Dig ita l Da rk ma tte r By Sha ne Gre e nste in F o r CAI DA WI - - PowerPoint PPT Presentation

▶
dig ita l da rk ma tte r
SMART_READER_LITE
LIVE PREVIEW

Dig ita l Da rk ma tte r By Sha ne Gre e nste in F o r CAI DA WI - - PowerPoint PPT Presentation

May 19, 2023 195 likes •307 views

Dig ita l Da rk ma tte r By Sha ne Gre e nste in F o r CAI DA WI E T he a g e nda GDP is the to ta l flo w o f fina l g o o ds a nd se rvic e s in a n e c o no my o ve r a spe c ific pe rio d o f time (witho ut do ub le c o unting

slide-1
SLIDE 1

Dig ita l Da rk ma tte r

By Sha ne Gre e nste in F

  • r CAI

DA WI E

slide-2
SLIDE 2

T he a g e nda

 GDP is the to ta l flo w o f fina l g o o ds a nd se rvic e s in a n e c o no my o ve r

a spe c ific pe rio d o f time (witho ut do ub le c o unting ).

 E

ve ry de ve lo pe d c o untry me a sure s GDP, c a pita l sto c ks, la b o r sto c k, c a pita l flo ws a nd la b o r flo ws. Bro a d e ffo rt to unde rsta nd c o ntrib utio ns fro m va rio us pa rts.

 Digital dar

k matte r: Asse t w/ o a ppa re nt mo ne ta ry va lue , tha t c a n b e

re plic a te d witho ut limit, & a c t a s I T input into pro duc tio n.

 if pric e is ze ro , it c o ntrib ute s no thing to GDP  I

nc re a sing ro le o f “inta ng ib le ” c a pita l in the e c o no my.

 At this sta g e : de mo nstra ting pr

  • of of c onc e pt.
slide-3
SLIDE 3

One wa y to mo tiva te it: One o f se ve ra l me a sure me nt issue s due to a ze ro pric e .

Ze r

  • Pr

ic e in most of the output Positive pr ic e in most of the output

Ze ro Pric e in mo st o f the inputs

  • Wikipe dia.
  • Apac he / L

inux F

  • unda tio ns
  • Github.
  • And mor

e … .

  • Andr
  • id pho ne use s L

inux ke rne l.

  • Unlic e nse d spe c tr

um.

  • Alg o rithms fo r mac hine le ar

ning

Po sitive Pric e in mo st o f the inputs

  • Google / F
  • B. F

re e o utput, a ds ra ise re ve nue .

  • Comme r

c ial WE

  • B. Se e a b o ve .
  • Ma nufa c turing iPhone.
  • Ne tflix pa y lic e nse s. Sub sc rib e rs pa y

fe e .

  • Ca rrie r pa id lic e nse for spe c tr

um.

slide-4
SLIDE 4

Wha t the first study do e s (Gre e nste in a nd Na g le , 2015)

 T

a ke s a o ne pe rc e nt sa mple o f a ll I P a ddre sse s a nd e stima te s the to ta l numb e r o f Apa c he se rve rs in the US in 2011. Pr

  • of of c onc e pt e xe r

c ise .

 Atta c h a va lue to o pe n so urc e se rve rs using sta nda rd pro c e dure s fo r “ne a r

ma rke t g o o ds,” a s de fine d b y No rdha us, 2006.

 Apa c he a c c o unts fo r a misme a sure me nt o f so me whe re b / w $2 b illio n

a nd $12 b illio n in so ftwa re in the US in 2011.

 L

ar ge . E q ua te s to b / w 1.3 % & 8.7 % o f sto c k o f pre pa c ka g e d so ftwa re in priva te fixe d inve stme nt in the US

 High ROR. Re turn o n Apa c he a lo ne wo uld ha ve g e ne ra te d suffic ie nt ra te o f

re turn to justify inve stme nt in I nte rne t R&D b y F e de ra l g o ve rnme nt.

 Omission biase s. L

a rg e .

 Attrib utio n b ia s? No .

slide-5
SLIDE 5

Wha t the se c o nd study do e s (Ac ke rma nn a nd Gre e nste in, 2018)

 T

a ke s a b ig sa mple o f o utwa rd fa c ing se rve rs in the pla ne t in 2012. Apa c he , ng inx, & I I S in 2012. Pro o f o f c o nc e pt a t a global le ve l.

 L

  • c a te s a se rve r in a c o untry using me tho ds fo r a ssig ning

long/ lat to I P a ddre sse s,. Co mpa re a c ro ss c o untrie s.

slide-6
SLIDE 6

Ma ps: whe re se rve rs a re lo c a te d

slide-7
SLIDE 7

Wha t the se c o nd study do e s (Ac ke rma nn a nd Gre e nste in, 2018)

 T

a ke s a b ig sa mple o f o utwa rd fa c ing se rve rs in the pla ne t in 2012. Apa c he , ng inx, & I I S in 2012. Pro o f o f c o nc e pt a t a global le ve l.

 L

  • c a te s a se rve r in a c o untry using me tho ds fo r a ssig ning

long/ lat to I

P a ddre sse s,. Co mpa re a c ro ss c o untrie s.

 Ske we d a c ro ss the wo rld.  US ha s 44% o f se rve rs. Ne xt b ig g e st is China , whic h is six time s

sma lle r.

 Ana lyze who ha s mo re a nd why?

 Be st pre dic to rs o f se rve r a nd pe r-c a pita se rve rs a re the q ua lity o f the

ne two rk a nd the so phistic a tio n o f the la b o r ma rke t fo r te c hnic a l g o o ds.

 Rule o f la w a nd indic a tio ns o f “fa ile d” institutio ns do no t ma tte r

 Ve xing . Ca nnot e a sily fore c a st ope n sourc e sha re

slide-8
SLIDE 8

T hird study fro m Wa yb a c k Ma c hine (Gre e nste in, Murc ia no -Go ro ff & Zhuo , 2018)

Sma ll & me dium firms L a rg e firms L a rg e st firms * Pre limina ry: Usa g e sha re s fo r we b site s a t 200k+ US firms. A who le a sse t c la ss fo r whic h we ha ve NO inve stme nt da ta . Me a sure d fo r the first time . Ve ry diffic ult to c le a n up. * Pro o f o f c o nc e pt. L a rg e (? ) misme a sure me nt o f inta ng ib le c a pita l inside firms. * Ve ry diffe re nt a do ptio n a nd upg ra de b e ha vio r a c ro ss diffe re nt type s o f firms.

slide-9
SLIDE 9

Ope n q ue stio ns:

 Othe r e xa mple s?

 F

ra nk Na g le ha s sho wn tha t muc h OS ha s a dire c t pr

  • duc tivity impa c t
  • n the firms who de plo y it.

 T

he sa me me c ha nisms. Any ope n sour

c e? E.g ., L

inux, F ire fo x, PE RL , PHP, 802.11 so ftwa re .

 Spillove r

s fr

  • m unive r

sity R&D witho ut lic e nsing . E

.g ., WWW, b ro wse rs, T CP/ I P so ftwa re pro to c o ls, e tc .

 Cr

e ative c ommons: Wikipe dia , K

a hn Ac a de my, Github ? Unme a sure d g a ins c o uld b e la rg e due to me a sure d b y a udie nc e size .

 T

he spe c tr

um in Wifi e q uipme nt? Sta nda rds fro m I

E E E

 Be yo nd pro o f o f c o nc e pt? Mo re tha n a ro unding e rro r. L

a rg e e no ug h va lue to ma tte r.