The 100-Hour Rule

A recently popularized meme is the 10,000-Hour Rule, which describes the amount of time required to achieve mastery of a field. This rule has several implications:

  1. Since the required time commitment is so high -- three hours per day for a decade! -- a person can only master a few things in their lifetime.
  2. Because time moves at the same speed for everyone, you can‘t accelerate the acquisition of mastery. If you‘ve mastered something (e.g. sales or programming or product management) and your competition has not, that‘s a huge competitive advantage.
  3. The commitment for mastery is so high that a lot of people become intimidated and give up early -- or never even begin. For every violin virtuoso, there are many more people that quit after a few lessons, or who never picked up a violin in the first place.

Mastery of various disciplines is very important for startups. A typical startup has many areas it has to handle well, including software engineering, user experience design, product strategy, sales and marketing, and recruiting. Failing in any one of these areas can be enough to cause the whole company to fail. For example, if founders can‘t recruit a good team efficiently, then they won‘t have the resources to implement their plans, regardless of how brilliant those plans may be. Or, if the founders can build a useful product but it‘s not user friendly or aesthetically appealing, then the product usually struggles to get traction. Of course there are exceptions (e.g. Craigslist and its old school UI), but they are... well, exceptional.

This poses a dilemma: few founding teams have mastered each of the five or ten areas critical to their success, but it takes a long time to achieve that mastery.

To address this dilemma, I‘d like to propose the 100-Hour Rule:

For most disciplines, it only takes one hundred hours of active learning to become much more competent than an absolute beginner.

For example:

  • Cooking: it takes years to become a master chef, but one hundred hours of cooking lessons and classes and deliberate practice will make you a better cook than most of your friends.
  • Coding: it takes years of study and practice to become a strong software engineer, but going through a couple of Codecademy or Udacity courses will make you a good enough programmer for a lot of basic applications.
  • Sales: it takes years of experience to become great at sales, but reading a few key books and shadowing experienced sales people for several hours will help you learn enough to avoid the most common/dangerous sales mistakes.

The sales example is something I‘ve experienced personally. Before becoming a VC, I was a software engineer for ten years. I never interacted with the sales side of a business and didn‘t know the first thing about it. Once I moved to the investing side, I quickly learned that the bottlenecks for most companies were sales/marketing/customer acquisition, and not technology. As a result, I‘ve tried to educate myself on sales and related fields. I‘ve read books like Traction and gone to conferences like SalesConf (notes). I‘ve probably invested 50 or 100 hours into learning sales. What‘s the result? While I‘d still pale compared to any experienced salesperson, I‘ve picked up enough to know more than many non-salespeople. I now know that most software products should be priced based on value and not cost, that it‘s better to talk about benefits instead of features, and that the most important part of sales islistening to what customers want instead of telling them what you have. If an expert salesperson can turn 80% of prospects into sales while a newbie can only close 10%, I‘d probably be at 30% or 40% right now. Far from what the expert can do, but way more effective than the newbie. That‘s a great ROI for 1-2 weeks of full-time learning.

A few observations about the 100-Hour Rule:

  • While "100" is an easy-to-remember round number, it‘s just an approximation. Some fields might require ten or twenty hours to achieve a level of medium competence, while other fields might require several hundred hours. Either way, that‘s much less than the 10,000 hours required for full mastery.
  • The 10,000-Hour Rule is based on absolute knowledge -- it takes about that long to learn a field inside and out. In contrast, the 100-Hour Rule is based on relative knowledge. Since 95+% of people don‘t know anything about most fields, it takes very little time to jump from "the naive 95%" to the 96th percentile. The long part of the journey is moving from the 96th percentile to the 99.9th.
  • As with the 10,000 Hour Rule, learning has to be active and deliberate. That is, you‘re not just skimming a book or mindlessly practicing the motions of some technique, you‘re reading and practicing with the deliberate goal of learning and improvement.

To tie all of these observations back to building companies, make a list of the areas where your company will have to excel in order to succeed (sales, engineering, UI design, knowledge of a specific domain, and so on). If your team lacks experience in some of those areas, don‘t just wing it and hope for the best. Invest a little bit of time into learning and basic competence so that you don‘t hamstring yourself by making rookie, easily avoidable mistakes. In the long run, you‘ll want to hire experts; in the short run, invest enough time into learning so that you can adequately fill any critical gaps.

时间: 2024-11-15 21:03:55

The 100-Hour Rule的相关文章

ORACLE优化器RBO与CBO介绍总结

RBO和CBO的基本概念 Oracle数据库中的优化器又叫查询优化器(Query Optimizer).它是SQL分析和执行的优化工具,它负责生成.制定SQL的执行计划.Oracle的优化器有两种,基于规则的优化器(RBO)与基于代价的优化器(CBO) RBO: Rule-Based Optimization 基于规则的优化器 CBO: Cost-Based Optimization 基于代价的优化器 RBO自ORACLE 6以来被采用,一直沿用至ORACLE 9i. ORACLE 10g开始,

关于Netfilter NF_HOOK宏的outdev参数bug

1.首先指出,NF_HOOK系列宏的outdev参数的传递方式(直接传递一个net_device结构体指针)是不正确的 正确的方式要么是不传递,要么是传递指针的地址,即地址的地址. 2.接下来指出,仅仅传递一个地址为何不对 因为在该HOOK点可能存在多个HOOK函数,每一个函数都有可能改变skb的路由,即调用reroute,比如NAT,比如IP Mark等,这样后续的HOOK函数看到的依然是旧的outdev参数,而不是reroute之后的skb_dst(skb)->dev. 3.然后看一个实际出

Drools Expression 介绍

用好Drools 中的表达式是你的规则引擎能否强大的必要条件 http://docs.jboss.org/drools/release/6.1.0.Final/drools-docs/html_single/index.html 下面列出了几个个人认为比较重要的点: AND Drools 默认的并列表达式就是 前置的AND (and Cheese( cheeseType : type ) Person( favouriteCheese == cheeseType ) ) when Cheese(

Linux 基于策略的路由

Linux 基于策略的路由(Linux Policy Routing) Linux 有传统的基于数据包目的地址的路由算法,和新的基于策略的路由算法新算法优点:支持多个路由表,支持按数据报属性(源地址.目的地址.协议.端口.数据包大小.内容等)选择不同路由表 # 查看规则命令,后面可跟其它参数,默认为 show(list) 显示全部ip rule 系统默认有3条记录0: from all lookup local32766: from all lookup main32767: from all

sql语言复习2

一.查询 select 字段列表 from 表名列表 [where 条件表达式][group by 字段列表[having 条件表达式]] [order by 字段列表[asc|desc]] 含义:在from后面给出的表中,按照where给定的条件,select出规定的字段列表.如何有group by 字段列表,则结果集按照group by 后面的字段列表排序,having 后面的条件表达式是分组时结果输出条件.order by 表示结果集按照其后面的字段列表进行升序或者降序排列.各个字段列表与

CentOS学习笔记--防火墙iptables

Linux 的防火墙:iptables iptables是封包过滤软件,Linux内核2.6以上都是这款软件.本节节选自 鸟哥的 Linux 私房菜 -- 服务器架设篇  第九章.防火墙与 NAT 服务器 封包进入流程:规则顺序的重要性! 假设你的 Linux 主机提供了 WWW 的服务,那么自然就要针对 port 80 来启用通过的封包规则,但是你发现 IP 来源为 192.168.100.100 老是恶意的尝试入侵你的系统,所以你想要将该 IP 拒绝往来,最后,所有的非 WWW 的封包都给他

华为设备acl配置

拓扑图: 需求: 1.-vlan10内所有的主机,只能通过http访问vlan30-server的服务器;不能访问vlan40-server服务器2.-vlan20-pc1主机,可以访问vlan40-server服务器,不能访问vlan30-server服务器3.-vlan30-pc1主机,不能访问vlan20-server服务器,可以访问vlan40-server服务器4.-PublicServer服务器对vlan10和vlan20 仅仅提供ftp服务5.-PublicServer服务器对vl

16、基于状态的iptable+高级路由(重点)

-- 基于状态的iptables 如果按照tcp/ip来划分连接状态,有12种之多 但iptables里只有4种状态:ESTABLISHED.NEW.RELATED及INVALID 这两个分类是两个不相干的定义.例如在TCP/IP标准描述下UDP及ICMP数据包是没有连接状态的,但在state模块的描述下,任何数据包都有连接状态. new(第一次) |      |-------------------->|      | |      |   established(回来)  |      |

MyCat 学习笔记 第十一篇.数据分片 之 分片数据查询 ( select * from table_name limit 100000,100 )

1 环境说明 VM 模拟3台MYSQL 5.6 服务器 VM1 192.168.31.187:3307 VM2 192.168.31.212:3307 VM3 192.168.31.150:  3307 MYCAT 1.5 服务部署在宿主机上 MYCAT 192.168.31.207 :8806[SQL执行端口] / 9066[管理端口] 2 应用场景 2.0 MYCAT配置 schema.xml <schema name="TESTDB" checkSQLschema=&quo

基于RULE的优化器(学习笔记)

崔华<基于Oracle的sql优化学习笔记> 1.1 基于RULE的优化器 (1) CBO (2)RBO 和CBO相比,RBO是有其明显权限的.在使用RBO的情况下,执行计划一旦出了问题,很难对其做调整.另外,如果使用了RBO则目标SQL的写法,甚至是目标SQL中所涉及的各个对象在该SQL文本中出现的先后顺序都可能影响RBO执行计划的选择我,更糟糕的是,Oracle数据库中很好的特性.功能不能再RBO中使用因为他们不能被RBO锁支持. 只要出现如下情形之一,那么即便修改了优化器模式或者使用了R