The Workshop Never Sleeps: Designing Artisan Mastery
In Valenya, forging takes real time. We wanted artisans to invest in producing more without letting wealth replace patience. Here are the three levers we chose, and what we had to repair before introducing them.
Valenya's crafting rests on a wager: making something costs real time. An ingot, twenty minutes. A high-tier cuirass, ten hours. Meanwhile, the player does something else: farms, sells, logs off if they like. The workshop queue advances on its own.
That wager has an awkward consequence. A player logging in twice a day had three twenty-minute slots to start, one hour of production per visit. The rest of the time, their workstations sat empty. The workshop rewarded availability rather than dedication, which is a very different thing.
Hence the question that opened this release: how do we let an artisan invest to produce more?
Three levers, and why these three
The answer is three mastery paths, learned from the profession master, with four ranks each.
- Lines add a workstation that runs alongside the others.
- Batches put more copies into a single production run.
- Pace shortens each crafting job by 5% per rank.
Each serves a different player. Lines help those juggling several projects. Batches suit volume production for the Auction House. Pace serves those who want a particular piece as quickly as possible. Since you cannot advance all three at once, there is a real choice.
Training is paid for and learned
This is the decision we are happiest with. It also took the most discussion.
Training costs gold: from 5,000 coins for the first rank to 320,000 for the last. Had we stopped there, we would have built a tollgate. A rich player could buy all twelve ranks in one evening, and the market would face a multiplied production rate overnight. In a player economy, that cannot be undone.
Training therefore also costs real time: from six hours to two days of learning, with only one course at a time across all paths. Gold acts as a sink, removing coins from the economy. Time guarantees that no fortune can compress the curve.
Learning progresses offline, of course. Technically, it uses the same principle as the workshop queue: no timer runs on the server. We record an end date and detect completion at the next opportunity with a simple timestamp comparison. A server restart loses nothing. A player returning after three days finds their training finished.
Five chest pieces, five quality rolls
Ordering five copies at once tempted us toward the wrong implementation: roll quality once and duplicate it. Simpler to write, but it would have drained the system of its interest.
Each item in a batch rolls its own quality and affixes. Batching saves trips, nothing else. Among five chest pieces there may be a failure and perhaps an exceptional piece, just as if they had been forged individually.
A less visible but significant detail: materials are checked together before anything is removed. Checking along the way could have consumed the first three copies' materials, then failed on the fourth, leaving an inventory state nobody asked for.
Two workstations running together
The queue had three slots, but only one project progressed while the other two waited. It now has five slots, with two workstations progressing in parallel.
The real work was elsewhere. Offline catch-up hides two pitfalls you never see while playing online, because they appear only after a long absence:
- The earliest overdue completion must be processed first. It frees its line first, allowing the next project to start.
- A freed line must resume at the exact time the previous project ended, not when the player reconnects.
Without those rules, a player away for three days would return to one finished item instead of three. The system would seem to work for everyone except those using it most.
What had to be repaired first
Half this release is missing from the list above. Before adding those three levers, we had to repair high-level crafting that did not hold together.
The production chain stopped at level 9
The highest refined material in the game required level 9. All equipment from tiers 24 to 80 used the same steel ingot, a level 3 recipe. A level 80 chest piece was made of the same metal as a level 24 one.
Nobody had complained, for good reason: no crafting loop ran above level 16, and three gathering professions had nothing left to refine after level 9.
Each profession now has its own tier ladder: ingots for miners, leather and silk for trappers, planks for carpenters, from silver to Origin. Each step is refined from the previous one, so an Origin ingot contains the entire ladder beneath it. The silk weaving agent comes from the cook. This is the only ladder that requires two professions to work together.
An Origin harness took three minutes to forge
High-tier equipment batches had been generated later and left with demonstration values. The result: an Origin harness in 3 minutes, against 7 hours for its dragon equivalent. A hundred and forty times faster, for forty more levels.
Worse, those recipes required no workstation. Field crafting immobilizes the character and cancels on disconnect: increasing durations without fixing that would have made recipes unplayable, not merely long. Anything over ten minutes now uses a workbench.
A level 1 recipe remained optimal until level 59
Workshop experience was a flat 25 points regardless of recipe. A ten-hour cuirass earned as much as a twenty-minute ingot. The shortest recipe was therefore always best for progression, and level 1 leather leggings, crafted in five seconds, offered the game's best efficiency across nearly the entire curve.
Two rules fix this. Experience now follows the recipe's real cost. And a recipe far below your level crafts faster but teaches less. Midrange goods produced in volume are no longer penalized, while repeating a trivial recipe earns little.
Duration calculation was duplicated
Field and workshop crafting each calculated their own duration. With Pace joining specialization and recipe mastery, we would eventually have had two different durations for the same recipe. There is now just one calculation, and the recipe description finally shows exactly the duration used by the progress bar.
What we learned
An investment system in a player-driven economy is judged by what it does to the market two weeks later, not just what it gives its buyer. Hence learning time: it is the one brake money cannot buy.
And a lesson that returns every release: when a feature looks easy to add, the underlying layer often has not been pushed far enough to reveal its gaps.