"I read about it on HackerNews" ain’t gonna cut it in 2025

Or: how to convince your CTO to adopt new tech...and actually make it a success.

Transcript

1. O r : how t o c on v i n c e y o u r C T O t o a do pt new te c h . . .a n d a c t ua ll y m a k e it a s ucces s "I read about it on HackerNews" ain’t gonna cut it IN 2025 twn sn d . com / le ed s r ust 2025
2. • 20 y e a rs i n we b d ev e l op m ent , p rim ar i l y e C o mm e r c e & S aa S • S t a rted i n P H P , m o ve d o n t o Ru b y on Ra i l s • 10 y e a rs as C T O • Co - fo unde d S H/ FT Co mme r ce • Sp oke n a t > 20 c o n fer e nce s gl o b all y • Lif t e r , F oo d ie, C of fee , W ine , Tr a v el
3. G o E l i x i r P H P R u st R u b y A pach e K a f ka A n y thi n g but g odd a m n R e a ct H o w c a n I c on v i n c e le ad e r sh ip Ru st i s a g oo d c h o i ce ?
4. T h e re are just two s ilver bull ets
5. H o w do I k n ow Rust i s th e c or rec t c h o i ce ? . . .a n d m a k e a s ucces s o f a do p tin g it?
6. TALE 1: INHERITING ENGINEERS
7. – C E O “Slap some e-commerce on it”
8. What c o u ld p os s ibly go w ro n g ?
9. 🔥 🔥 🔥
10. “We’ve estimated how long it’ll take”
11. “3 months... for a team of 4”
12. “3 months... for a team of 4”
13. Yi k e s 😬
14. – T h e . NE T N i nja “~3 weeks would be fair, .NET is no Rails”
15. “Rails doesn’t scale”
16. Sou r ce : hi g hsc a l ab i li t y. com / s c al in g - t w i t t e r - mak i n g - t w i t t e r -10000- pe r c e nt - fas te r / ( 2009 ) “Rails and Ruby haven’t been stumbling blocks [...] the performance boosts associated with a “faster” language would give us a 10-20% improvement” – B l ai ne Co o k , Twi t t e r 's Le a d Ar ch ite ct (a t the t ime )
17. O u r u ncache d perf o r ma nc e
18. Lessons learned • T h e re are d i f f er ent ki n d s o f d ev e l o p e rs • Get a d v i ce f rom an e xp er t • A lign w it h the b u s i n es s comme rcia l p ri o r it ies • Un d e rst a n d y o ur c on str aints/ e xpe c ta tio n s , e.g. n o n - fu n ctio n al re q s • I nte rro g a te op ini on ( p o si t ive o r n eg ative )
19. TALE 2: A SELLER'S MARKET
20. – Fr ont - e n d T e am L e a d “We should be using a JavaScript framework”
21. 6 WEEK BOOTCAMP TO 6 FIGURE SALARY
22. A COMPELLING ARGUMENT ✅ J av a S cri p t / T yp e S cri p t we re whe re the inc r e a s i ng ly- co m pet it ive ma r ket was h ea d i ng – h i r ing / r et ention wou ld be bett e r ✅ We a lre ad y h a d J av a S cri p t e xp erie nc e f rom our ex is ting f ront - e n d ✅ We h a d the do m a in e xp er t ise a n d an ex is ting f ront - e n d f o r s co p i ng ✅ Web perf o rma nc e corr elat es w it h e- comme r ce KP Is a n d Next . j s pr omis e d " ins t ant " a p p -l ike e xp erie nce s
23. It wa s n 't a l l b a d .
24. Lessons learned • Ag r ee how we me as u re s ucces s • Co n s i der the t al ent a v ailabi li t y / job ma r ket • Pr e p a re t o in ve st i n e xp er t ise (t r aini ng , c o n fer e nce s etc) • Don 't c hang e the s cop e AND the te c h – a do pt o n e u nkn o w n a t a t ime • Don 't b u y the h yp e , do y o ur ho mew o rk
25. TALE 3: GO, NO GO
26. – B a c k - e n d T e am Le a d “Things are getting stale and I worry we’ll lose long-tenured sta ff —we’d like to consider a di ff erent language for this new service”
27. FO M O i s real a n d it i s vali d.
28. Rails NODE.JS ELIXIR / PHOENIX GOLANG ✅ Most e xp erie nc e ☑ A cce p t ab le perf o rma nc e ❌ T h e stat us q u o ❌ Dy i ng? ? ? ✅ S tro n g t al ent a v ailabi li t y ✅ Go o d e xp erie nc e f rom f ront - e n d ✅ Inc r e a s i ng ly l ar ge e cosy s t em ☑ A cce p t ab le perf o rma nc e ✅ S ide -proj ect e xp erie nc e ✅ La nguag e d ev e l o pe d by ex - Rail s d e v ✅ Ex c ellent perf o rma nc e / c onc urre nc y ❌ L i m ited t al ent p o ol ✅ Ex c ellent perf o rma nc e ✅ S igni f ic a nt comme rcia l u s e 🤷 Co m p letely new t o us ❌ L i m ited t al ent p o ol
29. We c o u l dn 't s a cr if ice pr o du cti v it y.
30. I c o u ld o ff e r cal m.
31. Sou r ce : hi g hsc a l ab i li t y. com / s c al in g - t w i t t e r - mak i n g - t w i t t e r -10000- pe r c e nt - fas te r / ( 2009 ) “Tried Erlang. Problem: How do you get a broken server running at Sunday [morning] with 20,000 users waiting? The developer didn't know.” – B l ai ne Co o k , Twi t t e r 's Le a d Ar ch ite ct (a t the t ime )
32. P us h th e b ound aries o f e x is ti ng to o ls Sou r ce : l e ssons o f a c to. c om /vi d eo s /007- s w eat - y o ur - t e c h -st a c k /
33. De v e l o p e rs o ften u nd er - sel l the m se l v e s
34. Sou r ce : t a il s c a le. com /bl og / ne w - i nter n et ( 2024 ) “I read a post recently where someone bragged about using kubernetes to scale all the way up to 500,000 page views per month...” – A v e ry P enn a r un , Tails c a le
35. Sou r ce : t a il s c a le. com /bl og / ne w - i nter n et ( 2024 ) “...that’s 0.2 requests per second. I could serve that from my phone, on battery power, and it would spend most of its time asleep.” – A v e ry P enn a r un , Tails c a le
36. Sou r ce : t a il s c a le. com /bl og / ne w - i nter n et ( 2024 ) “The truth is, most things don’t scale, and never need to.” – A v e ry P enn a r un , Tails c a le
37. Lessons learned • Put t ing it d o w n on p ap e r cr ea t ed a pr o du ct iv e di s cus s i on • Co n s i der the e cosy s t em , not j ust t he te chn o l o g y i n i so la ti on • Pe r s on al u sag e i s bett e r th a n no thi ng , but h ow d oe s it br e a k ? • A g a in, a l ign w it h comme rcia l p ri o r it i es • A g a in, u nd e rst a n d c o n t e xt : we di dn 't n ee d t o s av e 20 ms • If y o u c a n 't 'w i n': di s a g r ee a n d c omm i t – go d eep er r at her th a n w ide r • Ag r ee g u a rdra il s f o r mo re au t o n o m y el se w he re • Ret ro spe c tiv e s are c r it ical t o l o n g - t e r m m o ra l
38. TIMES CHANGE, FADS CYCLE
39. • N o SQL • M i cr o s e rvi c e s • G ra p hQL • Se r verle s s • T h e Rail s Re n a i s sanc e • P H P i s c o ol a g a in • S ing le P ag e A pp s ( d e a th tbc )
40. BEWARE OF MARKETING
41. • Rep r e se n ts a s m al l but v oc al s ubs et o f the in du str y • T h e r ise o f the ' De v e l o p er A dvoc a te' • Few p u bli ci se f a il ure • Ca s e st ud i es u su all y la c k important c o n t ext
42. AI
43. THE AVAILABILITY OF TALENT STEALABLE COPYRIGHTED CONTENT TO FEED LLMS
44. THE AVAILABILITY OF TALENT STEALABLE COPYRIGHTED CONTENT TO FEED LLMS
45. A BLUEPRINT
46. • T ak e m a r k eting w it h a pi n c h o f sal t – d o n 't b u y h yp e • M a k e it a c onv e r sat io n , it s ho u l dn 't be a z e ro- su m ga m e • ' P o p t he why st a c k ', a li gn w i t h b u s i n es s p r i o r it i es a n d b ui l d a pl an f o r me as uri ng s ucces s • D o c u m e nt fa vo u r e d o pt io n s a n d a lt er n ati v e s • ' To ta l Co st o f O wne r s h ip' (TC O) in c l u de s sa la r ies , t r aini n g , h i r in g , r et ention , i n f ra s tru c ture, o p e ra tio n s , m a int e nanc e , p ackage s a n d w ide r e cosy s t em • ( Pl an t o) in v o l v e o r h ir e an e xp er t • A v o id ' T h e B i g R ewr i t e'; what c o u ld be a st e p p i ng st o n e ? • W ha t are our s h o r t-, med i um - a n d l o n g - t e r m s tr at e g ic g o a ls?
47. REMEMBER: WE'RE ALL TRYING TO REDUCE RISK, JUST IN DIFFERENT AREAS
48. Thank you 🙏 t w n sn d . c om / le ed s r us t 2025