Home · Map
The map Areas of learning first. Open an area to see its claims, or use the diagram below for how beliefs connect.
The two big questions: Q1 How do humans actually learn, and what helps them learn best? Q2 How does that work in practice: in school, with screens, and with AI as one of the tools?
Memory and practice How we store what we learn and bring it back later — retrieval, spacing and mixing practice. 3 claims Claims: Retrieval beats restudy , What counts as learned , Spacing and interleaving
Teaching and feedback What teachers and tutors say and do that changes later learning, not just tonight's performance. 3 claims Claims: Task feedback vs praise , Step-based tutoring , Tutoring software in school use
Motivation and curiosity What keeps people wanting to learn — and what quietly kills interest. 2 claims Claims: Growth mindset for struggling students , Tangible rewards and interest
Sleep and the body How sleep, school start times and the body shape what sticks. 2 claims Claims: Later school start , Sleep before learning
Learning from other people Learning by explaining, teaching and working with others. 1 claim Claims: Learning by teaching
Expertise and practice How much deliberate practice explains skill — and what else matters. 1 claim Claims: Practice explains only part
Reading and comprehension Paper or screen, phonics, and the knowledge that makes text make sense. 3 claims Claims: Paper vs screen, deep reading , Systematic phonics , Background knowledge
Screens, devices and attention Phones, laptops and on-screen extras — when they help and when they get in the way. 5 claims Claims: Phones in lessons , Handwritten notes , Distracting extras · +2 more
AI and thinking When AI tools help practice and when they hollow out the skill they replace. 2 claims Claims: Answer-giving AI in practice , AI deskilling
Metacognition and self-regulation Teaching people to plan, check and steer their own learning. 1 claim Claims: Metacognitive strategies
Early childhood and language Shared reading and talk that grow vocabulary before formal school. 2 claims Claims: Interactive book reading , Preschool brain training
Language learning How people learn a second or foreign language — spacing, practice and what kind of teaching helps. 2 claims Claims: Spaced L2 practice , Explicit L2 teaching
Emotion, stress and arousal How anxiety, acute stress and arousal change what we can learn and bring back under test. 2 claims Claims: Test anxiety and results , Stress at the test
Learning outside school Homework, museums, field trips and other structured learning beyond the classroom. 2 claims Claims: Homework and achievement , Learning outside the classroom
Special education and inclusion How placement and targeted teaching affect pupils with special educational needs — what helps, and what does not. 2 claims Claims: Inclusive placement , Targeted help for special needs
Adult / workplace learning Training at work: what raises learning and what makes it transfer to the job. 1 claim Claims: Workplace training
Music and arts training Music, drama and the arts — what transfers beyond the art form itself, and what does not. 2 claims Claims: Music lessons and thinking skills (killed) , Music lessons and school results
Bilingualism and dual literacy Learning to read and learn in two languages — programmes for language-minority children and the skills that transfer. 1 claim Claims: Home language in school
How the beliefs connect Arrows show that one claim depends on or relates to another. Each claim shows its attacks as dots, oldest first.
Claim map The two guiding questions, the claims that serve them, and how the claims depend on or relate to each other. Newly added claims are grouped at the bottom and name their questions inside the box. A text version follows the diagram. Newly added, being tested (fewer than 3 survived attacks) depends-on depends-on relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to relates-to Q1 How do humans actually learn, andwhat helps them learn best? Q2 How does that work in practice: inschool, with screens, and with AI as one of the tools? C0: Retrieval beats restudy. Status narrowed. 14 attacks: 6 survived, 8 narrowed. C0 narrowed Retrieval beats restudy Attack 1 on C0, 2026-09-08: narrowed (Roediger & Karpicke 2006) Attack 2 on C0, 2026-09-08: survived (no named source (attacker's own inference)) Attack 3 on C0, 2026-09-08: survived (Dunlosky et al. 2013) Attack 4 on C0, 2026-09-08: survived (Bjork & Bjork 2020) Attack 5 on C0, 2026-09-08: survived (Roediger & Karpicke 2006) Attack 6 on C0, 2026-09-08: survived (Adesope et al. 2017) Attack 7 on C0, 2026-09-08: narrowed (Roediger & Karpicke 2006) Attack 8 on C0, 2026-09-08: narrowed (Dunlosky et al. 2013) Attack 9 on C0, 2026-09-08: narrowed (Butler & Roediger 2008) Attack 10 on C0, 2026-09-08: narrowed (Dunlosky et al. 2013) Attack 11 on C0, 2026-09-28: survived (Gupta, Pan & Rickard 2026) Attack 12 on C0, 2026-09-28: narrowed (McCarter, Huber & Cowell 2025) Attack 13 on C0, 2026-10-03: narrowed (Higham et al. 2023) Attack 14 on C0, 2026-10-05: narrowed (Abel et al. 2019) 14 attacks: 6 survived, 8 narrowed C8: Distracting extras. Status narrowed. 5 attacks: 3 survived, 2 narrowed. C8 narrowed Distracting extras Attack 1 on C8, 2026-10-08: narrowed (Maloy et al. 2019) Attack 2 on C8, 2026-10-10: survived (Cheng, Wu, Wang & Wang 2026) Attack 3 on C8, 2026-10-13: narrowed (Wesenberg et al. 2026) Attack 5 on C8, 2026-10-23: survived (Wesenberg, Krieglstein, Schneider & Rey 2025) Attack 7 on C8, 2026-10-27: survived (Bender, Renkl, Scheiter, Richter & Eitel 2026) 5 attacks: 3 survived, 2 narrowed C16: Practice explains only part. Status narrowed. 4 attacks: 3 survived, 1 narrowed. C16 narrowed Practice explains only part Attack 1 on C16, 2026-10-16: narrowed (Ericsson & Harwell 2019) Attack 2 on C16, 2026-10-16: survived (Macnamara & Maitra 2019) Attack 3 on C16, 2026-10-27: survived (Macnamara, Moreau & Hambrick 2016) Attack 4 on C16, 2026-10-27: survived (Platz, Kopiez, Lehmann & Wolf 2014) 4 attacks: 3 survived, 1 narrowed C19: Metacognitive strategies. Status narrowed. 4 attacks: 3 survived, 1 narrowed. C19 narrowed Metacognitive strategies Attack 1 on C19, 2026-10-27: survived (de Boer, Donker, Kostons & van der Werf 2018) Attack 3 on C19, 2026-10-27: survived (Eberhart, Ingendahl & Bryce 2025) Attack 4 on C19, 2026-10-27: survived (Tuero, Núñez, Vallejo, Fernández, Añón, Moreira, Martins & Rosário 2022) Attack 7 on C19, 2026-11-15: narrowed (Torgerson, Ainsworth, Bell, Elliott, Fountain, Gascoine, Hewitt, Kasim, Kokotsaki & Torgerson 2018; IPEELL: Evaluation report and executive summary, Education Endowment Foundation, London, November 2018 (independent evaluation, Durham University and University of York; full report PDF from the EEF project page read in full and stored privately)) 4 attacks: 3 survived, 1 narrowed C20: Interactive book reading. Status narrowed. 4 attacks: 3 survived, 1 narrowed. C20 narrowed Interactive book reading Attack 1 on C20, 2026-10-27: survived (Mol, Bus & de Jong 2009) Attack 2 on C20, 2026-10-27: narrowed (Noble, Sala, Peter, Lingwood, Rowland, Gobet & Pine 2019) Attack 3 on C20, 2026-10-27: survived (Mol, Bus, de Jong & Smeets 2008) Attack 4 on C20, 2026-10-27: survived (Dowdall, Melendez-Torres, Murray, Gardner, Hartford & Cooper 2020) 4 attacks: 3 survived, 1 narrowed C30: Music lessons and thinking skills (killed). Status killed. 2 attacks: 1 narrowed, 1 killed. C30 killed Music lessons and thinking skills (killed) Attack 1 on C30, 2026-10-31: narrowed (Bigand & Tillmann 2022) Attack 2 on C30, 2026-11-05: killed (Jamey, Foster, Hyde & Dalla Bella 2024) 2 attacks: 1 narrowed, 1 killed C2: Answer-giving AI in practice. Status narrowed. 7 attacks: 5 survived, 2 narrowed. C2 narrowed Answer-giving AI in practice Attack 1 on C2, 2026-09-27: survived (Bastani et al. 2025) Attack 2 on C2, 2026-09-27: survived (Barcaui 2025) Attack 3 on C2, 2026-09-29: survived (Kreijkes et al. 2026) Attack 4 on C2, 2026-09-29: survived (Bergh, Tag, Vassar & Renzella 2026 (preprint)) Attack 5 on C2, 2026-10-03: narrowed (Contractor & Reyes 2026 (preprint)) Attack 6 on C2, 2026-10-05: narrowed (Lira, Rogers, Goldstein, Ungar & Duckworth 2025 (preprint)) Attack 8 on C2, 2026-10-16: survived (Kazemitabaar, Chow, Ma, Ericson, Weintrop & Grossman 2023 (CHI '23)) 7 attacks: 5 survived, 2 narrowed C3: Paper vs screen, deep reading. Status narrowed. 7 attacks: 4 survived, 3 narrowed. C3 narrowed Paper vs screen, deep reading Attack 1 on C3, 2026-09-27: narrowed (Florit et al. 2025) Attack 2 on C3, 2026-09-28: survived (Li & Yan 2024) Attack 3 on C3, 2026-09-29: survived (Foss, Støle & Magyari 2026) Attack 4 on C3, 2026-09-30: survived (Ronconi, Altoè & Mason 2025) Attack 5 on C3, 2026-10-03: survived (Seo, Rhim & Gweon 2026) Attack 6 on C3, 2026-10-05: narrowed (Lauterman & Ackerman 2014) Attack 7 on C3, 2026-10-06: narrowed (Delgado & Salmerón 2021; Ackerman & Lauterman 2012) 7 attacks: 4 survived, 3 narrowed C6: Phones in lessons. Status narrowed. 5 attacks: 3 survived, 2 narrowed. C6 narrowed Phones in lessons Attack 1 on C6, 2026-10-07: narrowed (Allcott, Baron, Dee, Duckworth, Gentzkow & Jacob 2026 (preprint)) Attack 2 on C6, 2026-10-09: narrowed (Shi & Villarroel 2026 (preprint)) Attack 3 on C6, 2026-10-09: survived (Figlio & Özek 2025 (preprint)) Attack 4 on C6, 2026-10-11: survived (Lichand, Moreno-Louzada, da Costa & Gentzkow 2026 (preprint)) Attack 5 on C6, 2026-10-14: survived (Sungu, Choudhury & Bjerre-Nielsen 2026) 5 attacks: 3 survived, 2 narrowed C10: Tutoring software in school use. Status narrowed. 7 attacks: 4 survived, 3 narrowed. C10 narrowed Tutoring software in school use Attack 1 on C10, 2026-10-07: narrowed (Steenbergen-Hu & Cooper 2013) Attack 2 on C10, 2026-10-08: narrowed (Pane et al. 2014) Attack 3 on C10, 2026-10-10: narrowed (Pane et al. 2010) Attack 4 on C10, 2026-10-13: survived (Campuzano et al. 2009; NCEE 2009-4041 (ERIC ED504657)) Attack 5 on C10, 2026-10-15: survived (Feng, Huang & Collins 2023 (preprint); WestEd research report; AIED 2023 late-breaking result; ERIC ED658161) Attack 6 on C10, 2026-10-16: survived (Wijekumar, Meyer & Lei 2012; Wijekumar, Meyer, Lei, Lin, Johnson, Spielvogel, Shurmatz, Ray & Cook 2014) Attack 7 on C10, 2026-11-15: survived (Leite, Zhang, Rana, Hao, Busch, Kong & Kuang 2026) 7 attacks: 4 survived, 3 narrowed C1: What counts as learned. Status definition. 7 attacks: 3 survived, 4 narrowed. C1 definition What counts as learned Attack 1 on C1, 2026-09-14: narrowed (Roediger & Karpicke 2006) Attack 2 on C1, 2026-09-14: narrowed (Tulving & Pearlstone 1966) Attack 3 on C1, 2026-09-14: survived (Karpicke & Blunt 2011) Attack 4 on C1, 2026-09-14: survived (Rowland 2014) Attack 5 on C1, 2026-09-14: narrowed (Morris, Bransford & Franks 1977) Attack 6 on C1, 2026-10-01: narrowed (Gupta, Pan & Rickard 2022) Attack 7 on C1, 2026-10-02: survived (Racsmány, Szőllősi & Marián 2020) 7 attacks: 3 survived, 4 narrowed C4: Task feedback vs praise. Status narrowed. 2 attacks: 1 survived, 1 narrowed. C4 narrowed Task feedback vs praise Attack 1 on C4, 2026-10-16: narrowed (Butler & Nisan 1986) Attack 3 on C4, 2026-11-14: survived (Abe & Usami 2024) 2 attacks: 1 survived, 1 narrowed · serves Q1 C5: Step-based tutoring. Status alive. 1 attack: 1 survived. C5 alive Step-based tutoring Attack 1 on C5, 2026-10-10: survived (Steenbergen-Hu & Cooper 2014) 1 attack: 1 survived · serves Q2 C7: Handwritten notes. Status narrowed. 3 attacks: 3 narrowed. C7 narrowed Handwritten notes Attack 1 on C7, 2026-10-07: narrowed (Urry et al. 2021) Attack 2 on C7, 2026-10-17: narrowed (Artz, Johnson, Robson & Taengnoi 2020) Attack 3 on C7, 2026-10-21: narrowed (Morehead, Dunlosky & Rawson 2019) 3 attacks: 3 narrowed · serves Q1+Q2 C9: One laptop per pupil. Status narrowed. 1 attack: 1 narrowed. C9 narrowed One laptop per pupil Attack 1 on C9, 2026-10-08: narrowed (Hull & Duch 2019) 1 attack: 1 narrowed · serves Q2 C11: Spacing and interleaving. Status narrowed. 4 attacks: 2 survived, 2 narrowed. C11 narrowed Spacing and interleaving Attack 1 on C11, 2026-10-15: narrowed (Kremer, Gray-Lobe, de Laat & van der Haar 2026 (preprint)) Attack 2 on C11, 2026-10-16: narrowed (Ebersbach & Barzagar Nazari 2020) Attack 3 on C11, 2026-10-17: survived (Murray, Horner & Göbel 2025) Attack 4 on C11, 2026-11-03: survived (Ruitenburg, Goos, Lambert, Kirschner, Jarodzka & Camp 2026) 4 attacks: 2 survived, 2 narrowed · serves Q1 C29: Preschool brain training. Status narrowed. 1 attack: 1 narrowed. C29 narrowed Preschool brain training Attack 1 on C29, 2026-10-31: narrowed (Birtwistle, Chernikova, Wünsch & Niklas 2025) 1 attack: 1 narrowed · serves Q1 C35: Music lessons and school results. Status alive. No attacks yet. C35 alive Music lessons and school results untested no attacks yet · serves Q1 C36: Phone bans and lower achievers. Status alive. No attacks yet. C36 alive Phone bans and lower achievers untested no attacks yet · serves Q2 C12: Systematic phonics. Status narrowed. 2 attacks: 1 survived, 1 narrowed. C12 narrowed Systematic phonics Attack 1 on C12, 2026-10-16: narrowed (Torgerson, Brooks & Hall 2006; Department for Education and Skills Research Report RR711 (ISBN 1 84478 659 6)) Attack 2 on C12, 2026-10-20: survived (Mathes, Denton, Fletcher, Anthony, Francis & Schatschneider 2005) 2 attacks: 1 survived, 1 narrowed · serves Q1 C13: Growth mindset for struggling students. Status narrowed. 1 attack: 1 narrowed. C13 narrowed Growth mindset for struggling students Attack 1 on C13, 2026-10-16: narrowed (Macnamara & Burgoyne 2023) 1 attack: 1 narrowed · serves Q1 C14: Later school start. Status narrowed. 3 attacks: 2 survived, 1 narrowed. C14 narrowed Later school start Attack 1 on C14, 2026-10-16: survived (James, Erickson, Lammert & Widome 2023) Attack 2 on C14, 2026-10-17: narrowed (Thacher & Onyper 2016) Attack 4 on C14, 2026-10-22: survived (Dou, Gihleb, Giuntella & Lonsky 2026 (preprint)) 3 attacks: 2 survived, 1 narrowed · serves Q1 C15: Learning by teaching. Status narrowed. 1 attack: 1 narrowed. C15 narrowed Learning by teaching Attack 1 on C15, 2026-10-25: narrowed (Lachner, Russ, Hübner, Sibley & Scheiter 2025) 1 attack: 1 narrowed · serves Q1 C17: Background knowledge. Status narrowed. 5 attacks: 1 survived, 4 narrowed. C17 narrowed Background knowledge Attack 1 on C17, 2026-10-16: survived (Grissmer et al. 2023 (preprint)) Attack 2 on C17, 2026-10-18: narrowed (See, Gorard & Siddiqui 2017) Attack 3 on C17, 2026-10-22: narrowed (Cabell, Kim, White, Gale, Edwards, Hwang, Petscher & Raines 2025) Attack 5 on C17, 2026-10-30: narrowed (Kim, Burkhauser, Mesite, Asher, Relyea, Fitzgerald & Elmore 2021) Attack 7 on C17, 2026-11-04: narrowed (Kim & Cao 2025) 5 attacks: 1 survived, 4 narrowed · serves Q1 C18: AI deskilling. Status narrowed. 1 attack: 1 narrowed. C18 narrowed AI deskilling Attack 1 on C18, 2026-10-16: narrowed (Pedersen et al. 2026) 1 attack: 1 narrowed · serves Q2 C21: Sleep before learning. Status narrowed. 3 attacks: 1 survived, 2 narrowed. C21 narrowed Sleep before learning Attack 1 on C21, 2026-10-27: survived (Cousins, Wong & Chee 2019) Attack 2 on C21, 2026-10-27: narrowed (Lo, Bennion & Chee 2016) Attack 4 on C21, 2026-11-01: narrowed (Crowley, Alderman, Javadi & Tamminen 2024) 3 attacks: 1 survived, 2 narrowed · serves Q1 C22: Tangible rewards and interest. Status alive. 1 attack: 1 survived. C22 alive Tangible rewards and interest Attack 1 on C22, 2026-10-27: survived (Cameron, Banko & Pierce 2001) 1 attack: 1 survived · serves Q1 C23: Spaced L2 practice. Status narrowed. 1 attack: 1 narrowed. C23 narrowed Spaced L2 practice Attack 1 on C23, 2026-11-05: narrowed (Kim & Webb 2022) 1 attack: 1 narrowed · serves Q1 C24: Explicit L2 teaching. Status alive. No attacks yet. C24 alive Explicit L2 teaching untested no attacks yet · serves Q1 C25: Test anxiety and results. Status alive. 1 attack: 1 survived. C25 alive Test anxiety and results Attack 1 on C25, 2026-11-03: survived (Caviola, Toffalini, Giofrè, Mercader Ruiz, Szűcs & Mammarella 2021) 1 attack: 1 survived · serves Q1 C26: Stress at the test. Status narrowed. 1 attack: 1 narrowed. C26 narrowed Stress at the test Attack 2 on C26, 2026-11-04: narrowed (Shields, Sazma, McCullough & Yonelinas 2017) 1 attack: 1 narrowed · serves Q1 C27: Homework and achievement. Status narrowed. 1 attack: 1 narrowed. C27 narrowed Homework and achievement Attack 1 on C27, 2026-11-03: narrowed (Guo, Li, Xu, Hu, Liu, Xing, Li, White & Yang 2024) 1 attack: 1 narrowed · serves Q1 C28: Learning outside the classroom. Status narrowed. 1 attack: 1 narrowed. C28 narrowed Learning outside the classroom Attack 2 on C28, 2026-11-08: narrowed (Delen, Özüdoğru, Şen & Biasutti 2026) 1 attack: 1 narrowed · serves Q1 C31: Inclusive placement. Status narrowed. 1 attack: 1 narrowed. C31 narrowed Inclusive placement Attack 1 on C31, 2026-10-31: narrowed (Krämer, Möller & Zimmermann 2021) 1 attack: 1 narrowed · serves Q1 C32: Targeted help for special needs. Status narrowed. 2 attacks: 2 narrowed. C32 narrowed Targeted help for special needs Attack 1 on C32, 2026-11-01: narrowed (Hall, Dahl-Leonard, Cho, Solari, Capin, Conner, Henry, Cook, Hayes, Vargas, Richmond & Kehoe 2023) Attack 3 on C32, 2026-11-07: narrowed (Van Herwegen, Masterman, Dockrell, Marshall, Gordon & Thomas 2026) 2 attacks: 2 narrowed · serves Q1 C33: Workplace training. Status alive. No attacks yet. C33 alive Workplace training untested no attacks yet · serves Q1 C34: Home language in school. Status narrowed. 2 attacks: 1 survived, 1 narrowed. C34 narrowed Home language in school Attack 1 on C34, 2026-11-01: narrowed (Slavin, Madden, Calderón, Chamberlain & Hennessy 2011) Attack 3 on C34, 2026-11-14: survived (Valentino & Reardon 2015) 2 attacks: 1 survived, 1 narrowed · serves Q1 survived narrowedone dot per attack, oldest first
Claim map as text Q1. How do humans actually learn, and what helps them learn best? Served by: C0 (Retrieval beats restudy) : 14 attacks: 6 survived, 8 narrowed, C1 (What counts as learned) : 7 attacks: 3 survived, 4 narrowed, C4 (Task feedback vs praise) : 2 attacks: 1 survived, 1 narrowed, C7 (Handwritten notes) : 3 attacks: 3 narrowed, C8 (Distracting extras) : 5 attacks: 3 survived, 2 narrowed, C11 (Spacing and interleaving) : 4 attacks: 2 survived, 2 narrowed, C12 (Systematic phonics) : 2 attacks: 1 survived, 1 narrowed, C13 (Growth mindset for struggling students) : 1 attack: 1 narrowed, C14 (Later school start) : 3 attacks: 2 survived, 1 narrowed, C15 (Learning by teaching) : 1 attack: 1 narrowed, C16 (Practice explains only part) : 4 attacks: 3 survived, 1 narrowed, C17 (Background knowledge) : 5 attacks: 1 survived, 4 narrowed, C19 (Metacognitive strategies) : 4 attacks: 3 survived, 1 narrowed, C20 (Interactive book reading) : 4 attacks: 3 survived, 1 narrowed, C21 (Sleep before learning) : 3 attacks: 1 survived, 2 narrowed, C22 (Tangible rewards and interest) : 1 attack: 1 survived, C23 (Spaced L2 practice) : 1 attack: 1 narrowed, C24 (Explicit L2 teaching) : No attacks yet, C25 (Test anxiety and results) : 1 attack: 1 survived, C26 (Stress at the test) : 1 attack: 1 narrowed, C27 (Homework and achievement) : 1 attack: 1 narrowed, C28 (Learning outside the classroom) : 1 attack: 1 narrowed, C29 (Preschool brain training) : 1 attack: 1 narrowed, C30 (Music lessons and thinking skills (killed)) : 2 attacks: 1 narrowed, 1 killed, C31 (Inclusive placement) : 1 attack: 1 narrowed, C32 (Targeted help for special needs) : 2 attacks: 2 narrowed, C33 (Workplace training) : No attacks yet, C34 (Home language in school) : 2 attacks: 1 survived, 1 narrowed, C35 (Music lessons and school results) : No attacks yetQ2. How does that work in practice: in school, with screens, and with AI as one of the tools? Served by: C2 (Answer-giving AI in practice) : 7 attacks: 5 survived, 2 narrowed, C3 (Paper vs screen, deep reading) : 7 attacks: 4 survived, 3 narrowed, C5 (Step-based tutoring) : 1 attack: 1 survived, C6 (Phones in lessons) : 5 attacks: 3 survived, 2 narrowed, C7 (Handwritten notes) : 3 attacks: 3 narrowed, C8 (Distracting extras) : 5 attacks: 3 survived, 2 narrowed, C9 (One laptop per pupil) : 1 attack: 1 narrowed, C10 (Tutoring software in school use) : 7 attacks: 4 survived, 3 narrowed, C18 (AI deskilling) : 1 attack: 1 narrowed, C36 (Phone bans and lower achievers) : No attacks yetC0 depends-on C1 C2 depends-on C1 C3 relates-to C1 C4 relates-to C0 C5 relates-to C2 C5 relates-to C4 C5 relates-to C10 C6 relates-to C9 C6 relates-to C36 C7 relates-to C3 C8 relates-to C6 C11 relates-to C0 C11 relates-to C2 C25 relates-to C26 C27 relates-to C28 C29 relates-to C30 C30 relates-to C35 C31 relates-to C32 C36 relates-to C6 C36 relates-to C9 Browse all areas → · Browse all sources → · Where it's weakest →