Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Part II:

Okay, maybe that monitoring work would not be very important.

But, maybe the company is facing some challenges in routing, scheduling, planning that, as is commonly the case, boils down to linear integer optimization, right, known to be in NP-complete. So, an impressive example might be a 0-1 integer linear program with, say, 40,000 constraints and 600,000 variables. Then, asking for a solution that is optimal, down to the last tiny fraction, might be a bit much, but coming within 1% of optimality might be nice. So, suppose Joe came withing 0.025% of optimality? Might want to chat with Joe about any problems in looking for good combinations.

So, Joe also put that work on the 1000 resume copies he sent.

Ah, usually the real data is arriving in ways that are not fully predictable. So, there is a stochastic process and want to know some of what the results will be. Once Joe was in touch with some people in touch with the US Navy, and they wanted an evaluation of how long the US fleet of SSBN missile firing submarines would be under as special scenario of global nuclear war but limited to sea. From an old B. Koopman report and some common sense assumptions, Joe saw a continuous time, discrete state space Markov process subordinated to a Poisson process, wrote and ran Monte Carlo software to do the evaluations, and had the work please the US Navy and be sold to a leading US intelligence agency (could say which one but then would have to ...).

Joe put that on his resume, too.

Then Joe got a call back, and they wanted to know what he knew about C, Python, and R.

Joe said that he'd one some C programming, but the language was so primitive that it was usually next to useless for his work in analysis. For R, that was just simple, standard statistics, and the work he'd done is statistics was beyond what R offered and he wrote his own code. For Python, that was similar -- instead he'd written his own code.

The interview died.

Okay, there's a fundamental lesson here:

Fundamentally, in US business, the attitude remains much as back in a Henry Ford plant 100 years ago: The company and the company managers know more than any candidate employee, about the work, how to do the analysis, how to write the software, and an employee is there just to add routine labor to what the company already knows. So, a resume that mentions work beyond what the company knows conflicts with this fundamental attitude and results in rejection.

In US mainline business (there are exceptions elsewhere), they are just very strongly against hiring as an employee someone able to do something the company can't already do.

So, dumb down the resume.

Then inside the company, fit in, don't expose real potential.

Then look for a good, new problem, say, needing analysis such as in the OP.

Then, largely independently, after hours, make some progress through prototype code and some impressive real output from real data.

Then develop some high level foils, go to the relevant manager, and give a short talk.

Then expect 'incoming' attacks, vicious, whisper campaigns, sabotage, etc., from jealous people. Then discover if the company really wants good analysis or not. If so, likely will have to transfer to a much better position, say, on staff of the CEO. Else, will likely soon get FIRED. No joke.

If say anything about such analysis before actually have the results, then people will either laugh or attack. The situation is quite general, say,

"First they ignore you, then they laugh at you, then they fight you, then you win."

Mahatma Gandhi

So, what the heck to do?

Sure, face the market directly. Do a startup. Pick your own problem, do you own analysis, write your own code.

Back to it. Work for today: Use a .NET collection class for a fast, easy to program way to find duplicates. So, a Web server, a Web session state store (wrote my own with just TCP/IP sockets, class instance de/serialization, and two instances of a collection class instead of using, say, SQL Server or Redis), SQL Server, and two specialized back-end servers, 18,000 programming language statements in Visual Basic .NET in 80,000 lines of typing.

Going live is getting to be very visible!

Let's see: The work is intended to provide the world's first good, a must have, solution for a problem pressing for nearly everyone connected to the Internet. To start, if I can get on average one user a second, 24 x 7, then, ballpark, I should get monthly revenue of

2 * 8 * 5 * 3600 * 24 * 30 / 1000 = 207,360

dollars. Now, in the OP, what annual salary ranges where they talking about?

The core analysis work? Silicon Valley has more hen's teeth than entrepreneurs who would understand that work even if I explained it to them, and many more such entrepreneurs than such VCs!

If you have something and explain it, then, if it is really bad, no one will like it. But, curiously, if it is really good, the same, no one will like it. Remember what Gandhi said. Or, easier and much the same lesson, just go back to kindergarten and to Mother Goose and "The Little Red Hen" -- same stuff. The hen wouldn't get hired, either.



Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: