出版与作家生涯2011radar.oreilly.com编辑精选
我如何自动化自己的写作生涯
How I automated my writing career
导读
这篇 O'Reilly 文章介绍作者如何用自动写作软件（Automated Insights）来「自动化」自己的写作生涯，评论区则把话题引向更有价值的方向：哪些写作会被机器取代、哪些不会。对写作者来说，它值得一读，因为它逼你思考自己写的东西究竟属于「可替代的信息搬运」还是「不可替代的风格与判断」，并顺带给出用数据管理写作进度的实用做法。
可以直接用的写作要点
- 01
把写作分成两层：新闻快讯、清单、摘要、教程这类「内容重于风格」的稿子最容易被自动化，而小说、特稿、人情味故事、依赖专业判断的评论才是难以替代的部分。
- 02
标题不要承诺正文做不到的事——用夸张标题换点击，读者读完后只会觉得被欺骗。
- 03
用电子表格记录每日写作量等指标，据此预测交稿日期并对照进度，能清楚看到自己是超前还是落后。
- 04
靠拼凑信息加耸动标题的量产写作虽然来量快，但廉价、一次性、可替代，别把它当成长期饭碗。
HN 评论精选
写作会分成「可自动化的信息型」和「难替代的风格型」两层，写作者要往后者走。
我做个预测，姑且听之：未来写作行业会分裂成越来越多细分领域。最底层的「低价值写作」是软件终能自动生成的那类文章：新闻快讯、信息汇总、操作指南、清单、摘要等，传统新闻业的「新闻」和杂志的「信息型」稿件多属此类。这类文章内容重于风格，讲的是事实或事实的归纳；在内容农场那里，则是用煽动性标题不断重组信息。你不需要普利策级别的作者来产出这些，很快甚至不需要人来产出。这类写作报酬低并不奇怪——它廉价、对读者是一次性的，所以付得也少。另一头的高价值写作是不易被自动化、风格与内容同等重要的写作：好小说、特稿、人情味故事、依赖专业知识的社论等。这类写作要么报酬很低，要么报酬很高，取决于写作者……
jonnathanson查看英文原文
Just a prediction here, so take it for what it's worth, but my hunch is that the profession of writing is going to fragment into more and more subsets in the coming years. At the low end of the totem pole -- what we'll call "low-value" writing -- will be the sorts of articles that software can eventually automate. Things like news updates, information dumps, how-to pieces, lists, summaries, and so forth. Much of what traditional journalism would call "news" stories, and what magazine journalism would call "informational" pieces, fall into this bucket. In these sorts of articles, substance is more important than style. These pieces are all about the facts, or summations of the facts. Or, in the case of content farms, they're about relaying and recombining information in endless mixes, using provocative headlines. You don't need a Pulitzer-caliber author to crank these out. Hell, pretty soon you won't even need a human to crank them out. It's no surprise that this type of writing doesn't pay well, because frankly, it's the fast food of journalism. It's cheap, it's disposable to the consumer, and so it pays cheaply. On the other hand, higher-value writing will be that which isn't easily automated, and for which style is every bit as important as substance. Fiction (good fiction, at least), features, human-interest stories, editorials (especially those relying on expertise), and so forth. This will be the kind of writing that either pays crap, or pays big, depending on the writer's skill level -- and his or her ability to build a market or following for it. There will always be a need for this kind of writing, and until such time as software AI becomes genuinely creative, it'll be very hard to automate the highest-quality, most interesting, and most innovative stuff. Low-value writing will, if anything, see its value decline even further. It is the equivalent of the man on the assembly line who can be replaced by a tireless, hyper-efficient machine. High-value writing will not, on average, find itself paying more handsomely than it used to. It will still be a high-variance profession. But it will be what remains for professional writers in the age of content farms, automated news, social networking, and so forth. Essentially, the way to earn a decent living in the future will be: 1) be damned good, 2) build and maintain a following, 3) differentiate yourself, and 4) produce at high volume.
标题夸大其词；能自动生成段落不等于能写整本书，小心一开始就把目标定错。
那篇文章的标题非常不准确。我们的软件现在能生成八个段落，但能生成八章的内容吗？答案是「能」，但至少目前还不是我以前写的那种技术书。……这也是上一段说「因为我一直忙于经营 Automated Insights，最近没时间写新书」的原因。这有点像《卡尔文与霍布斯》里那个铺床机器人：我强烈怀疑他最终会像卡尔文一样，做出来的东西并不按计划运转。运气好的话，他会像卡尔文那样发现自己达成了目标，却发现一开始的目标就定错了。
gvb查看英文原文
That was a very inaccurate headline on the story. Our software can create eight paragraphs now, but is it possible to create eight chapters' worth of content? The answer is "yes," but not quite the same kind of technical books I used to write, at least right now. ...which is why the previous paragraph says "[b]ecause I've been so focused on running Automated Insights, I haven't had time to write any new books recently." That is a variant on the Calvin and Hobbes bed making robot[1]. I strongly suspect he will end up like Calvin, with something that doesn't work as planned. If he is lucky, like Calvin he will find he accomplished his goal, but discover he started with the wrong goal in mind. [1] http://books.google.com/books?id=NV4WEqQtvTYC&pg=PA126&lpg=PA126&dq=calvin+and+hobbes+bed+robot&source=bl&ots=xev_NyjpE1&sig=4cu_iiiCbl4jYAAJ1fL28akEUEQ&hl=en&ei=rcayTpPWBa-OsALw06z6Aw&sa=X&oi=book_result&ct=result&resnum=5&ved=0CDkQ6AEwBA#v=onepage&q&f=false
用数据指标记录写作量，可以预测交稿时间并实时对照进度。
稍微跑题：我写《The Geek Atlas》时做的一件事，是记录自己写作的各项数据指标，好知道进度在哪，再用这些指标向 O'Reilly 预测交稿日期、衡量自己相对截止日期的进展。全部用一个电子表格完成，让我能看出自己是超前还是落后。结果非常非常有用。
jgrahamc查看英文原文
Slightly off topic, but when I was writing The Geek Atlas one of the things I did was keep metrics about my writing so that I knew where I was, and then I used those metrics to predict the book's delivery date to O'Reilly, and measure how I was doing against the required delivery date. This was all done in a spreadsheet and it enabled me to see whether I was ahead or behind on my writing. Turned out to be very, very useful.
同类还值得读
写书：值得吗？
Writing a book: is it worth it?
Martin Kleppmann（《设计数据密集型应用》作者）复盘写书到底值不值：金钱回报低、耗时极长，但带来了可见度、信任和职业机会。
出版与作家生涯2020 · 642 分 · 287 评4 条写作要点编辑精选写书并不是个好主意
Writing books is not really a good idea
这篇文章讨论靠写书谋生的现实难度，评论区从版税数学、收入预期和写作目标三个角度补充了作者视角。
出版与作家生涯2021 · 577 分 · 609 评4 条写作要点编辑精选如何自助出版一本书：实用资源清单
How to self-publish a book: A handy list of resources
这是一份关于「如何自助出版一本书」的资源清单帖，涵盖从编辑、排版到 ISBN、销售渠道的实操工具与花费。
出版与作家生涯2018 · 558 分 · 88 评4 条写作要点编辑精选我自助出版一本Ruby on Rails书，赚了7万美元
How I Made $70k Self-Publishing a Book about Ruby on Rails
这是一篇 Ruby 开发者自述：他把一套 Rails 性能课程（书＋视频＋答疑）自出版，一年毛收入约 7 万美元，并总结了选题、定价、营销等经验。
出版与作家生涯2017 · 539 分 · 175 评4 条写作要点编辑精选石黑一雄警告年轻作家因“恐惧”而自我审查
Sir Kazuo Ishiguro warns of young authors self-censoring out of 'fear'
石黑一雄提醒年轻作者因“恐惧”而自我审查，即这条帖子讨论的核心。
出版与作家生涯2021 · 396 分 · 641 评5 条写作要点编辑精选