3 ms·
It is an interesting question which, in my opinion, hinges entirely on a fairly puritanical notion of 'work.' One of my 'problems' is that I can't "not work" i
by ChuckMcM 10y ago
It is an interesting question which, in my opinion, hinges entirely on a fairly puritanical notion of 'work.'
One of my 'problems' is that I can't "not work" in the sense of a laborer who is no longer building widgets. As a person who is asked to solve complex problems with a high degree of dimensionality inside of an arbitrarily constrained solution space, much of the 'work' I do consists of turning the problem over and over in my head while I explore the solution space.
A good example of this was an early review I got at Intel by my manager (a really solid EE type guy). He added some criticism of my time management (considered a 'ding' in the vernacular) for a embedded compiler/assembler/driver thing I wrote as part of the evaluation of a graphics processor. He said I had 'sand bagged' the time estimate.
When I asked him to describe that a bit more he explained that I had told him it would take 6 months to do, and it was 2 weeks late, and I had spent 5 months "goofing off and not working" and then about 6 weeks doing the work. So my estimate should really have been '8 weeks' and if I had started on time it would have been done two weeks earlier than that.
I thought about that for a long time. And explained to him the for five months I had no idea what the best way to write the software was, and in that five months I had learned about 8 different technologies that all came together into the final solution. I had to learn how to write device drivers in Xenix, how to map I/O space memory into the kernel, design a language which was human understandable and could be compiled into the odd little RISC instruction set of the Graphics chip. And until I had figured out all of that precursor information, I didn't have a clue how it would be written, but then after figuring out that information actually writing the code was fairly mechanical.
In this one case the problem was that hardware has so many great milestones you can call out, parts captured, schematics done, netlists verified, layout started, design rule verification, first films, films checked, first boards, boards checked, first assembly. Bringup in stages 1, 2, and 3 etc. All along the way there are pleasant milestones to say "this is done" now on to the next thing.
But software is never like that for building something that nobody has ever built before. And it is even rarely like that when you have the same software but you are building it on a different system. The linkages, the entanglement between the system and the software (and now the network and the services) makes each new implementation its own special snowflake, with its own kinds of problems.
Have you ever woken up and "knew" the solution to a tricky software issue? Or had an idea for a change to an existing system that might make it better? That happens to me all the time when I'm designing stuff. And a case could be made that I'm working even when I'm asleep! Not because I'm sitting there typing in lines of code but because I'm going through the solution space, somewhere in my subconscious, looking for clues to places that hold better answers than the answer that is currently checked into the git repository.
As a result I tend to measure my own productivity by 'solutions over unit time' versus 'hours typing into employer owned equipment'. It still bites me from time to time when a supervisor needs a constant stream of 'still flying' type status messages to feel comfortable.
- tutufan 10y agoInteresting. So despite the apparent success, your manager put on paper that you were "goofing off and not working"? Did he mention that verbally one or two times beforehand? How did you react? Once things hit paper, I generally start interviewing. Not sure whether that was right. But looking back over my career, I usually should have left earlier rather than later in those cases.
- ChuckMcM 10y agoIt was verbal vs on paper. What was in the review was that I needed to improve my time management. And oddly enough I left for Sun Microsystems not to long after that, but not strictly because of the review, rather it was that I preferred a work culture that understood what programmers did in addition to understanding what EEs did. And for what its worth, I took it to mean that I wasn't communicating how I was spending my time well enough and suggested ways that would work for both of us. I expect that had I stayed at Intel it would have been fine.
- gtirloni 10y agoI've a hard time communicating what I'm doing in that "research" phase. During standups, most people are reporting concrete tasks that got done and I feel a bit embarrassed to say I spent the whole day trying something out that's remotely related to the final product but was interesting for the ideas. It's stressful to have to report those things, I feel like I have to justify them, what's my train of thought, etc, and a 1-minute update has to turn into an essay. Not really sure how to improve on that. During your time at Intel did you have to report on your work quite frequently?
- ChuckMcM 10y agoIntel was my first exposure to "management by objective" and "key results". And yes, there were weekly reviews of your progress. Every quarter, a set of objectives and under each objective key results, and then weekly you reported on how you were doing with respect to those. I think the key to your stress is either that you aren't clear in your head what you're trying to figure out, or your trying to figure out things that aren't actually relevant. In my life what I have done is to do get very very disciplined about what I'm to understand and how that will relate to the final goal. Not surprisingly that is different as a manager than it is as an individual contributor. If I were your manager, and you came to me with this (and I would hope your relationship with your manager is good enough that you could talk to them about it) I would start with three questions; 1) What parts of your assignment are you completely confident you can build/write? 2) What parts are you unsure of how to build? 3) What things that you are unsure about are between you and building the things you know you can build? Perhaps you can imagine how the conversation goes after you answer those questions. But lets say for this hypothetical that you were unsure if you could build a database fast enough to respond in time to meet the response requirements of the product. At that point we'd talk about what steps you could take to understand what sort of performance to expect out of a database, what variables had the biggest impact on that performance, and which databases were designed to be fast. And you and I would agree that you would spend this sprint perhaps developing your understanding of database performance. So at the stand up I'd expect you summarize and article you read, a set of benchmarks you set up, a set of test tables you created in the existing system, or maybe the top 5 blogs/books/videos you've found on analyzing database performance. At the retro I'd want to know what sources gave you the most information for the time invested, what were the time wasters, if you were more or less confident about the database choice and its performance and maybe how you had, or would, quantify your understanding with something objective. From your perspective you would probably have spent the time 'surfing the web' to find out various sources of information or perhaps prototyping some things on AWS or on a local server. If instead you came back and said, "I really don't know anything more about databases yet but I learned Rust, and got stuff running in the new Angular release and updated my server to the latest Ubuntu and read a book on containers. Then we're going to have a different talk :-)