabil or sa mai apara cu cat numarul persoanelor care mineaza creste.

"Hello,

Cliff here, I wanted to give you an update now that the network and product have launched.

First, I want to thank all of you for your continued support and enthusiasm for the HADRON.cloud project. As of today, we have over 8,000 Telegram users and one of the hottest threads on BitcoinTalk, exceeding Stellar and closing fast on Ripple. Thanks to the early efforts of some excellent community members, we also have a thriving subreddit.

The network went live and mining began on-schedule on January 15th, and we’ve extended invitations to over 5,000 miners out of the many more who have applied.

We are grateful to those who have participated in the test program. Collectively, your devices have earned thousands of coins and processed well over 150,000,000 images through the network, which is an extraordinary number. That's 6000% faster than our powerful Nvidia Titan test cluster could muster. This computation rate will only increase as we scale out to more people.
Throughout this process, you’ve helped us find many ways to improve the service. We're also surprised by the range of devices used by our miners, including unusual phones, miniature SoC systems, and even a smart mirror! So far, the mirror is winning the "most eye-catching HADRON.cloud miner" contest as suggested by our Alpha Echelon testers.

We know many of you are eager to start mining and participating. Thanks for your patience as both the community and Hadron team grows.

In the coming weeks we'll be focused on the following:
1) Final audits of our ERC20 smart contract
2) Further growth of our community in Europe, Korea, and Japan
3) Inviting more people into the mining group

On behalf of the Hadron Team, thanks again for your support and enthusiasm."

https://hadron.cloud/ -> De azi, site-ul este functionabil!

Am minat si de pe telefon, dar nu prea fac asta regulat. Nu mananca cine stie ce memorie si baterie, dar daca il tii 2-3 ore se simte gen 40-50% baterie. Nu am testat sa vad cat net mananca, ca l-am lasat mereu pe wi-fi.
Ideea e ok, se bazeaza pe cantitate, adica sa mineze cat mai multa lume, decat sa fie putini cu rig-uri si placi video performante.
Vedem peste cateva luni ce si cum, ca momentan, proiectul e intr-o faza incipienta. E promitator, in orice caz.

Da. Stai linistit ca o sa mineze toata lumea in alpha testing, doar ca adauga oameni gradat, pentru a vedea cum se comporta platforma. Eu mi-am postat configuratia pe la pagina 26-27 si la limita am prins primul val de alpha testeri.

ideea in sine nu este rea, sunt deja programe prin care iti inchiriezi calculatorul pentru minerit in general. ei in loc sa mearga pe procent, au mers pe o moneda

dar la partea cu minatul de pe telefon, puteti sa incercati sa intrati pe coinhive si vedeti cum se comporta un telefon cand este pus la minat

Tu poti mina? Am postat cu ceva zile in urma pe topic-ul lor configuratia mea si era ok. Am si cont pe site-ul lor, dar momentan nu pot mina, imi apare un mesaj cu "Alpha release coming soon!".

Pana acum, hadron.cloud se comporta bine, avand in vedere ca mineaza in jur de 100-200 de persoane. Zilele acestea se da drumul la urmatorul wave, din care vor face parte mult mai multe persoane decat pana acum.
Aveti aici si whitepaper-ul: https://hadron.cloud/whitepaper.html (care momentan nu fura ochii)

Astazi se da drumul la minat. (https://bitcointalk.org/index.php?topic=2142232.1680)
Este un hype destul de mare cu proiectul asta, in mare parte si datorita amatorilor de "airdropuri", dar asta e mai putin important. In 3 zile de la 300 de oameni pe telegram, s-a ajuns la 6,3k. Pe reddit, la fel, deja ~1k subscriberi.

Mai sunt 3 zile pana la terminarea inscrierilor pentru alpha mining. De retinut, cei care participa la aceasta faza, au beneficii pe termen lung.

Va recomand un proiect care mi-a atras atentia. Inca nu se stie nimic despre pre sale, nr. de tokeni, dar nu asta e important. Cele mai mari plusuri sunt echipa din spate si ideea. Pe 15 Ianuarie dau drumul la alpha mining. Trebuie sa specificati, pe topicul din linkul de mai jos, ce specificatii aveti si vi se confirma acolo daca sunteti acceptati. Sunt si niste airdropuri frumusele.  
https://bitcointalksearch.org/topic/hadron-decentralizedai-ecosystem-2142232 - mai multe detalii aici.

Telegram: https://t.me/joinchat/GahFuELlfJoWE1vsFQf3VQ
Twitter: https://twitter.com/hadron_cloud
FB: https://www.facebook.com/HadronCloud/