HandmadeCon 2015 Q&A (Part 2) — 2016 — Episode Guide

0:00Casey Muratori: Welcome back

0:31Research papers often have a math-wall. How do you get past this barrier? Assuming you've got there, how do you find and evaluate papers?

🗪

0:31Research papers often have a math-wall. How do you get past this barrier? Assuming you've got there, how do you find and evaluate papers?

🗪

0:31Research papers often have a math-wall. How do you get past this barrier? Assuming you've got there, how do you find and evaluate papers?

🗪

2:00Tommy Refenes: Finding unfamiliar symbols on Wikipedia

2:43Pat Wyatt: Drawing on expertise of biostatistician wife, and slowing down when reading papers

3:27Jonathan Blow: No math-wall for notation, but for concepts

6:31CM: Getting pointers to useful stuff by word-of-mouth

6:48CM: Often either the math is so obvious, or redoing it is the way to go

8:42How do you find research?

🗪

8:42How do you find research?

🗪

8:42How do you find research?

🗪

9:18JB: Web search for, e.g. "interpolation of unstructured control points" for The Witness, and learning the subject space

10:41PW: Gauging the author's perspective from the conclusion

11:05CM: Going through known publications – e.g. SIGGRAPH – and reference-crawling bibliographies

13:34JB: I don't know what I don't know, e.g. Ken Shoemake's quaternion paper¹

14:31CM: Pulling in mathematical concepts from other fields than programming, and further seeing what other practical problems they may apply to

15:32PW: We're still waiting for your paper on using Taylor series expansion for estimating the division coordinate for quaternions

15:46CM: I remember this incident, but I never used the Taylor series expansion for it

15:55PW: It was an estimation of the division term

16:01CM: A simple first-order approximation to renormalise a vector whose length was close to 1

16:54PW: Spending a month trying to optimise quaternions for Guild Wars' animation system, before Casey came along with the solution

17:28CM: Renormalising vectors all the time

17:52Ron Gilbert: I have no idea what you guys are talking about

18:00JB: It's what lets the hand go on the doorknob

18:10CM: If it hadn't been over a decade, I could re-derive it

18:41How do you evaluate a research paper?

🗪

18:41How do you evaluate a research paper?

🗪

18:41How do you evaluate a research paper?

🗪

18:58JB: It's a steep curve with really good papers clustered at the top

19:30CM: You were saying to look at the conclusion

19:45PW: Look at the abstract, then the conclusion to see the results, and see if that warrants reading the rest

19:53CM: So the middle gets paged in if the two ends look good?

19:57JB: Some authors are reliable, e.g. Kahan

20:16CM: Maybe we don't use research as much as we think we do

20:25JB: We used to research more in the olden days

20:59ApoorvaJ When creating an engine or shipping a game, what do you look for in a file format for asset data? How do you balance compression vs speed vs readability? What is the best file format you've worked with, and why?

🗪

20:59ApoorvaJ When creating an engine or shipping a game, what do you look for in a file format for asset data? How do you balance compression vs speed vs readability? What is the best file format you've worked with, and why?

🗪

20:59ApoorvaJ When creating an engine or shipping a game, what do you look for in a file format for asset data? How do you balance compression vs speed vs readability? What is the best file format you've worked with, and why?

🗪

21:46RG: Gravitating towards ASCII formats, having gotten bitten by endianness

22:35CM: So your room layout tool will save out to a text description?

22:49RG: Speedy text manipulation vs GUI tools

23:10PW: Guild Wars's model exporter, storing unprocessed source data in addition to the compressed model

24:14CM: And that saved you having to export again?

24:27PW: Gaspar Games automated 3DS Max export

24:46CM: Your system also protected against 3DS Max upgrades

25:11TR: ASCII for config files, and chunked file format

25:51CM: And the difference with JSON is that you don't want to have to start parsing things?

25:57TR: Using what's documented

26:37JB: Different file formats for release and for tools pipeline, e.g. entities in The Witness

29:55CM: So you're saying, whatever you look for in a file format is what makes the systems on either side of it work well?

30:15JB: Contextual file format, text-only for small amounts of data

31:48CM: How do you know if it's been edited?

32:07JB: printf() in Windows is worse than 1-bit error for floats

32:14CM: How do you know the reason why the hex and float don't agree?

32:27JB: Tactics in file format, omitting container IDs to play nicely with version control

34:27CM: Still trying to nail down the reason why the hex and float don't agree

35:16RG: When the user edits the floating point value, you delete the hex value

35:24JB: Maybe also measure the magnitude of the difference

35:38CM: So the user must remember to delete the hex value?

35:49JB: Entities are rarely changed, and variables are read-only

36:31CM: Storing a variable length list, with items recording their parent

37:52JB: Marker entity type in The Witness

38:52occamin If you could go back in time and give yourself only one piece of advice at the beginning of your career, what would it be?

🗪

38:52occamin If you could go back in time and give yourself only one piece of advice at the beginning of your career, what would it be?

🗪

38:52occamin If you could go back in time and give yourself only one piece of advice at the beginning of your career, what would it be?

🗪

39:16CM: Deep thoughts

39:26PW: If anybody tells to anything related to finance or legal, get a lawyer to review it

39:43TR: Maybe spend the money on lawyers

39:58RG: I think I would have told myself to be a lawyer

40:07CM: So 50% of the panel says, "talk to a lawyer"

40:31RG: Don't be afraid to fail

41:12CM: Do this come from not having experienced success?

41:48RG: Watching people we have respect for fail

42:33CM: Internalising the fact of other programmers making mistakes

43:15JB: Don't reason yourself into a corner, but trust your observation that things are wrong

45:30CM: From Jeff: Actually finish pieces of code before moving on

48:29JB: Avoid piling up shortcuts that you intend to go back and refine

50:28CM: Distinguish between "adding stuff to the system" and "making the system good"

51:10JB: Implementing systems later than in Handmade Hero

52:44CM: Getting confused with new systems

53:14JB: Context-appropriate decisions

54:12CM: I don't understand the punctuation

54:24JB: When somehow asks you how you design a game, you have to plug them into the matrix

54:48CM: So you have a complex set of changing, internal trade-offs?

55:18JB: I don't do that too often, but I have done it

55:21PW: Step 1: Write a Jon Blow simulator and run your problem through it

55:24JB: Chris Butcher's asset system is the technical version of this

56:13CM: We have fifteen minutes left

56:18JB: I'm gonna not talk for the rest of the fifteen minutes

56:24ClamorousVole Regarding the prevention of cheating, how much of a difference does encrypting gameplay network traffic realistically make? I often hear it is important, but it seems to me it would only catch the lowest hanging fruit

🗪

56:24ClamorousVole Regarding the prevention of cheating, how much of a difference does encrypting gameplay network traffic realistically make? I often hear it is important, but it seems to me it would only catch the lowest hanging fruit

🗪

56:24ClamorousVole Regarding the prevention of cheating, how much of a difference does encrypting gameplay network traffic realistically make? I often hear it is important, but it seems to me it would only catch the lowest hanging fruit

🗪

56:48PW: Encryption is used to prevent personal data being stolen. It can be useful as an anti-cheat mechanism on consoles, but doesn't save you at all on PCs

57:51CM: So encryption is for keeping user-known data from being snooped?

58:00PW: That's right

58:01CaptainKraft When debugging your code, are there obviously shortcomings in the debuggers you use? Are there any obvious things that debuggers should do to make debugging easier?

🗪

58:01CaptainKraft When debugging your code, are there obviously shortcomings in the debuggers you use? Are there any obvious things that debuggers should do to make debugging easier?

🗪

58:01CaptainKraft When debugging your code, are there obviously shortcomings in the debuggers you use? Are there any obvious things that debuggers should do to make debugging easier?

🗪

58:58PW: Watcom C/C++ 10.0² had breakpoint scripting

59:31RG: Trying to track down a bug that only happens when the game is running in fullscreen mode

59:51CM: So you don't use a secondary monitor?

59:54RG: I do not, no

59:59CM: Are you café-only programming?

1:00:03PW: EGA monitors, you can't dual monitor them

1:00:09CM: So literally some way to overlay the text on the screen would solve this problem?

1:00:17RG: It's difficult to debug on a laptop when in fullscreen mode

1:00:25CM: I guess I'll say Tommy, not Jon here

1:00:31TR: Creating data breakpoints by right-clicking in the watch window and clicking "Watch"

1:00:52Yeah!

🗪

1:00:52Yeah!

🗪

1:00:52Yeah!

🗪

1:00:55TR: Why isn't that a thing?

1:01:11JB: You used to be able to drag a breakpoint from line to line

1:01:42Oswald_Hurlem What are the costs / benefits of shipping a mod API or level editor with a game?

🗪

1:01:42Oswald_Hurlem What are the costs / benefits of shipping a mod API or level editor with a game?

🗪

1:01:42Oswald_Hurlem What are the costs / benefits of shipping a mod API or level editor with a game?

🗪

1:02:07PW: Other companies benefit hugely, but Warcraft III also spawned an extended life cycle

1:03:19CM: It seems like a lot of the value seems to derive from the ability to change the actual game

1:04:21PW: Warcraft II with its level editor didn't spawn any derivative games, whereas Warcraft III with its modding capability did

1:04:37CM: Tommy, do you have anything?

1:04:41TR: Mods vs level editor is based on the context of the game, and Super Meat Boy's level editor works well for the game

1:06:11CM: What about the cost part of shipping a level editor?

1:06:50TR: Wanting to have a break from developing the game after launch

1:07:58How do you know what it is to be at different levels of productivity?

🗪

1:07:58How do you know what it is to be at different levels of productivity?

🗪

1:07:58How do you know what it is to be at different levels of productivity?

🗪

1:09:12RG: The number of times per hour that I open up twitter

1:09:21TR: Three times an hour is a good day

1:09:26CM: What is a good number?

1:09:37RG: Logging out of twitter, and using two-factor authentication as a productivity measure

1:10:20TR: And the same way, if you're not feeling it, you can continue to stay logged it

1:10:36CM: Would you say that we could get productivity gains by moving to three-factor authentication?

1:10:52PW: Checking emails and searching the internet for solutions are signs of burnout

1:11:21CM: So searching for other people's answers is a sign?

1:11:38What old school books did you like back when people still read books about programming?

🗪

1:11:38What old school books did you like back when people still read books about programming?

🗪

1:11:38What old school books did you like back when people still read books about programming?

🗪

1:12:09RG: The Commodore 64 manual³

1:12:47PW: Liking an AI programming book in college; but not liking K&R⁴

1:13:34CM: So the K&R C book⁵ was too language-specific?

1:13:48TR: A QBasic book, for a Panasonic computer

1:14:22CM: Jon?

1:14:35JB: Getting into programming using books for the RadioShack TRS-80 Color Computer⁶

1:15:03CM: Thanks so much for coming back, everyone

Keyboard Navigation

Global Keys

Menu toggling

In-Menu and Index Controls

Quotes and References Menus and Index

Quotes, References and Credits Menus

Filter Menu

Filter and Link Menus

Credits Menu