A single-segment real-time speedrun of Papers, Please in 2:47:08 (in-game timer). Played in 2014-08-26.

Papers, Please is a puzzle video game developed by indie game developer Lucas Pope. It focuses on the emotional toll of working as an immigration officer, deciding whom to let in and whom to exclude from entering the fictional dystopian country of Arstotzka. The game was released on August 8, 2013, for Microsoft Windows and OS X, and was released for Linux on February 12, 2014.

I am playing the Linux version (v1.1.67).

Goals:

• The most glorious ending: Ending 19, member of EZIC, which requires completing four of the five EZIC tasks.
• All family members alive and well in the end (does not matter if they suffer some days during the run)
• Fastest possible under these conditions.

Incidental goals (which I did not aim for, but accomplished nonetheless):

• Obristan Token
• All Booth upgrades

Notes:

• There are nine days where a terrorist attack is scheduled to happen during the processing of a certain entrant. Day 2 is first of those. Those days, I deliver swift judgments. Right or wrong, just go quickly. This way the day can end quickly. Number of citations is not a factor in the goals of this speedrun; time is. I try to choose judgments that make statistical sense though.
• There's also seven days where an entrant can be a smuggler, who will turn into a suicide bomber if admitted. They are randomly generated after a certain number of entrants have been processed. Those days, I usually process regular entrants correctly, but if one would demand too much time, I deliver a random judgment. If I do admit a smuggler, they will explode spectacularly and the day will end short, which is a fantastic thing for speedrun. Luck is a very significant factor here. Because it is entirely possible that you can process 20 entrants without running into a single smuggler, I cannot afford to turbo-approve everyone; the heap of citations would grow prohibitively expensive, with a cumulative cost in hundreds of credits. (Getting 14 citations in a single day costs 335 credits in penalties + 70 credits in missed income, as seen on my Twitch channel.)
• Most days there are some special, predefined entrants, who always have correct papers or wrong papers. I can quickly stamp their passport without even looking at it. Example is day 3, where entrants 1, 3, 5 always have correct papers and entrants 2, 6, 8 always have incorrect papers. Entrants 4, 7, and 9 onward have random and unpredictable papers. Every day, after entrant 10 everyone is always completely random and unpredictable.
• In general, I choose my denial/approval options for special entrants based on the number of lines the entrant will speak now or later, to maximize the number of entrants I can process during that day. If I seem too heartless when dealing with some particular case, it's just because they talk too much.
• When the first entrant of the day is a guard, the boss, or the blue creeper, clock does not run during their visit. This means that the day will also end so much later. Sergiu is one of these special entrants, and I make sure to kill him as soon as possible, so he won't waste my time talking on any following days.
• Jorji always involves lots of talking, and hence lots of time spent waiting. I approve Jorji's obvious fake passport on day four, so he won't show up on day six. This means I can get the terrorist attack on day six faster, and day six will be shorter. He also won't show up on days 8 or 11.
• Because I don't give Corman Drex the paper on day 8, (s)he appears again on day 9. This is intentional. This way I can quickly approve entrant 9's papers on day 9 without studying them.
• On day 20, I poison the first possible entrant, to end the day in record time. I chose this to be the one EZIC task that I was allowed to fail and still get the glorious ending.
• Many days I close early. If you have an entrant in the office when the clock hits 6pm, you must work overtime. This is bad for a speedrun, so I try to make sure I'm always completely done at 6pm.
• Heating / feeding the family every day is not necessary. See http://papersplease.wikia.com/wiki/Bills#Details for my detailed analysis. You can save average 15―25 credits per day by alternating.

End stats: 485 entrants, 255 approvals, 207 denials, 12 detains, 104 citations, 672 stamps, 405 credits, 1 token, 20 naked photos, 48 onscreen deaths.

The in-game reckoning of the gameplay time (2:47:08) matches exactly the timing counted manually from calling entrant #1 in on day 1 (first possible control action after intro cutscene) until the Sleep button is hit on day 31 (last possible control action before ending cutscene).

Possible future improvements, in order of plausibility:

• Forget about saving credits by alternating with bills. Just click sleep as soon as the day is over. You can overcome the money difference easily by doing your job well. There are only a few days where you must click some items on the budget screen (such as days 11 & 12 where you must burn the EZIC gift, and any day when you wish to upgrade the booth). This is the single largest consistent source of potential improvement. Note that when you do wish to upgrade the booth, you better not do it after a terrorist day or on a day where a house upgrade becomes available, or the budget screen will cost more time. For entertainment reasons, I recommend not turbo-clicking Sleep on the final day, so people can actually see the state of the budget in the of the game.
• Close the booth earlier, maybe 17:15 each day. The day will still run until 18:00, but you don't want to risk running overtime because of a last-minute citation.
• Better control of the quick reply trick. When your action triggers dialog ― such as returning the last document to the entrant, or pointing out a discrepancy ― if the entrant is still talking, the new dialog will be queued and rendered after a standard between-dialog delay (2-3 seconds). However, if you do it right as the last bit of dialog is getting rendered, the new dialog will be performed immediately without delay, as if the two lines are spoken at the same time. This saves a few seconds each time, very important on terrorist days.
• Acknowledge Corman Drex on day 8 instead of day 9. Dealing with a random entrant on day 9 is probably still faster than dealing with Corman's dialog, and day 9 is a smuggler day.
• Record on a faster computer. Oh, yes. This game does not compensate lag. If you lag, the day will last longer. This may account for a few seconds in my run.
• Play more risky on smuggler days. You can afford a bit more citations than I was getting.
• RNG. So much RNG. The immigration lottery. The random number generator of doom. The difference between a no-smuggler day and a good smuggler day can be as much as five minutes, and there are seven of those days. The difference between a diplomat whose papers are all approvable right away, and a worker who fails to produce a passport, is about ten seconds, per entrant; utmostly important on terrorist-attack days. There's absolutely nothing you can do about who walks in, aside from the choice between a scripted entrant and a randomized entrant in some cases. My sum of best splits is many minutes faster than this run. If you grind the 2-3 hour run 100 times, chances are you will get a better time than this run at least once, assuming you don't fumble one of the EZIC tasks. After you've got the strategy down, and you have enough practice that you can actually verify the papers quickly enough, it all comes down to how much effort you wish to spend appeasing the RNG.

Runs of other categories for this game that I would like to see people attempt:

• Any% Escape to Obristan ending (expected to be several minutes faster, because you can escape on an earlier day), equally challenging as this ending (note: escape alone and escape with all family members are different categories).
• 99.8% Glory to Arstotzka ending (no citations, except the one that is forced on you by Jorji near the end of game).