In ten words: the inability to set up correct experiments and an extremely hypertrophied sense of one’s own importance. I will not describe a specific case that I encountered, but I will describe a fictional case with the same features.

Let’s say you were approached by a friend who wants you to give him the contacts of managers and top decision-making algorithmists in the conventional Microsoft. Why? After an exchange of lines, where he initially balks, it is revealed that he has invented a new sorting algorithm. As proof, he shows a Python program that he says sorts 10 numbers faster than Python’s default sort.

I tell him that it might be an artifact of the python interpreter and the particular set of numbers, which he agrees, writes it in C, and says that his sort is 15% faster than qsort despite being based on bubble sort. I said to him: “Ah, so your sorting is also based on bubble sort? Yes, bubble has complexity O(n**2), and qsort – O(n*log n), what are we even talking about?”

At this the inventor begins to accuse me of being associated with the Microsoft compiler group, which does not want to lose the monopoly on qsort in their library, and that I do not understand anything about sorts. And demands that I give him the email of someone who understands. If not in Microsoft, then in Google or Facebook.

I told him, “I don’t want my colleagues or people in other groups/companies to start complaining that I send them people who load them. They won’t dig into it, because:

1. They already use qsort, and it is sufficient for them. They deal with algorithms in more specialized things that qsort has been known for 50 years and the holerite punched card sorting problem has been known for 150 years.

2. You have no articles or other achievements on the topic.

3. Figuring out where you went wrong there is work. The error may be in incorrect accounting of the program start time, cache effects, data size, amount of data, the array may already be partially sorted, etc.

That is, you want top experts to deal with you and then teach you to program for free.”

Then I suggest that he write an article on Habré, so that it will be criticized there by those who are not lazy. He resolutely refuses “why do you offer me public disclosure again and again?” That is, he wants to agree with the big boss that he will be paid royalties from billions of copies of some software based on his idea, which top experts will develop into a product.

I then have him email the author of qsort in the GCC library. He refuses because the open-source community is boring for him, he wants to work with serious large commercial companies. And he says “well, yes, I will contact venture funds!”

What do I do to him:

“Always write to venture capital funds! They get ten emails a day with fifty-page business plans written by marketing consultants who charge a hundred thousand dollars for each business plan (I’m not talking abstractly, I’m I’ve been through and gotten investments in my time, so I know the inside scoop.) I mean, VC partners usually hate reading these business plans explaining TAM (total available market), TSM (total serviceable market) and SUCA (Sustainable Unfair). Competitive Advantage).

What to do to become a successful inventor, and not an inventor of a perpetual motion machine?

For the past 30 years, the digital logic of microprocessors, GPUs, network chips is not drawn with a mouse on the screen, but synthesized from code in the hardware description languages ​​Verilog and VHDL. Here’s a quick cheat sheet on how it differs from programming:

You will learn about all this at the School of Synthesis. If you decide to register, welcome!