第３回予測市場勉強会資料・Googleにおける社内予測市場

2019/09/09
第３回予測市場勉強会
「Googleにおける社内予測市場」
https://eagna.io/

Transcript

1. Googleʹ͓͚Δࣾ಺༧ଌࢢ৔ ༧ଌࢢ৔ษڧձୈ̏ճ ݹᖒ ༏໵ 2019/09/09
2. Table of Contents • ࣗݾ঺հ • ࣾ಺༧ଌࢢ৔ʹ͍ͭͯ • Googleʹ͓͚Δ༧ଌࢢ৔ • ͓ΘΓʹ • Q&A
3. ࣗݾ঺հ • ݹᖒ ༏໵ • ౦େܦࡁM2 • ઐ໳ɿήʔϜཧ࿦ɺωοτϫʔΫཧ࿦ • ؔ৺ɿ༧ଌࢢ৔ɺ҉߸௨՟ɺҼՌਪ࿦ • ༧ଌࢢ৔αʔϏε”Eagna”ΛӡӦɾ։ൃͯ͠·͢
4. ࣾ಺༧ଌࢢ৔ʹ͍ͭͯ
5. ༧ଌࢢ৔ • ʮ܈ऺͷӥஐʯͱʮࢢ৔ϝΧχζϜʯΛ༻͍ ͨɺকདྷͷग़དྷࣄΛ༧૝͢ΔͨΊͷࢢ৔ͷΑ ͏ͳ΋ͷ • ਖ਼֬ͳ༧ଌɺϦΞϧλΠϜͷ༧ଌ͕Մೳ • ৄ͘͠͸ୈ̍ճͷεϥΠυΛࢀর
6. ࣾ಺༧ଌࢢ৔ • ձࣾʹͱͬͯେࣄͳग़དྷࣄΛ༧૝͢ΔͨΊʹ ձࣾ಺ʹઃஔ͞Εͨ༧ଌࢢ৔ • ࢀՃऀ͸ࣾһ͓Αͼؔ܎ऀʹݶΒΕΔ • ϧʔϧ΍ใुͳͲ͸ձ͕ܾࣾΊΔ • ࣾ಺༧ଌࢢ৔ઐ༻ͷιϑτ΢ΣΞΛൢചͯ͠ ͍Δձࣾ΋͋Γ·͢(Inklingࣾ)
7. ࣾ಺༧ଌࢢ৔͍Ζ͍Ζ • Google • Ford • HP • Microsoft • ͳͲͳͲɺɺɺ
8. ࣾ಺༧ଌࢢ৔ಛ༗ͷ໰୊ • τϨʔμʔͷগͳ͞ʢThin Market Problemʣ • औҾ૬ख͕ݟ͔ͭΒͳ͍Մೳੑ • ࢀՃऀ͕ݶΒΕ͍ͯΔ • ৘ใͷଟ༷ੑ͕ࣦΘΕΔՄೳੑ • ࣾ಺༧ଌࢢ৔͸͜ΕΒͷ໰୊Λ๊͑ͳ͕Βਫ਼ ౓ͷߴ͍༧ଌ͕Ͱ͖Δ͔ʁʁ
9. Googleʹ͓͚Δ༧ଌࢢ৔
10. Google Prediction Market • 2005೥ɺBo CowgillΒʹΑͬͯ։࢝ • ܦࡁֶऀHal Varianͷαϙʔτ • ͦͷ༧ଌͷਖ਼֬ੑͳͲ͔Βଞͷاۀ΋ࣾ಺༧ ଌࢢ৔Λͭ͘ΔΑ͏ʹ • ༧ଌࢢ৔͕޿͘஌ΒΕΔΑ͏ʹͳͬͨܖػ
11. Why Google? • GoogleͷΧϧνϟʔ • ʮۈ຿࣌ؒͷ20ˋΛࣗ෼ͷ޷͖ͳϓϩδΣΫ τʹ౰ͯͯ΋ྑ͍ʯ • ྫ: GmailɺGoogle News
12. GPMͷ໨త • “Objectives and Key Goals(OKR)”ͱݺ͹ΕΔ ଌఆՄೳͳࣾ಺ͷॏཁࣄ߲ʹ͍ͭͯͷ৘ใΛ ू໿͢Δ • ࣾ಺ͷOKRͷ͏ͪɺ͓Αͦ60%ΛΧόʔͯ͠ ͍ͨ
13. Types of Markets • Demand Forecasting • ྫɿࠓ࢛൒ظͷGmailͷϢʔβʔొ࿥਺͸͍ ͘Β͔ʁ • Company News • ྫɿGoogleͷϞεΫϫΦϑΟε͸Φʔϓϯ ͢Δ͔ʁ
14. Types of Markets • Industry News • ྫɿApple͸IntelϕʔεͷMacΛൃച͢Δ͔ʁ • Fun • ྫɿNBAϑΝΠφϧͷ༏উνʔϜ͸ʁ
15. GPMͷ࢓૊Έ • μϒϧΦʔΫγϣϯํࣜͰചങ • `` Gooble”ͱݺ͹ΕΔαʔϏε಺௨՟Λ༻͍ͯऔ ҾΛ͓͜ͳ͏ • ̏ϲ݄ʹҰ౓ɺ10,000GoobleΛड͚औΔ • Gooble͸͘͡ͷνέοτͱަ׵͞ΕΔʢޙड़ʣ
16. GPMͷ࢓૊Έ • GoogleࣾһͳΒ୭Ͱ΋ࢀՃ͢Δ͜ͱ͕Ͱ͖Δ • ͋͑ͯࢀՃऀʹ੍ݶΛઃ͚ͳ͍ʢΠϯαΠ μʔΛڐ͢ʣ • ࣾһ͕͍࣋ͬͯΔ৘ใΛޮ཰తʹϚʔέοτ ʹ൓өͤ͞Δ
17. ใु • ࠷ऴతʹอ༗͍ͯ͠ΔGooble͸͘͡ͷνέοτʹ ม׵͞ΕΔ • ӡӦνʔϜ͕̒͘͡ݸΛϥϯμϜʹબͿ • બ͹Εͨνέοτ͸$1000ͱަ׵͞ΕΔ • GoobleΛগ͠Ͱ΋૿΍ͦ͏ͱ͢ΔΠϯηϯςΟϒ
18. Tγϟπใु • ۚમతใु͸͋Μ·Γັྗత͡Όͳ͍ɻɻɻ • ใुΛTγϟπʹ͢Δ͜ͱͰࢀՃΛଅ͢ • ʢগͳ͘ͱ΋GoogleͰ͸ʣTγϟπ໊͕੠ͷ ୅ΘΓɺۚમΑΓ΋໊੠
19. ར఺ɿਖ਼֬ͳ༧ଌ
20. Two-outcome
21. Five-outcome
22. ͕࣌ؒܦͭʹͭΕinformativeʹ
23. ར఺ɿࣾ಺ͷަྲྀ͕׆ൃʹ • ैۀһͷਓؒؔ܎͕޿͕Δ • ใुΛಘΔͨΊʹ͍ΖΜͳͱ৘ใަ׵Λߦ͓ ͏ͱ͢Δ • ձ࿩ͷ͖͔͚ͬʹͳΔ • ʮ༧ଌࢢ৔͸ैۀһಉ࢜ͷձ࿩ͷ৔ͳͷͩʯ
24. ൃݟɿ෺ཧతڑ཭ͷॏཁੑ • ಉ͡৔ॴʹ͍ΔτϨʔμʔͨͪ͸ಉ࣌ʹಉ͡ औҾΛߦ͏܏޲ʹ͋Δ • σεΫͷ৔ॴ΍पғͷ஥͕ؒมΘΔͱߦಈ΋ มΘΔ • ʮߟ͑ํ͸෺ཧతͳۙ͞ʹΑܾͬͯ·Δʯ
25. GPMͷݱࡏ • ݱࡏ͸ऴྃʢ͓ͦΒ͘ʣ • ΋ͱ΋ͱͷӡӦϝϯόʔ͕࣌ؒΛׂ͚ͳ͘ ͳͬͨͨΊ • Bo Cowgill͕PhDऔΓʹߦͬͨ
26. ·ͱΊ
27. ࣾ಺༧ଌࢢ৔ͷར఺ • ਫ਼౓ͷߴ͍༧ଌ͕ಘΒΕΔʢ܏޲ʹ͋Δʣ • ࢢ৔ௐࠪ΍ઐ໳Ո΁ͷώΞϦϯά౳ɺैདྷͷ खஈʹൺ΂ͯɺۚમతɾਓతίετ͕௿͍ • ैۀһؒͷަྲྀ͕׆ੑԽ͢ΔͳͲͷ෭࡞༻
28. Remaining Problems • ಘΒΕͨ༧ଌΛͲ͏ҙࢥܾఆʹ༻͍Ε͹ྑ͍ ͷ͔ʹ͍ͭͯ·Ͱ͸ڭ͑ͯ͘Εͳ͍ • ಋೖͷ೉͠͞ • ηϯγςΟϒͳ৘ใΛެ։͢Δ͜ͱʹͳΔ • طಘݖӹΛڴ͔͢
29. ͓ΘΓʹ
30. ΋ͬͱ஌Γ͍ͨਓ͸ • ʮී௨ͷਓͨͪΛ༬ݴऀʹม͑Δʰ༧ଌࢢ৔ʱͱ͍ ͏৽ઓུʯɺυφϧυɾτϯϓιϯ • Google Official Blog ``Putting crowd wisdom to work” (https://googleblog.blogspot.com/2005/09/ putting-crowd-wisdom-to-work.html) • Prediction markets at Google(https:// www.slideshare.net/nimesh94/prediction-markets- at-google-gpm)
31. ΋ͬͱ஌Γ͍ͨਓ͸ • “Corporate Prediction Markets: Evidence from Google, Ford, and Firm X”, Cowgill and Zitzewitz, Review of Economic Studies, 2015 • “Using Prediction Markets to Track Information Flows: Evidence from Google”, Cowgill, Wolfers and Zitzewitz, 2009
32. Eagna • ”Eagna”ͱ͍͏αʔϏεΛӡӦɾ։ൃͯ͠·͢ • PCɺεϚϗͷϒϥ΢β্Ͱ༧ଌࢢ৔ΛແྉͰ ମݧͰ͖·͢ʢsign upඞཁʣ • Ϛʔέοτ͝ͱʹίΠϯΛ഑෍͢ΔͷͰɺͦ ΕΛͨ͘͞Μ૿΍͍ͯͩ͘͠͞ʂ
33. Eagna • ใु΋͋Γ·͢ʂ • ֫ಘͨ͠ίΠϯʹൺྫͯ֬͠཰తʹΪϑτ݊Λ഑ Γ·͢ • eagna.ioͰݕࡧʂ • ϑΟʔυόοΫେ׻ܴͰ͢ʂ
34. ༧ଌࢢ৔ษڧձ • ೔ຊͷ༧ଌࢢ৔ίϛϡχςΟͱͯ͠ຖ݄ߦͬ ͍ͯ͘༧ఆͰ͢ • 10݄΋΍Γ·͢ • ೔࣌ɺձ৔ɺςʔϚ͸connpass౳Ͱʂ • ੋඇ࣍ճ΋͝ࢀՃԼ͍͞ʂ
35. Q&A
36. ͋Γ͕ͱ͏͍͟͝·ͨ͠ʂ