2026-05-29T16:43:43Z https://yabu.me sean_ae https://yabu.me/npub1j7hqvlftgffj4ueh4vl85dy0krf3x2lact3344a6uctt9kkuqhmstt8x2e https://post.lurk.org/system/accounts/avatars/000/185/936/original/197be651281d1396.jpeg https://post.lurk.org/system/accounts/avatars/000/185/936/original/197be651281d1396.jpeg https://yabu.me/nevent1qqsfyk7cjlffn9nm4n2q96erakr4rfwt5l6rjj004cvkwxzgaduyewczyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwtpr45g dammit<br/> <img src="https://post.lurk.org/system/media_attachments/files/116/696/769/047/604/478/original/f14aa2ffe46ac8ef.png"> <br/> 2026-06-05T09:24:02Z https://yabu.me/nevent1qqsgkef7q437ytak238ht6ymhan6jk630akp3zjm0akxd3yn5adp9gszyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwng9ucs In reply to <a href='/nevent1qqsf3xf56ttlc0sw8lx9cjypfmm8h8gr8eex0jpvwgtqe6q0sxg9x2gn99rae'>nevent1q…9rae</a><br/>_________________________<br/><br/>well sure, but complex systems are not just their constituent parts are they 2026-05-06T19:29:13Z https://yabu.me/nevent1qqs0vp8gjwxpuhn42zluz4gwtmjgvg9tycrf08azpcfln3688xrpcqczyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwqjzanh In reply to <a href='/nevent1qqsdgqcw9rryeueudfr8zmgmaqly2hw0t7n62psll83mks75kc9dwwglj392e'>nevent1q…392e</a><br/>_________________________<br/><br/>i see a clear difference between curating and designing though<br/><br/>i mean you can either be very careful with curation or not careful at all, depends what it is you&#39;re trying to do<br/><br/>and i think <span itemprop="mentions" itemscope itemtype="https://schema.org/Person"><a itemprop="url" href="/npub1frqcduvp5g4cpmu4yuqj7qje4vnz3l0g72fhaewccfr39xeezx7qcjkzu5" class="bg-lavender dark:prose:text-neutral-50 dark:text-neutral-50 dark:bg-garnet px-1"><span>emenel</span> (<span class="italic">npub1frq…kzu5</span>)</a></span> is right that most of the big LLM companies are working hard to achieve their objectives via tuning and curation, but i&#39;m not buying the notion of the resulting models being designed<br/><br/>they&#39;re extremely black boxy, esp at that kind of scale<br/><br/>you can see this esp with grok because every time they try to rein it in it just becomes a new kind of unhinged 2026-05-06T19:21:48Z https://yabu.me/nevent1qqsrs3eklsp22w6eky3kh7uc3h3tpqgtyuquy7ytuptaa76nhrmrmeqzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwju45t4 In reply to <a href='/nevent1qqst40uwzeyupmjpu3hdyj5w9g69j53wk95u9gq8hswrhd6etdhu2cscafl7x'>nevent1q…fl7x</a><br/>_________________________<br/><br/>yeah all of those things will affect your observation - but it&#39;s still an observation<br/><br/>you can say the same for any kind of analysis, recording, data gathering and whatnot (at least at the macro scale)<br/><br/>but none of that implies design<br/><br/>and the data you observe will exist regardless of whether or not you choose to observe it, or however you choose to<br/><br/>i suppose this is just a difference in philosophical approaches or something, we&#39;re prob never gonna agree :D 2026-05-06T17:43:29Z https://yabu.me/nevent1qqsd7uvrjr2jm7j4dcf4ay47e96qnwh5s0lsf6wya04sganatjqpwsqzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwngzwaz In reply to <a href='/nevent1qqswh3a7rna4jj97t8vuas8y730m0kmazwysralu54ensjpvameaxpcsy6h7a'>nevent1q…6h7a</a><br/>_________________________<br/><br/>ok sure, i guess you *could* say that performing an analysis on some data is in some way creating something<br/><br/>i don&#39;t see it that way personally, i see it as an act of observation<br/><br/>i think the thing you are observing exists regardless of whether you choose to observe it or not<br/><br/>what you&#39;re describing is more akin to editing than writing 2026-05-06T17:28:34Z https://yabu.me/nevent1qqsrh7waypm3epmcsg6vrg87u7z2t4z0ru8n5xk4lp7uvcl7pdkc7nqzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlw0fkkaq In reply to <a href='/nevent1qqsrpcrgt7qmny7fr8755eljuy95hnls7xc7j3xx2zy0acg288fkqwgfutcld'>nevent1q…tcld</a><br/>_________________________<br/><br/>no one said it was neutral though, and lack of neutrality (whatever neutrality is supposed to be) isn&#39;t in itself an indication of design<br/><br/>i may just be having a hard time understanding what you mean by these terms like &#39;natural&#39;, &#39;neutral&#39; and &#39;objective&#39; 2026-05-06T17:14:30Z https://yabu.me/nevent1qqsv4ptgzm49kylkp8l2vlyna6nw4sxf9ekp64r3d0eun0u47cwd24szyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwjhdwc6 In reply to <a href='/nevent1qqsgfkek8mu9hmsavfnwz78zmnn5kardvkkve0lhqft4n0jksscc48gmfh7n9'>nevent1q…h7n9</a><br/>_________________________<br/><br/>yeah, if you carry out some analysis you will get the results of that analysis<br/><br/>idk what you mean by <br/>&gt;natural representation of the data 2026-05-06T17:04:23Z https://yabu.me/nevent1qqsyd9smjtyysrggej9y0lvswuxd8wdpvyecw50jaspx29mssfh8h3czyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwcc7yxd In reply to <a href='/nevent1qqsfh47yn22ug7uzvlqasd04vane9fq09sl0fg4du00c6qfyhwrw05s2ag6sh'>nevent1q…g6sh</a><br/>_________________________<br/><br/>i mean this is why we repeat experiments isn&#39;t it?<br/><br/>i&#39;m not really understanding why you think the results of [some analysis] are themselves designed - i get that the analysis may be - but the results?<br/><br/>and you seem to be disregarding that the data itself was not designed either<br/><br/>and/or you seem to be implying that &#39;selected&#39; and &#39;designed&#39; are the same thing, and i don&#39;t think they are 2026-05-06T16:07:38Z https://yabu.me/nevent1qqsgtj46nket0wka6mvwhfqyrgcl92w0fx4rcw2xwyfhvj9f0dwrlcszyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwqku98c In reply to <a href='/nevent1qqsgsqztcy8zvcudn9wsnufalfqgw3lhzlr5wyeketlgvy7e8dvds7snvfgww'>nevent1q…fgww</a><br/>_________________________<br/><br/>well you can tune the weights later on anyway (just ignore the ones you don&#39;t want to use) <br/><br/>but yeah you&#39;re performing [some analysis] and then working with that data<br/><br/>i mean, what you seem to be saying is that the output of any statistical analysis is designed - and this is kinda like saying the data you get from doing a scientific experiment is designed - and idk if you can really say that<br/><br/>i mean you can say the experiment was designed but the data is just [the data you got from that experiment] 2026-05-06T15:41:16Z https://yabu.me/nevent1qqs9kf3m86n2urcnfhvar7npf033apztnt8u7c9fw4w3sleyv7s9txszyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwxk22mn In reply to <a href='/nevent1qqspccg23hzphu6k3054fvf5pkf4hhkrtyj42jyw6fnw607gazdc44sxjst5p'>nevent1q…st5p</a><br/>_________________________<br/><br/>well yeah, i guess there are levels of arbitrary<br/><br/>i can work with some pretty arbitrary sound sets but there are always gonna be questions about how or why files ended up in the set - i get that<br/><br/>i just find this idea that [the model is therefore designed] to be a bit weak, but maybe we can agree on &#39;weakly designed&#39; or something<br/><br/>and i obvs get where you&#39;re coming from wrt the bigger LLMs being horribly skewed and manipulated - but to me there&#39;s a big difference between manipulating something and designing it 2026-05-06T15:29:13Z https://yabu.me/nevent1qqs9czquw6epnpfct5vf6pun5c7geytnqwhn3vucps9dp77vlj0n3nszyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwwtqg4p In reply to <a href='/nevent1qqsp94ce6wdjshx80a83ywxukxgaskvja476jqavz3duuej8tn7w4hqp0nj36'>nevent1q…nj36</a><br/>_________________________<br/><br/>but what if the data is arbitrary? <br/><br/>(sorry for repeating myself)<br/><br/>i realise &#39;arbitrary data&#39; is a very spherical cow here, but just hypothetically, if you had arbitrary data - would you still consider the model to have been designed? 2026-05-06T15:07:02Z https://yabu.me/nevent1qqsre622e2zmvlpnxvt3zqsmv6j6s2ew6q0auwekpavk3yklvlmvylczyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlw03hg8g In reply to <a href='/nevent1qqsru4xmywg9hctt05a3rgvg66hwwlef0lttu9yccgc29hf3jjvgrtc7qasky'>nevent1q…asky</a><br/>_________________________<br/><br/>the network is the thing you are training, the model is a result of training that network<br/><br/>i suppose what you are saying is that since networks are designed, some of that design is affecting the quality of the model (and it is, but i would not call the model designed - otherwise they would actually work, and they frequently don&#39;t) 2026-05-06T14:31:23Z https://yabu.me/nevent1qqs9cc2tqz9yg0an752lxa9nw38za3mygaw9tkz5nvuqzlw2pcry4ugzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwy87n9n In reply to <a href='/nevent1qqs97cmeh79zsmvd2endg386flsw20ux438rgem3477jyyg2arej5tgpxdwmk'>nevent1q…dwmk</a><br/>_________________________<br/><br/>ah ok, i see<br/><br/>when you say &#39;model&#39; you mean what i would call a network 2026-05-06T14:25:47Z https://yabu.me/nevent1qqsx2chkhp8j9chf32tgwtscplj4wzsdvyg3ylrjts5p89yuq2mra6qzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwlf05wc In reply to <a href='/nevent1qqsruhqeje44hw905zscrfffce6stwc698sr60g7ara22lm53lckk2szvav8y'>nevent1q…av8y</a><br/>_________________________<br/><br/>can you explain a bit more what you mean there?<br/><br/>i don&#39;t have to specify anything when training a network locally, i can just chuck any data in (sounds/gestures/images/text), not label anything and get data out the other end 2026-05-06T14:17:11Z https://yabu.me/nevent1qqsqjh4j5j5wzsnsxl63jkwfcwf3yc6c4us3vv6l40wj6whs95j2wkqzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlw794a5f In reply to <a href='/nevent1qqsgcgadj6ya3036n2q5dr6xlwf6usey8gls2njxuw4spvha5gk3s3sprwdur'>nevent1q…wdur</a><br/>_________________________<br/><br/>hmm idk about this - you seem to have ignored the word &#39;arbitrary&#39; in my question 2026-05-06T14:13:55Z https://yabu.me/nevent1qqs8jfhulcnma4upx8nhxzxu4m2lrx8h2zs4ldmhvvfawer759gw3nszyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwlh966e In reply to <a href='/nevent1qqsqld6t2j4jltjgxwsz2l5jq9wtxpawjre79uc97nwnzen0gxavxzqmuyvu3'>nevent1q…yvu3</a><br/>_________________________<br/><br/>yeah ok, let me reframe this<br/><br/>if you didn&#39;t do any curation at all - and just trained a network on some arbitrary data - could the model be described as having been designed? 2026-05-06T13:59:31Z https://yabu.me/nevent1qqszaskgq6yrr8784amn95d9tgvnup0egmrvy4jx67q38h7yqtfv8tgzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwppk9wm In reply to <a href='/nevent1qqs8v2tgz5fyv9rs5jx7sryu5fe2jeqw0vwylwdf30aezhl5z5flxzssjllp5'>nevent1q…llp5</a><br/>_________________________<br/><br/>if it&#39;s entirely designed then why are people having debates about mass scraping of art etc? 2026-05-06T13:50:27Z https://yabu.me/nevent1qqs82gt0fc9qkt5lyfgv75w7sxgv09aqx5hdxvxra0f5ygf95j3uftczyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwr4rac8 In reply to <a href='/nevent1qqsypg249t5uwkydvulrrnffecqf7v62sqh3hpjt6qanuw0gyk9tqcsvkk3uz'>nevent1q…k3uz</a><br/>_________________________<br/><br/>right, but what you&#39;re saying there is that the design rests entirely in the commercial application of [thing] - but the application of what [thing] exactly? 2026-05-06T13:43:55Z https://yabu.me/nevent1qqsqhka294u7lg5uq3g33vkx3rtyp8czax7g4z028jn6hzdf0j6d6jqzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwqklkeq In reply to <a href='/nevent1qqsqr96nkf9su8nv4a38eg8ngu5ncq3kykk4t3m4s922a62pxuzq8rcweq4mu'>nevent1q…q4mu</a><br/>_________________________<br/><br/>idk if you can say designed or controlled really, maybe slightly<br/><br/>you could maybe say built<br/><br/>it&#39;s not really a set of instructions, more a set of weights (most of which are not being set intentionally unless you only include human-generated data and even that would be a bit of a stretch imo)<br/><br/>not often i end up in a semantic hole, pls forgive me 2026-05-06T13:34:58Z https://yabu.me/nevent1qqsd5waeydxy89ght2hrdclc9jpw6dwhmvvcnm36kn5vmh6tma9sy4szyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwfqcdlf In reply to <a href='/nevent1qqswl9vlle926rymrhdv2k9x2exvaxqjzarpk7vz5clrrrtageje2fc0a07lw'>nevent1q…07lw</a><br/>_________________________<br/><br/>idk - do they even qualify as technology?<br/><br/>they&#39;re weird, statistical models, not designed or controlled, they just kind of *are* 2026-05-06T13:04:44Z https://yabu.me/nevent1qqsfs3y4l85rf4fjsvr8636n9rvwldx6a8yz84t8sxpls3krp29nceszyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlw7pkjd8 In reply to <a href='/nevent1qqsydhlt340g8d8vemn7w0y70p3gm7g0yqkafacm3scsww7cqvvpgqsmr5n6m'>nevent1q…5n6m</a><br/>_________________________<br/><br/>i see people describe LLMs as algorithms quite often but i&#39;m pretty sure that&#39;s not the right word <br/><br/>i suppose they can *mimic* algorithms though 2026-05-06T11:52:53Z https://yabu.me/nevent1qqsw8e55spzn697t2advpdqrl7nmp0du3wsje2l8m6r9alm74ktwzfqzyzt6upna9dp9x2hnx74nu73537cdxyetlhpwxxkhhtnpdvk6mszlwpctm7g in celebration of confield&#39;s 25 year anniversary we will be releasing 25,000 ball bearings into a gong factory 2026-04-30T12:24:50Z