constantcompile » Ulasan » while True: learn()
Lihat Halaman Toko
Lihat Hub Komunitas
Cari Produk Serupa
4 orang menganggap ulasan ini membantu
Tidak Direkomendasikan
0.0 jam dalam 2 minggu terakhir / Tercatat 7.8 jam (Telah dimainkan 7.7 jam saat ulasan ditulis)
Diposting: 27 Jun 2020 @ 6:40pm

Left me scratching my head - and not in a good way.

It's difficult to review, let alone compare, systems-engineering games. Most Zachtronics games can run well into the double- or triple-digits, and this game inhabits a similar genre. That said, even a mere ~7.5 hours into this game, I'm baffled by the design choices that were made for the gameplay. Here's just a few reasons why:
  • The difficulty curve is a nightmarish rollercoaster. Some of the most finicky levels I encountered were followed by single-step puzzles with a solution I'd previously learned.
  • You're given tools that - so far, in my playthrough - you're very rarely incentivized to use, because they will always be far slower than the simpler functions.
  • Lore-wise, you're given cutesy flavor text about programming self-driving cars for sentient cats in one level, then Orwellian descriptions of political imprisonment in the next. It's all over the place.
  • No matter the data you're sorting in-universe, though, 99% of the time it shows up on the UI as red, green, and blue shapes - so why even bother?
  • Why are performance upgrades even a thing? It raises doubts about whether gold medals can be earned on the first playthrough of the level with clever building, or whether the only choice is to return after upgrading.
  • If performance upgrades ARE a thing, why also tie exorbitant amounts of money for cutesy decorative tat?
  • Why are startups essentially "golden parachute simulator?" It isn't fulfilling to try to make perfect code the first time, discover the company is failing anyway, then try to bail before it crashes. Unlike every other level, you're not allowed to try again once the company folds, unless you start a new save file (but then you have to do all the other levels over again).
  • Why are Schemes faster? Why are DLLs faster? Why?
I could go on. I want to like this game, and I admire the aspirations to educate that no doubt inspire it, but as a product - as a videogame - I'm not having very much fun so far. Looking at the reviews, most people also seem to hit a dozen hours or less before moving on.

I would implore the developers to get really good play-testing of this game, start-to-finish, asking at each point whether things are flowing smoothly. It's been rough going for me so far.

Not recommended.
Apakah ulasan ini membantu? Ya Tidak Lucu Penghargaan
Tanggapan pengembang:
ElJorro  [pengembang] Diposting: 7 Jul 2020 @ 2:46am
Hi constantcompile,

And thanks for your very deliberate feedback!

Making a fun game about machine learning was definitely a challenge and if you ask us, it did work for many people: actually, seeing people spend 12 hours on a game like that is inspiring to us :)

Some of the problems that you’ve listed on your review are well known to us, but we’ll have to correct them in a sequel if and when it happens: things are tied to each other in a way they can’t be dealt with without breaking the game.

Some of the issues are there by design: for example, we’ve tried making data representation more graphical, but in the end we’ve decided to stick with shapes and colours. ML specialists are working with data and it’s rarely spectacular. The idea was that almost everything can be turned into data. It’s like in The Matrix movie: if you keep doing this long enough, you’ll start seeing actual images in that data. Or, at least, that was our idea

My bottom line is, it’s a shame that the game didn’t feel quite right for you. But we do thank you for spending time writing a review. The point of our game is that one should always keep learning and that’s exactly what we shall do, thanks to feedback from our players!