TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

A Formal Apology for Metaphysics [pdf]

34 点作者 lainon超过 6 年前

5 条评论

dvt超过 6 年前
Probably one of the few people that read the whole thing, but the argument is, in my opinion, pretty weak. It basically boils down to making some kind of equivalence between metaphysics and pure mathematics. Of course, it ignores the main force of mathematical progress: theorem-building. Metaphysics might be &quot;interconnected in much the same way&quot; as Dr. Baron argues, but let&#x27;s not kid ourselves. To argue that metaphysics is as &quot;conceptually interrelated&quot; as mathematics is is a real stretch. Just consider volume: it&#x27;s estimated that around <i>a quarter of a million</i> theorems are proved every year. That&#x27;s real, measurable, progress; what&#x27;s the metaphysics equivalent?<p>The distinction between &quot;internally applied metaphysics&quot; and &quot;externally applied metaphysics&quot; on page 14 is a shameless red herring. The <i>only</i> thing we ought to care about, given the preceding section on applications of pure mathematics is &quot;externally applied metaphysics,&quot; but in typical philosopher fashion, Dr. Baron equivocates for a few unnecessary pages. Finally, let me leave you with this gem: &quot;All three cases are examples in which scientists appear to be doing metaphysics. Computer scientists build models of objects and categories in order to provide the resources for artificial intelligence to successfully navigate the world, or make judgements [sic] about it.&quot; Yeah, that&#x27;s not metaphysics; it&#x27;s pretty much just regular taxonomy (and, by the way, the term of art is <i>labeling</i>). TensorFlow models have nothing to do with metaphysical models, and this is exactly why philosophers get such a bad rap. They extend far beyond the reaches of their knowledge base and don&#x27;t even have the courtesy to look up &quot;machine learning model&quot; on Wikipedia.
评论 #18746626 未加载
评论 #18746368 未加载
评论 #18745540 未加载
评论 #18745319 未加载
评论 #18746834 未加载
评论 #18746050 未加载
AltruisticGap超过 6 年前
Maybe readers will enjoy this related post from Bernardo Kastrup. It may argue the point better, or not. I can&#x27;t argue myself as it&#x27;s too obvious for me. It&#x27;s a unwinnable argument in my opinion, since it drives right into what it means to know something.<p>Why dismissing philosophy threatens the integrity of science<p><a href="https:&#x2F;&#x2F;www.bernardokastrup.com&#x2F;2018&#x2F;12&#x2F;why-dismissing-philosophy-threatens.html" rel="nofollow">https:&#x2F;&#x2F;www.bernardokastrup.com&#x2F;2018&#x2F;12&#x2F;why-dismissing-philo...</a>
评论 #18746372 未加载
gglon超过 6 年前
Meta theories should be evaluated by their fruits, theories they help cognitive system to discover. By how good the theories are and how easy it is to discover them. For example: instead of teaching a child a great bunch of theories one can teach it a simple&#x2F;short meta-theory and see if the child, based on its experience with the world, discovers a greater bunch good theories.
klyrs超过 6 年前
This isn&#x27;t an apology, it&#x27;s blame-shifting.<p>&quot;Don&#x27;t hit people! Apologize to your sibling NOW!&quot;<p>&quot;Sorry mom, my sibling plays baseball and gets congratulated whenever they hit the ball; so we&#x27;re really no different&quot;
josh_fyi超过 6 年前
The actual justification for non-directly-observable metaphysics is Occam&#x27;s Razor.
评论 #18746288 未加载