Elon Mush’s ‘the algorithm’

Elon Musk calls it “the algorithm,” a distillation of lessons learned while relentlessly increasing production capacity at Tesla’s Nevada and Fremont factories.

According to Walter Isaacson in his new book Elon Musk, there is a “nontrivial chance” that Musk will trot out the algorithm during any given production meeting.

“I became a broken record on the algorithm,” Musk says. “But I think it’s helpful to say it to an annoying degree.”

The next time you’re trying to be more efficient and effective, whether professionally or personally, give Musk’s algorithm a try. Just make sure you complete each step in order. (Sections below in italics are from Isaacson’s book, quoting Musk.)

1. Question every requirement

Each should come with the name of the person who made it. You should never accept that a requirement came from a department, such as the “legal department” or the “safety department.” You need to know the name of the real person who made that requirement. Then you should question it, no matter how smart that person is. Requirements from smart people are the most dangerous, because people are less likely to question them. Always do so, even if the requirement came from me.

Then make the requirements less dumb.

When I took over manufacturing at a new plant, supervisors had to sign off on quality before a production line could start up. The crews often waited five or ten minutes for a supervisor to be found. (Which was another problem that needed to be solved; leaders should be on the floor, not in offices.)

Why? The CEO of the company had established the rule after one expensive mistake. But if operators couldn’t be trusted to know whether their work met quality standards, they shouldn’t be operators.

Many blanket requirements are based on a one-off event that didn’t require a process, guideline, or rule in response. Instead, just deal with the specific situation.

Learn from it – but don’t respond by creating a box everyone must forever fit inside.

2. Delete any part or process you can

You may have to add them back later. In fact, if you do not end up adding back at least 10 percent of them, then you didn’t delete enough.

When I first became a supervisor, one of my jobs was to prepare, print, and deliver a daily report to 20 or so people. The whole process took over an hour. One day I wondered whether anyone actually read the report, so I created it but didn’t print or deliver.

No one noticed.

So I stopped delivering a few other reports. Created them, but didn’t deliver them. No one noticed.

Often we do things simply because we’ve always done them. Or because we think we need to. Or because it’s our job, and therefore it must be important. (Everything about our jobs is important, right?)

3. Simplify and optimize

This should come after step two. A common mistake is to simplify and optimize a part or process that should not exist.

A couple weeks after I’d stopped delivering those reports, I asked a few people if they needed me to start delivering them again. Nope. Then I asked if we still needed to collect the data involved. In most cases we didn’t, because it was already being collected elsewhere. (My department had been doing double work because we didn’t think we could trust other departments to get it right.)

In a few cases, we did need occasionally need certain data, so I found a way to automate the collection process. And I found a way for production crews to not be involved in the collection process, which meant they could spend more time producing and less time serving as data entry clerks.

As you’ll see in a moment, make sure you don’t automate or optimize a process that doesn’t need to exist in the first place. Sure, you can make percentage gains by making something better, but why not save 100 percent of the time, effort, and cost involved in an unneccessary process by eliminating it altogether?

4. Accelerate cycle time

Every process can be speeded up. But only do this after you have followed the first three steps. In the Tesla factory, I mistakenly spent a lot of time accelerating processes that I later realized should have been deleted.

At my old job, we constantly worked to wring time out of job changeovers; the faster we could switch from job to job, the more units we produced per day. In simple terms, the two major ways to increase productivity are to speed up the rate of production (think increasing the miles per hour) and to speed up the time it takes to switch from producing one widget to another.

We spent a ton of time trying to make a set of conveyor guides easier to adjust. A few seconds here, a few seconds there, until one day a junior operator said, “I don’t see why we need to adjust them at all. If we change the shape a little, they’ll work for any size product we run.”
Turns out he was right; we were trying to accelerate a step that should have been removed altogether.

5. Automate

That comes last. The big mistake in Nevada and at Fremont was that I began by trying to automate every step. We should have waited until all the requirements had been questioned, parts and processses deleted, and the bugs shaken out.

After you’ve completed the first four steps, what’s left – the things you really need to do, that are truly important, that genuinely add value – can then be optimized and automated. Get rid of all the fluff, and then make what’s left as effective and efficient as it can be.
Although my former co-workers and I didn’t realize it, we were loosely following Musk’s algorithm at that job. But in our case, as in Musk’s, we often made our lives more difficult by inadvertently skipping a step or two and having to go backwards. (Even so, over the course of ten months, we cut job changeover times in half.)

There are also a few corollaries to the algorithm. Here are a few of my favorites:

All technical managers must have hands-on experience. For example, managers of software teams must spend at least 20 percent of their time coding. Solar roof managers must spend time on the roofs, doing instalations.

Leaders with practical experience tend to make better leaders. (One study found that if your boss can do your job, you’re more likely to be happy at work.)

It’s OK to be wrong. Just don’t be confident and wrong.

Whenever there are problems to solve, don’t go directly to your managers. Do a skip level, and meet with the level right below your managers.

Often, the best decision you can make is deciding who should make certain decisions, and then empowering those people to make those decisions. Almost always, those people are at least one level further down the hierarchy than you think.

The only rules are the ones dictated by the laws of physics.
Everything else is a recommendation.

Granted, you may not want to take that approach to extremes, but if you want to do something better, or faster, or cheaper – or if you simply want to make your business or life better – then you have to do things differently.

Because if you do what everyone else does, you can only accomplish what everyone else accomplishes.

馬斯克的演算法

不論是在特斯拉或是SpaceX，只要召開生產會議，馬斯克通常都會提到他的「演算法」(the algorithm)，就像是念口號一樣。這些主要是他在內華達和佛利蒙工廠進入「爆能」生產地獄期間學到的教訓。

他的高階主管有時候會張嘴默念這些演算法，就好像跟著牧師吟誦禮拜一樣。「我會像故障的唱片一樣不斷重複這些演算法，」馬斯克說，「但我認為反覆講到讓人覺得很煩的程度，確實很有用。」

馬斯克的演算法包含五大戒律:

一、質疑每一項要求。

每一項要求都應該附上提出這項要求的人名。任何來自部門單位的要求，例如「法務部門」或是 「安全部門」，你們都應該拒絕。你們必須知道提出這項要求的人是誰。不論這個人有多聰明，你們都要提出質疑。聰明的人提出的要求通常是最危險的，因為其他人多半不會去質疑他們。
一定要提出質疑，即使是我提出的要求也一樣。然後想辦法把這些要求變聰明。

二、如果可以，刪除任何零組件或流程。

之後你可能需要重新把某些刪掉的東西加回去。事實上，如果你後來沒有重新 加回去至少10%，就代表你一開始刪得不夠多。

三、簡化與優化。

完成第二步之後，才能開始這一步驟。常見的錯誤是去簡化或優化本來就不應該存在的零組件或流程。

四、縮短週期。

所有流程都應該要縮短時間，但前提是必須完成前三個步驟。

在特斯拉工廠，我的錯誤是花太多時間加 快流程，但後來發現許多流程根本應該要被刪除。

五、自動化。這一點必須放在最後。

我在內華達與佛利蒙工廠最大的錯誤，就是從一開始就想盡辦法要把所有步驟自動化。

我們應該要等到所有要求都被質疑過、不必要的零件與流程全都被刪除、所有漏洞都被解決之後，再開始自動化。

「演算法」有時還會伴隨著一些補充，

例如: 所有技術主管都必須具備實務經驗。

例如:軟體團隊的主管至少要花20%的時間自己寫程式。太陽能屋頂安裝團隊的主管必須花時間親自在屋頂上安裝太陽能板。否則他們會變成不會騎馬的騎兵團長，或是不會用劍的將軍。

同事情誼其實很危險。這會讓人很難去挑戰彼此的工作，大家都會覺得不要害到同事。必須避免這種情況發生。

犯錯沒有關係，但是不要犯錯還過度自信。

絕對不要要求部屬去做你自己不願意做的事情。

如果有問題需要解決，不要只找主管討論。要跳過這些主管，直接和他們底下的員工討論。

找人時，一定要尋找具備正確態度的人。工作技能可以之後培養，但是要一個人改變態度，只有換腦手術做得到。

一定要有瘋狂的急迫感，這是我們的營運原則。

受限於物理定律的規則才是真正需要遵守的規則。其他都只是建議。