2.3.8 Algorithm ididekile
Indlela yokuziphatha kwiinkqubo ezinkulu zeenkcukacha zendalo; iqhutywa yiinjongo zobunjineli zeenkqubo.
Nangona imithombo emininzi yolwazi ingasebenzi ngenxa yokuba abantu abaqapheli ukuba idatha yabo ibhalwe (icandelo 2.3.3), abaphandi akufanele baqwalasele ukuziphatha kwezi nkqubo ze-intanethi ukuba "zenzeke ngokwemvelo." Enyanisweni, iinkqubo zedijithali ezirekhodiweyo zokuziphatha ixinwe kakhulu ukwenza izinto eziphathekayo ezifana nokuchofoza kwiintengiso okanye ukuthumela okuqukethwe. Iindlela iindlela zokuqulunqa iinkqubo ezingenako ukuvelisa iipateni zibe nedatha zibizwa ngokuba yi- algorithmic confounding . Ukuphazamiseka kwe-algorithm akungaziwa kwizenzululwazi zentlalo, kodwa kuyinkxalabo enkulu phakathi kweengcali zengcali eziqapheleyo. Kwaye, ngokungafani nezinye iingxaki ngeendlela zedijithali, ukudibanisa kwe-algorithmic ingabonakali.
Umzekelo onqabileyo we-algorithmic ephazamisayo kukuba kwi-Facebook kukho inani elininzi elingabonakali labasebenzisi abangama-20 abahlobo, njengoko kwafunyanwa nguJohan Ugander kunye noogxa (2011) . Iingcali zenzululwazi zihlalutya le nkcukacha ngaphandle kokuqonda indlela i-Facebook isebenza ngayo ngokuqinisekileyo ivelise amabali amaninzi malunga nendlela engama-20 enjani inombumba yoluntu lomlingo. Ngethamsanqa, u-Ugandanda kunye nabalingane bakhe babenokuqonda okukhulu kwinkqubo eyenza idatha, kwaye beyazi ukuba i-Facebook ikhuthaze abantu abanxibelelwano oluncinane kwi-Facebook ukuze benze abangakumbi abahlobo baze bafikelele kubahlobo abangama-20. Nangona u-Uganda kunye nabalingane bangatsho oku kumaphepha abo, le nkqubo yayingadalwa ngu-Facebook ukuze kukhuthazwe abasebenzisi abatsha ukuba basebenze. Ngaphandle kokungazi malunga nokukho komgaqo-nkqubo, kunjalo, kulula ukuphetha kwisiphetho esingalunganga kwi-data. Ngamanye amazwi, inani elincamisayo labantu abanama-20 abahlobo basitshela ngakumbi nge-Facebook kunokuziphatha kwabantu.
Kulo mzekelo wangaphambili, ukudibanisa kwe-algorithmic kwavelisa umphumo we-quirky ukuba umphandi ochanekileyo unokufumanisa nokuphanda. Nangona kunjalo, kukho inguqu enkulu ye-algorithmic confounding eyenzekayo xa abaqulunqa iinkqubo ze-intanethi bezazi iingcamango zentlalo baze bhakaze le ngcamango ekusebenzeni kweenkqubo zabo. Inzululwazi zentlalo zibiza lo msebenzi : xa i-theory iguqula ihlabathi ngendlela ezisa umhlaba ukuba uhambelane nenkolelo. Kwimeko yokwenza i-algorithmic ephazamisayo, ubunzima obudityanisiweyo bwedata kunzima ukubona.
Omnye umzekelo womzekelo owenziwe ngumgangatho wokusebenza uguquka kwiintanethi zentlalo. Ngama-1970 kunye nee-1980, abaphandi bafumanisa ngokuphindaphindiweyo ukuba ukuba ungumhlobo kunye no-Alice noBob, ngoko u-Alice noBob banokuba ngabahlobo xa beba ngaba bantu ababini abakhethiweyo. Le ndlela efanayo ifumaneka kwigrafu yentlalo kwi-Facebook (Ugander et al. 2011) . Ngaloo ndlela, umntu unokugqiba ukuba iipatheni zobungane kwi-Facebook ziphindaphinda iintlobo zobungane obungekho phantsi kwe-intanethi, ubuncinane ngokubhekiselele ekutshintsheni. Nangona kunjalo, ubukhulu be-transitivity kwi-graph ye-Facebook yoluntu luqhutyelwa ngokukodwa ngolu hlobo lwe-algorithmic confounding. Oko kukuthi, izazi zesayensi kwi-Facebook zazi malunga nophando lwezobugcisa kunye nolwazi malunga nokuguquka komzimba kwaye ziyabhaka kwindlela i-Facebook isebenza ngayo. I-Facebook inomxholo othi "Abantu Ongaziyo" obonisa abahlobo abatsha, kwaye enye indlela i-Facebook inquma ukuba ngubani na onokubonisa ukuba usuqulelo. Okokuthi, i-Facebook inokubonisa ukuba ube ngabahlobo bakho nabahlobo bakho. Lo mbandela unesiphumo sokwandisa ukuguquka komfanekiso kwi-Facebook yoluntu; Ngamanye amazwi, imbono yokuguquka komsebenzi izisa umhlaba (Zignani et al. 2014; Healy 2015) kweengcamango (Zignani et al. 2014; Healy 2015) . Ngaloo ndlela, xa imithombo yolwazi emikhulu ibonakala ivelisa izibikezelo zentlalo yesifundo, kufuneka siqiniseke ukuba inkolelo ngokwayo ayibhalwa ngendlela inkqubo eyasebenza ngayo.
Kunokuba sicinge ngemithombo emininzi yolwazi njengokubheka abantu kwindawo yokumgangatho wendalo, isalathisi esithe xaxa sigcina abantu kwi-casino. I-Casino iindawo ezizimeleyo zenzelwe ukunyusa iimpawu ezithile, kwaye umphandi akanakuze alindele ukuziphatha kwi-casino ukubonelela ngefestile engapheliyo ekuziphatheni komntu. Ngokuqinisekileyo, unokufunda okuthile ngokuziphatha kwabantu ngokufunda abantu kuma-casinos, kodwa ukuba wawungayigcinanga into yokuba idatha yadalwa kwikhasithi, unokwenza izigqibo ezimbi.
Ngelishwa, ukujongana ne-algorithmic confounding kunzima kakhulu kuba iinkalo ezininzi kwiinkqubo ze-intanethi zimeyile, zibhalwe kakuhle kwaye zihlala ziguqukayo. Ngokomzekelo, njengoko ndiza kuchazela kamva kwesi sahluko, ukuchithwa kwe-algorithmic kwakuyiyo enye inkcazo yokuchithwa kweminye imithwalo ye-Google Flu Trends (icandelo 2.4.2), kodwa le bango kwakunzima ukuvavanya kuba ukusebenza kwangaphakathi kwe-algorithm ye-Google yosesho umnikazi. Uhlobo olushukumisayo lwe-algorithmic confounding yindlela enye yokutshintsha inkqubo. I-algorithmic confounding kuthetha ukuba kufuneka siqaphele nayiphi na ibango malunga nokuziphatha komntu okuvela kwisistim esisodwa sejithali, kungakhathaliseki ukuba sikhulu kangakanani.
