You Said, We Did: Ten Fixes From Player Reports

Most of this release exists because a player told me something was wrong. Two of you worked out that async duels rewarded the weaker fighter. A Security+ learner showed me a wrong answer lighting up two options in red. An instructor found a final boss that never unlocked the next one. Below is each report, what was broken, and what changed. The short version lives in the changelog.

Duels rewarded weak gear

Two players reported this on September 17, minutes apart. One of them put it better than I could:

You can really cheese the system by wearing no gear and doing as little damage as possible as long as you can defeat the opponent within the 20 strikes limit.

They were right, and it wasn’t an edge case. In an async duel you fight a stand-in enemy on 20 questions, your opponent fights one on the same 20, and the server compares the two runs. Each fight ends the moment the enemy falls. The comparison was how many questions you got right, times ten. A better weapon, more Precision or a lucky crit ended your fight sooner and left you fewer questions to score on. Taking your gear off made you win more often.

I checked how much that mattered. In the 120 days before the fix, 228 duels were settled. In 221 of them the attacker finished alive with fewer than 20 correct answers (the median was 8), so the fight stopped on a kill long before the last question. The broken rule shaped nearly every duel.

Here is the same learner on the same questions, once without a weapon and once with the Rootkit Scythe (+0.6 attack): level 20, eight seconds per answer, one wrong answer. Watch the last two rows.

A duel is now decided by the average damage you deal per question you faced. The server works it out from your graded answers with the damage formula the battle itself uses: your weapon’s attack, your crit and heavy-hit chances from Precision and answer speed, and a speed multiplier. Every point still comes from a correct answer. A wrong answer or a timeout deals nothing.

• It is a rate, so a longer fight can’t raise it and a quick kill can’t lower it.
• Crits count at their average chance, not the roll you happened to get. The roll happens on your device, where the server can’t check it. The average it can.
• Answer time comes from the server’s clock, with a second and a half of slack for your connection.
• Quitting on a hot streak doesn’t pay. The server counts at least as many questions as the fastest possible kill, so the ones you skipped count as zero.

Ties changed too. Of those 228 duels, 44 ended level, and every one went to the attacker. An exact tie now goes to whoever answered faster.

One trade I made on purpose: gear matters now, sometimes a lot. With a top weapon, getting 6 of 10 right can edge out 9 of 10 with no weapon at all. With ordinary gear, accuracy still wins. Gear is what you earn by studying, so I think it should count in a fight. If the balance feels wrong in real matches, tell me. After a duel settles, the Battle Log shows both scores at the top of the round-by-round exchange.

Live Arena was never affected. There, both players answer the same question at the same moment, so a harder hit really does end the fight in your favor.

Two answers flashed red

A Security+ learner sent this the day after signing up:

Both the correct and incorrect answers flash red. … I initially assumed it meant the question had multiple correct answers, and only later realized it was simply how the system was designed.

When you picked a wrong answer, the exam painted your pick red and the correct answer a slightly warmer red-orange, both as a light tint, for one second before moving on. On a phone they looked the same. The label that should have said which was which existed in the code and never rendered.

Now your pick is red and says Your answer, incorrect, the correct one is green and says Correct answer, and the other options fade back. A wrong answer also stays on screen until you move on, so you can see which answer was right. Correct answers still move on after a second. While I was in there I also moved the verdict out of a pop-up that covered the exam header.

The final boss that never unlocked

An instructor who teaches CompTIA certifications wrote:

When I beat the first final Exam Boss on Sec+, it doesn’t let me move on to the next one.

The win was saved, but the campaign map never read final-boss results back, so the second boss stayed locked on every path. When I checked on September 3, 23 players had beaten the first final boss and none of them had a single recorded fight against the second. It’s fixed, and each of the gauntlet’s three stages now has its own boss instead of the same one three times.

Reading comfort, and a VM that crawled

One learner asked for this:

It would be helpful to have either a dyslexic friendly font option or additional ways to increase/stop the timer to give ample time to read the questions.

The timer half shipped on September 15. Pausing a practice exam stops the clock, you can hide the timer, and practice exams never count down or submit on their own. The font half is new: Settings → Reading comfort lets you pick Atkinson Hyperlegible or OpenDyslexic, larger text, and wider line and letter spacing. OpenDyslexic is there because a learner asked for it. A study of children with dyslexia in Annals of Dyslexia found no improvement in reading rate or accuracy with it, so try it and keep it only if it feels better to you.

A member told me SecuSpark was smooth on a home PC and laptop but, on a work virtual machine, “veeeeery choppy.” Most virtual machines have no graphics acceleration, so the browser draws every blur, glow and animated background on the processor. Settings → Visual effects now defaults to Auto, which checks for that and turns the expensive effects down. On my test machine with graphics acceleration switched off, the home screen went from under 30 frames per second to a steady 120. You can force Full or Reduced.

Flashcards at your pace

Two reports were about flashcards. One learner found that new cards dried up until the next day and asked for an adjustable break. Settings → Study Preferences now has New cards per day and a Card break period of one day, four hours or thirty minutes.

Both of them saw True or False buttons under questions that weren’t True or False. That was 188 questions across Security+, A+, Network+ and CySA+ (78 of them Security+) whose card fronts had been reworded into open questions while the answer buttons still followed the original. Those cards don’t show the buttons anymore.

Smaller fixes you reported

• Review Mistakes Only jumped back to question 1 after your second answer. That story has its own post.
• A learner with the 3-day premium reward was still asked to upgrade when picking a certification. Earned and gifted premium now counts as premium at every gate. That fix went out at the end of July.
• Cosmetics didn’t scale with smaller characters. Every item has had a fit for every body since August (here’s how), and in this release flipped helmets tilt the way they were drawn.

Keep them coming

Most of the reports in this post came through the feedback button, which comes straight to me. I read all of them, the ones that say something is broken most of all. Detailed feedback earns 3 days of premium, once per account, whether the news is good or bad.

How this was made: I build SecuSpark with AI coding tools and drafted this post with one. The duel and final-boss numbers come from SecuSpark’s production database, and the worked example comes from the scoring code the server runs.