Ukubuyekezwa ku-Future of Software Development

Uma uhamba e-bar ne-friend who is working in IT, ungaphendula ukuxhumana nezimo zokusebenza ezivela ku-industry. I-first half of the year has given us a lot to talk about, futhi laphajust a few general observations we both made:

• besides the usual impostor syndrome that many software developers have, there's also a lot of anxiety about all the chatter around AI taking over jobs

• leaders and managers can’t help but get into conversations about how much of a performance boost (and when) to expect from developers, whether it’ll be x10, x40 or x200

• at the same time, the push for AI adoption makes both tears and laughter

• everywhere you look, you can find many near-absurd product features that use AI only to justify that it is from an AI-first company

• drought and nervous foot tapping are common in the startup world, yet discussing them is often considered bad form

• hiring is becoming increasingly difficult for everyone, especially for people just starting out in their careers

  
  

One of the biggest concerns about the impact of hype surrounding AI is that it could discourage new people from entering the software development industry.

"Lesi akuyona umsebenzi wokudala; ukufundisa ukuthuthukiswa kwe-software namhlanje kuyinto isithuthuthu kumadoda; umdlalo wahlukaniswa."

"This is no longer creative work; learning software development today is a scam for fools; the game is solved" (this kind of rumors).

Ukuze ukugcina umkhakha we-tech ngokushesha, kubalulekile ukuhambisa ulwazi olusebenzayo kwezinto ezisebenzayo futhi ukunceda ama-newcomers ukufumana izilinganiso ezidingekayo ukuhambisa konke ngokushesha kanye nokuthuthukiswa okwengeziwe kwe-tech (U-Jonathan Blow wabhala ngokugqithiselwe lokhuNgaphandle kokuzihlanganisa).

Ngicabanga ukuthi thina kuqala isixazululo ebalulekile kwebhizinisi yethu. Nokho, kum, kulinganiswa nokuthuthukiswa kwewebhu kusuka ku-2008 kuya ku-2014, engaphezulu kwe-PC boom ye-1980s (i-expansion, engaphezulu kwe-catalogue entsha).

Ngemva kwalokho, kukhona izindlela ezininzi zokwenza izinto eziyinhloko. Ngonyaka ngamunye, kubonakala ukuthi sinikeza indlela entsha yokwenza izinto. Kuyinto efana nokuqinisekisa kwe-AI yamanje, lapho ithuluzi entsha, imodeli, indlela, benchmark, umkhakha, noma inkampani ivela ngeviki, ukhangela ukuthi zithintela ukuhlangabezana konke.

Akukho umdlavuza, kodwa kum, izixhobo ezifanaWazeKuyinto kakhulu like whatUkusebenza kwe-Adobe DreamweaverUkusebenza kwe-Web Development ngo-2010.

Ngiyaxolisa

Ukusebenza kwe-Software kunazo iminyaka engaphezu kwama-70. Ngaphezu kwalokho, idume wonke umhlaba ubambe ezininzi. Thina ucwaningo oluthile indlela yakhelwe ngonyaka ngamunye.

Ngiyazi AinguquleloUkubuyekezwa kwe-image phezulu

Njengoba ungathanda, indawo lethu isebenza ngokushesha ku-abstraction eningi kanye nezindlela eziningi ezomathemikhali zokulawula ukucindezeleka. Ngaphezu kwalokho, izinsuku ezintsha zihlanganisa imibuzo ezintsha ezinikezela ukulawula imibuzo ezidlulileyo kanye nokushintshisa i-hyper-specialization kanye namaqembu amancane, okunikezela izindlela ezininzi zokufaka ukubuyekeza ngokushesha.

What complexities are we facing in the 2020s (so far)?

• Supply chain security
  • modern software development relies on numerous third-party components, with even simple applications often incorporating thousands of them, each posing a potential security risk that could compromise the entire application
• Observability data overload
  • metrics, logs, and traces generate vast amounts of data daily, but pinpointing the root cause is becoming increasingly challenging because dashboards display everything yet explain nothing (debugging is still hard)
• Configuration management
  • use of IaC, feature flags, env variables, and secrets has created massive configuration spaces that are hard to validate
• AI/ML integration complexity

  • integrating AI components with traditional software is still quite challenging, as SOTA and toolchains change every two weeks, and approaches evolve even faster

    
    

Replacing software developers with AI agents can't be a solution for any of today's complexities.

A few things to watch for:

• AI tooling consolidation

• commoditization of AI workflow creation

• even better ways of distributing software

• changes in the way we build software that cultivate higher quality as consumer expectations rise due to software abundance

  
  

Bet ku-Software Development

I-Code Generation yinto elihle yokuqala ye-LLMs, njengoko i-open data eningi ukwenza ukusebenza kanye ne-wow-effect eningi enikeza wonke umuntu ukhangela. I-performance ikhiqiza esitimela etholakalayo kusuka ku-copilots ingatholakala kalula ekhululekile emkhakheni ebizwa yi-media noma lapho abantu baqala ukufuna i-side-gigs ukuze uthole imali engaphezu nokuphephile elilandelayo.

Ngokuvamile, kufanele ukholelwa ukuthi umthamo ye-LLM yokukhiqiza ikhowudi iyahambisa kuphela ngesikhathi (Njengoba lokhu kubonakalaUkubuyiselwaImibuzo yayo elikhulungezikhathi ezidlulile), ukuhlangabezana nezimo zayo (njenge-Ukuphakama okusheshayo okukhuthaza ukusebenza kakhuluNgaphezu kwalokho).

The effect of AI on cutting costs might be tricky, as everyone has the same chances to optimize. What's most likely to happen is that AI will make the whole pie bigger.

Yini kwaba lapho abantu abaninzi abafundisa indlela yokufunda, ukhiye, ukhompyutha, ukhiye, bese ngempumelelo ukhiye umsebenzi yabo ku-intanethi nge-worldly? Yini kwaba lapho abantu abaninzi bafumana izivakashi, ukhiye, ukongeza amafutha kanye nemiphumela, bese ukhiye ngoba wonke abantu ukubuyekeza (noma ukuthenga futhi ukubuyekeza ngemva)? Noma okufanayo, kodwa nge-inthanethi yokufundisa?

UkubuyekezwaIzinto ezijulileKodwa ushiye: Abacwaningi, abacwaningi, futhi abacwaningi abanolwazi futhi abanolwazi kakhulu (ngezinye kakhulu).

Kubonakala ukuthi sinikeza isimo esifanayo emahora we-Web lapho wonke umntu waqala ukwakha iwebhu ngokufanayo. Lokhu kuya ukwandisa i-bar ye-business ne-individual (njenge-websites ne-webapps), okwenza ama-differentiators ezintsha, izixhobo nezindlela kanye nokuvumela izidingo ezintsha kumadivayisi.

Ngokuvamile, ngacindezela i-ideas eyenziwe ngokucacileyo kuma-articles ezidlulile: "Ukuphakama kwe-programming njengasehlanzekile« futhi »I-AI ne-Programming: Ukuqala Isikhathi Enye“

So, what is next?

Ingabe sinikeza isixhobo se- "IKEA" enikeza i-app container enezinsizakalo zinsizakalo zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza, okuvumela ukuba utshintshe umphathiswa we-AI ukufaka izici ezingatholakala futhi ukongeza izici ezivamile, bese ukucindezeleka ngempumelelo?

Noma i-crypto ngeke ikakhulukazi inikeze inethiwekhi ethu kakhulu bonke (mhlawumbe ngempumelelo zihlanganiswa nezinsizakalo zangaphambili zendawo)?

Noma kuyoba izinga lokusebenza ze-AI e-Excel-level?

Ingabe sinikeza ama-services eyenziwe futhi izakhiwo ezintsha ukuze zihlanganise nezidingo ezintsha ezivamile: i-community (noma i-network) ne-AI agents?

Nginguquko, kuphela umdlavuza we-phantom. Uma siphinde isigaba esitsha sokuthuthukiswa kwe-software spiral, kakhulu izinguquko, futhi ngaphezulu kuya kuba.

Ukungena ngemvume

Ngo-environment lapho ukwengeza ikhowudi ngaphezulu kunezimali, umbhali we-software enhle kubona umbhali we-classic kunokuba umbhali we-classic.

Thina zihlanganisa imibhalo amabili (The Big Blue, The Green and The Dragon, njll), zihlanganisa umbhalo we-magic (izakhiwo zabo zangaphambili kanye nezinsizakalo, izimfuneko ezinhle), zihlanganisa izilimi ezizodwa (izincwadi, i-acronyms, i-lexicon), zihlanganisa ezinye izibonelelo ze-esoteric (i-diagrames), zihlanganisa ezinye izikhwama embhedeni (i-interfaces), futhi uma zihlanganisa, zihlanganisa "ukuguqulwa kanye nokushesha" (abracadabra).

Izidakamizwa ezivamile zihlanganisa izidakamizwa ezivamile. Izidakamizwa ezivamile zihlanganisa izidakamizwa ezivamile. Izidakamizwa ezivamile zihlanganisa izidakamizwa ezivamile zihlanganisa izidakamizwa ezivamile.

Kodwa lokhu akuyona futhi akufunde indlela yokusebenzisa.

Some things can significantly enable you on this journey:

• When code is generated, it will eventually lack evident vulnerabilities and be validated against requirements using automated tests.

  • Your job is to ensure that the code is maintainable (this makes it easier for both machines and humans to troubleshoot and extend the codebase).

• You need to understand, appreciate, and delve into the fundamentals of software architecture and the core principles of computer science.

  • The best way to go is to a) study hard, b) build from scratch, and c) revise and exercise (regularly).

• The programming languages ​​won't matter much, but mastering two languages ​​- a dynamic, high-level one and a static, low-level one - will give you enough opportunities to practice all the essential concepts and broaden your perspective.

  
  

Qinisekisa ukuthi izinto akuyona okungcono kwebhizinisi. Kulesi xesha, ungakwazi ukufumana umsebenziUmthetho we-United States(UkuzaIningi Waze IzindawoOkunye) nomaZonke amabhange, njengoba zihlanganisa kakhulu ku-codebases ebizwa ku-COBOL programming language (Imininingwane njengesilinganiso esihlalweni kwi-80sI-Joke elinye, lapho umkhakha we-tech uhamba ngokushesha kakhulu, isivinini sokuthuthukiswa kwebhizinisi namanye kanye nezinqubo zokusebenza zayo zihlanganisa ngokushesha (isib. Windows XP, Excel, Fortran, Perl).

Izinhlelo zokusebenza zokusebenza ngokuvumelana nezimo ezintsha. Lokhu kubalulekile ukuba abantu abalandeli abalandeli abalandeli abalandeli abalandeli.

1. Ukushintshwa ama-patterns ezaziwa futhi eziholile zokusebenza kanye nezinhlangano.
2. Thola izimo ngokuvumela into efanele ukuba bafuna izinto ezintsha.
3. Yenza isikhathi eside futhi amandla okwengeziwe ukufundisa futhi ukufundisa lokhu izinto ezintsha ngokuvamile.

I-Heads-up ye-Newcomers

Nangona ama-analogies ezivamile angama-resonance ne-reality esidumile, akufanele siphindezile. I-reality iyinhlanganisela kanye ne-dynamic, nangona i-history iyinhlanganisela kanye ne-contextual. Ngingu-ke akwazi ukuthi izimo izivakasha.

Many people will likely attempt to automate software development tasks, futhi kungenzeka ukuthi ukukhiqizwa kwe-code iyahambisa ngokushesha.

Ingabe ufuna ukushumeka kokuthunyelwe kwebhizinisi?

Kuyinto efanelekayo ukubonisa ukuthisoftware development shouldn't be your end goalThola lokhu njenge-medium kuya ku-end, noma le-end kuyinto emayeni, ibhizinisi, umkhakha, umkhakha, noma eminye indawo.

Ukubuyekezwa kwebhizinisi sokuthuthukiswa kwekhompyutha namhlanje iyiphi ingozi ye-agent personal. Ingabe ufuna ukufinyelela kuleli khosi, ngakho-ke uqala ukwenza umsebenzi lakho elilandelayo ngokushesha.

The great way to prepare for it is to:

• Ukufundisa izakhiwo ezinzima kodwa zihlanganisa izakhiwo ezithakazelisayo bonke abalandeli ngoba zihlanganisa
• Thola i-boyfriend noma i-community ebonakalayo uhlobo lakho le-energy kanye ne-enthusiasm
• usebenzisa imishini ye-AI yokufundisa nokufunda prototyping kusukela ngosuku eyodwa
• Ukusebenza, Ukusebenza, Ukusebenza

Yini kuyoba ukuguqulwa kakhulu kuyinto ukufundisa nokufunda ibhizinisi engaphakathi nemikhiqizo yakho yokwakha (amakhasimende, izinhlelo, imibuzo, kanye nezidingo). Ukufundisa ukuxhumana nabantu, ukucacisa izidingo zabo, ukucubungula izixazululo emkhakheni, i-prototyping isixazululo, ukucubungula nge-architecture epholile, kanye nokuxhumana nge-AI yokwakha.

Okwangoku ( kodwa awuzange), kunzima ukuvelisa ukucubungula kwelinye indawo ezithile. Qaphela izinto ezihambisa (noma ukujabulela) nawe futhi udinga ukujabulela ngaphezulu.

Ngemuva

I-AI ayifinyelela izindlela ezintsha zokufundisa nokuvumelana. I-AI ibonise izindawo lapho ukuhlolwa kwangempela nokuvumelana kuyimfuneko, ngempumelelo ukuchithwa kwe-vacuum nge-simulations yayo eminye izindawo.

Ukuthuthukiswa kwe-software akukwazi ukwenza wonke umngane. Uma unayo amandla le-builder ne-appetite enhle yokufunda nokufunda, kufanele usize! Uyakwazi ukufundisa yonke into ngokushesha.

Noma kunjalo, Ngithanda yonke into enhle emzimbeni yakho. Have fun doing it!

P.S. Uma ungenza le post, sicela ukhangeleconnecting with me on X or LinkedIn.