Red Value Neural Engine V1.0 is finally here!

The Origin Story

Originally, I wanted to write you a short backstory on how this system came to be, but this “short” backstory ended up becoming longer than an astronomical unit. And yes, Astronomical Units (short AU) are not a made up name but an actual useful term in Astronomy (not Astrology) describing the distance between the Earth and the Sun. I will keep the backstory for another post and focus this one on the Neural Network itself.

How To Use

So how do you operate it? First, you need to register for the site or login with one of the provided social logins. If you didn’t notice, you are not able to log in with Facebook. That’s because they were trying to strong-arm me into making their button really huge and I’m not even kidding. I in turn maliciously complied by removing their login alltogether and nothing of value was lost. Once you registered, you can then go to a game’s page (search function is helpful here) or on the price recommender tool to vote on how much you would pay for the game. See it as a kind of review score. After you’ve voted on 10 games, the Neural Network will start to train and learn about you from the 10 votes you’ve given. This will usually take around 1 minute, maybe more if the site has a lot of load. The hamster is only able to run a limited amount of speed to keep the server going after all. And yes, I hired a hamster to keep the server going. It was the most cost-effective as well as the cutest way. Clearly the most logical choice.

And now, once the Neural Network has trained, you will find that the Price Recommender as well as the Game’s page now load price recommendations based on your previous votes. Then, you can either agree or disagree to the suggested price. Agreeing will instantly convert the suggested value into a vote. Disagreeing will open the voting interface in which you can then place your own value. And once you’ve done that, I will personally thank you for doing so. That is if you write me a small message, so I know who to say thanks to, through Twitter, Instagram, Facebook (yes even Facebook), Email or the Pidgeon Delivery Service. And yes, I literally mean physical and alive pidgeons by that. I’m genuinely greatful for you doing that as the more data I get, the more patterns will be found by the Neural Network. And that in turn will result in better and more accurate recommendations. Machine Learning is cool right? And so are buzzwords, only just noticed I didn’t use the word ‘Machine Learning’ in this post yet. Maybe I should mention ‘AI’ and ‘SkyNet’ too, but that might just scare some people off. Last, let’s go through some issues the system currently has.

The Issues

1) Slow? Slooow! Slooooooow!!!

Yes, this is Neural Engine 1.0 and not 2.0. The recommender page takes upwards of 10 seconds to load and the same with the inidividual game’s pages. This is due to the code not being fully optimized yet. Basically, the moment you enter the recommender, the website, coded in PHP, will launch a Python script which in turn will launch all the Tensorflow (The Neural Network) and other bloat accompanied by it. This will need to be done on every single time the recommender or game’s page is opened. You might be wondering: “Hey, I have an amazing idea!!!! Can’t you just open and launch the Neural Network running instead of launching and shutting it down all the time?”And the answer to that is “YES!!!”. It is possible to instead create an internal server that takes the recommender requests, calculates them and replies to them. Only issue is I need to figure out how to precisely code that. This should bring the loading down to acceptable levels, ideally with barely any difference to other websites. We’ll see about that soon.

2) Why am I getting so high values for cheap indie games?

Because indie games are totally the best and artsy and obviously worth donating a kidney for. Just kidding, indie games are fine games but all games are treated equally on this platform. Too equal actually as I don’t have a way to reliably fetch video game prices at this time. This is why all games default to an MSRP of $59.99. While I do change the price on some games manually, I have also decided not to use price as a parameter in the Neural Network because of this reason. This will hopefully change in the hopefully near future. And yes, I did notice that I wrote “hopefully” twice in the last sentence. It’s simply the best way to describe the current situation.

3) Where are my list of votes. Why is the Recommender only showing the latest games? Why did you not implement feature X, Y and Z yet?

Lack of time.

Future Changes

Notice I didn’t say anything about “Future Improvements”. Just “Future Changes”.

V 1.0

The features I described above

V 1.x

Smaller quality of life stuff. Think that’s how versioning is done. Never done this before.

V 2.0

The promised speed upgrade. Hopefully I can get this done by end of July 2019. Notice the use of the word “hopefully” here.

V 2.x

More quality of life stuff. Can’t think of a funny quip this time though.

V 3.0

A Metascore. No really. You know when people say ”This game got 92 on Metacritic, 93 on Opencritic and $1,000,000,000 on Red Value! It must be good”. Well, of course you don’t, because you most definitely haven’t heard that before. But that is definitely a point that I want to reach. I might also include functionality here that allows critics to put in their opinions somewhere. Further, a feature to allow for game recommendations, not just price recommendations, could also be developed here. I already have a pretty good idea on how.

V 3.x

No idea, my OCD just forced me to add this line here

V 4.0

Recommendation of critics with similar opinion to yours. Basically, the idea of this site is to provide users with personal recommendations, so why not also personell recommendations? I believe the best critic to follow is one with a similar opinion to yours and out of the brazillions of YouTubers and critics out there, it becomes a sheer impossible task to really figure out which critic is the best for you. So let the AI do that job for you.


This concludes this post. Maybe you’ve noticed that I don’t really review games anymore. I will be talking about that in one of the next posts on why that is. So instead, my mission is to create a useful and valueable site without needing to make any game reviews. Regarding this post, I might update it and add a couple of nice pictures. Maybe make it more professional by placing icons and stuff in random locations.

Finally, I hope you enjoy your stay on this site and if you like what is being done here, please feel free to share it. I am of the opinion that the best way to share something is to print a screenshot of the website (smartphone pic also ok) and then stick it to a lamp post. Not even kidding here, we usually just ignore all kinds of ads nowadays. But if something is glued to a lamp post, like a missing cat, we are way more compelled to read it! I most definitely do every time. So let’s do this! Stick Red Value Gaming to lamp posts and don’t tell any advertisers about this new method. I really don’t want to get tricked by lamp posts ads, the last safe space for actual relevant news.

Before I forget, a big shout-out to Giant Bomb and their users for providing their wiki and API with which this site and Neural Network are made possible. I will release details on how this works in the future so stay tuned.

— Niko aka Kaisernik aka The Red Value

Rate this Post:


Leave a Reply