Never Sink
patreon
Broken Bones Automation [part 1]
🕑 Added 2019-02-12 13:13:31 +0000 UTCAs some of you might know, I recently (~1.5 months ago) had an accident in which I broke both of my elbows.
At this point most of the damage is in the past, my right arm required an operation and I've received permanent 2 titan screws (it's an upgrade!) and my recovery is going steadily. I work, program and even do some sports again.
I was lying in the hospital bed and for once I had time to think, instead of just thinking all the time.
One of the topics I've put some thoughts into: I was not able to update the filters. I don't plan on breaking arms again, but life is striped. Things happen. Accidents, stress phases, illness etc.
And even if all of that won't happen - the game has become absurdly complex. The Shaper/Elder tierlist alone needs over 6000 cases to consider. Uniques have ~600 items, each with sevefral drop variations, price development etc.
Hand-made updates have become costly. It takes a ton of my "filter-work-time", that I'd rather use to design new features, instead of doing repetitive tasks.
Also: how much concentration can you invest into a repetitive task before you start "slipping" and have a chance to make errors?
In short I have decided that updating tierlists and repetitive tasks (which is about 75% of the work in mid-league updates) have become too costly, unreliable, cumbersome and also problematic.
However, a big part of PoE is knowledge based. Prices change depending on the league state, item type, streamer influence and tons of others factors and I believe one of the reasons my filter received it's popularity is because I know the game well enough.
So is it impossible/unrealistic to design a system that could automate such tasks?
After some thoughts - I believe it's actually very realistic to serialize game knowledge and create an automated pipeline.
I've done some architecture charts 1 month ago and started coding. I'm currently investing a lot of time to create this new infrastructure.
This is currently my #1 priority project and I hope to finish the prototype until the start of the new league.
What are the benefits?
- The filter will receive automated tier updates (daily)
- The filter will hopefully be more accurate overall
- Potential Hardcore filter versions
- Less chances of me screwing up
- The ability to generate "economy news"
- I'll spend less time on repetitive tasks and more time on new features.
- The serialized game knowledge can be used for other projects
How exactly is this going to work?
Find out in the next post :)
As always - thank you guys for your support!
Comments
Kataris
Wow, Neversink. You never cease to amaze with your dedication. Thanks for all your hard work, and this automation update sounds awesome.
