Short-blogs

A bunch of short blogs and thoughts, in no particular order. Insults are mostly self directed.

80-20 rule in small vs big companies

In a small company, 80% of what you do doesn’t really matter. In a large company, there is an 80% chance that nothing you do matters.

PS: In a small company, there is also a 95% chance that the company will eventually fail, making all your contributions useless. Unless you learned from it.

Discomfort

Heuristic for deciding between being an employee vs a founder: 

• discomfort being an employee » discomfort being a founder. 

If not, or if you don’t feel any discomfort as an employee, don’t become a founder!

Bell-curve pay

It doesn’t matter how good you are, if the mean (or median) pay is low, you will never get a high wage. The right tail (really good people) has to compensate for the left tail (not-so-good people that still get paid).

Hiring and false positives

Hiring is about avoiding false positives. They are extremely costly: 

• difficulty in firing people, 
• lost cost and time, and 
• now you have to recruit again. 

The benefit of an amazing candidate over an adequate one is not as much as the cost of a false positive. Irrational recruiting practices, with excessive barriers to entry, are actually quite rational. Recruiting is false positive elimination.

Indecision Ponzi scheme

Organisations get filled with people who don’t make any significant decisions. These people then rely on each other to realise short-term improvements, with long-term risks. Just worry about quarterly targets. You won’t get blamed if the company eventually fails.

Those that make radical changes get removed when something goes wrong.

80% of AI (and data science) projects fail

So best practices won’t matter 80% of the time, as the project will fail in any case. Hence the lack of best practices with the projects.

Money = information

Money as a metric compress a lot of info, which is good for forcing decision making. Money is a competitive and distributional resource. Money is what we use to determine the distribution and allocation of real resources. The only time when I know for certain what people want is when I give them money or when they give me money.

More info is bad for decision-making

The more info, the less likely a decision will be made unless there is a time limit.

Team players

HR really wants a team player. Not a changer, they are too risky.

Surviving by hiding

People survive in large organisations by minimising risk and their exposure to it. It’s risk-free to step into a vacant role. It’s much more risky to step into a new one.

Fire there not here

Promotions are directly related to the number of subordinates. More employees are always the end-state, resulting in fire there not here games. A case in point is AI automation. “Give me more money so that I can hire more people to build AI so that you can fire elsewhere…”

Calculating Return-On-Investments

If you don’t calculate ROI, someone else will for you. Or more likely, they will just assume it’s zero.

Distributions vs incentives 

Probability distributions and expected values reveal a lot about how organisations should behave. Incentives reveal why they don’t.

IT is essential, nothing more nothing less

Treat and see traditional IT as essential operations, nothing more and nothing less. They are there to keep things running smoothly and respond to incidents.

“You have to spend 10-15years on this…”

The typical general advice when starting a new business, but like all of this type of slob, it lacks details. So you should spend 10-15y on this, but:

• With what outcome in mind?
• Most don’t make it past 1y.
• A lot that do, shouldn’t.

Science in POCs

Science in POCs is like salt in a meal. Without it, a meal can taste a bit bland and underwhelming. Too much can completely ruin it.

Some more detail with systems thinking:

• box things and create boundaries;
• decouple as much as possible;
• isolate the most risky parts;
• investigate and experiment with the risky parts;
• thoroughly document and learn.

Successful POCs

You learned! You know the next steps, business outcomes and failures. You have a better idea as to whether to proceed or not, and if you do, you know what things need to change for it to work.

Unsuccessful POCs

You learned nothing. You treated it as a binary go/no-go project.

The dangers of AI readiness consulting

It’s a Trojan horse. When POCs fail, blame the data, and sell a new data strategy, in replacement of your previous big data, data science and machine learning data strategies. You don’t need a new strategy, you need to fix your data. Why is it bad? How good does it need to be? How do we fix it?

Don’t fuck with the coffee machine

It’s the quickest way to get white-collar workers to riot.

Don’t be the best, be the favourite

How to sell SaaS.

Consulting is anti-fragile

When there is variance and scope creep, you sell more hours, thereby benefiting from randomness.

Target easy markets

Cause it’s difficult enough as-is.

It’s never easy

If you think something is “easy” but you are critical of it, that it never works, chances are you are missing the trade-off curve. Easy means low investment and bad results. Difficult means high investment and good results.

Corporate mindset

Waiting for permission before doing anything. And when you have your own company, fabricating permission lines (external consultants, lawyers, etc.)

Problem-solving vs selling

It’s not just about finding existing problems to solve, it’s about finding existing solutions to sell. The selling part should excite you. Not just the solving part.

Automating vs replacing

Automating manual processes and replacing manual processes with automated ones is not the same thing. There are many more degrees of freedom and resistance to change to the latter.

Big vs small

Be patient when discussing big ideas with people who tend to think small.

Shitty competitive products

When analysing shitty competition and products, the real question is not “how do we make ours better, or why is ours better?” The real question is, why is the shitty product still successful? If it’s because of something else, then what’s the point of having a better product?

Playing startup vs building companies

One of them allows for more time to post stuff on social media.

Results may vary

The worst disclaimer in industrial AI systems.

Common mistakes

If common mistakes were easy to avoid, they wouldn’t be so common.

5min vs 20min presentations

Do you need to present for 5min or 20min? Why are you talking? To show people how knowledgeable you are? How wrong they are? That you know something they don’t? Or, to get them to act? To change direction? Adapt something they need? Do you really need to present for 20min?

Inertia

In corporate, inertia is the enemy. Getting people to do something different is already an achievement.

Communication in corporate

It stretches beyond written content. It has its own cultural identity. Is it via meetings? Short emails? Presentation? Demos? Memos? You have to fit in and work with it, not against it.

“The last career guide you will ever need”

It’s pretty bad actually. It should be titled “How to become a middle manager”.

US AI risk

The big risk for US AI tech companies is being sued for the actions of their AI, basically suing god.

Apps

Apps don’t compete on value add, they compete for attention.

Optimisation

It’s just an upfront investment for a later payoff, identical to automation. So how much is that investment? How much is the payoff? What is the maintenance like?

Knowing the answer vs having lived the answers

Outliers

There is a vast difference between being an outlier in a business vs being an outlier with a business.

“What would you do if you didn’t have to worry about money?”

A very classest job question. Up there with “money isn’t everything”, as said the rich person. It’s like putting food in front of a hungry person, and asking them: “If you were never hungry, what type of restaurant would you still visit?” For the hungry, there is no reference point and no answer.

All rumours are false until officially denied

Here be dragons

A big risk as a first-time founder is preparing for imaginary fairy-tail battles. It’s like worrying about how to celebrate scoring a goal, while you are still learning how to play.

Services vs products

In services, the customer is always right. In services, the customer is always wrong.