Zettelkasten Forum


What macbook pro

Hi Macbook users, can you tell me what spec you use for the work you do?

I'm in the market for a new device, and trying to work out what I need. I'm planning a refurbished second hand Macbook Pro, but I'm not sure what components are more important to spend on, and what I can get away with lesser specifications. For example, I need to decide between 1) i5 and i7 processors; 2) how much RAM; 3) how much SSD (or whatever the more recent chip thing rather than drive is called); 4) what year. Screen size is also a consideration but a lesser one, as I gather any machine with an i7 processor has a 15" screen anyway.

For example, my friend said that personally he would never get less than i7, but then again he edits a lot of video and sound, and I would not be doing that. The work I would be doing would be primarily text-based, with data analysis in R. I guess a classic cognitive psychology workflow. I want to get up and running with the Archive, with LaTex, git, and whatever else I need to get a robust knowledge system together and keep all my ideas and research efficient and dynamic... probably similar to what some people on here work as. So tell me, is the i5 enough for that? I'm guessing it will be fine, but please tell me anyone who finds differently.

Further, I'd be really interested in hearing from anyone of any programs you found not to work for you, or anything that does work well, also how long you have had your machine. How much memory you have and whether it has ever run out, whether you have ever overloaded it? If you were buying a second hand machine now, how old a year would you accept?

Any other advice?

Thankyou in advance!
Steph

Comments

  • edited August 26

    I switched to a MacBook Pro just last year after the specs got upgraded. I do lots of CPU-heavy work when I develop applications, though the CPU is taxed only for a short time. Most of the day, programming is like writing: you just manipulate text, and that's not a lot of work for a comnputer. The CPU idles most of the time :) I got a 13" MBP because I prefer this size and weight for carrying the device around. I have a monitor and keyboard attached at my desk anyway.

    This is not a device any writer needs.

    I previously worked for years on a Mac Mini from 2011 (!!) and it rolled just fine with 8GB RAM and an SSD.

    I think the recent MacBook Air upgrades are nice. The non-Pro MacBooks, which were thin and light, were removed from the lineup, so that might not be an option for you.

    Here's my personal advise: If you want to get a refurbished MacBook anyway, get one with the pre-2015 keyboards. Even last year's keyboards in the MBP did not fix the problem of repeated key presses. Even though I mostly use an external keyboard my device suffers from single hits registering as 2 presses for the Cmd key, for example. The best option, I think, is the MacBook Air with the metallic screen bezel. I think the modern ones with black bezels all have the new butterfly keys that cause trouble sooner or later.

    • USB-C/Thunderbolt is very fast. I can boot Windows in a virtual machine from an external SSD via USB-C and it feels like it's native. That's super useful and I virtualize old macOS versions a lot. But if you don't rely on external disk speeds, USB-C might not be a selling point.
    • My MBP handles video encoding and running multiple CPU-taxing processes very nicely. It also gets very warm. If you write and do office work and light image processing (e.g. sub-100MB-Photoshop files, to pull a random number out of my hat) this is overkill. The Air suffices, and is lighter.
    • I don't really like the hard edge of the MBP, but with 13", I can rest my palms comfortably. The Air has a nicer angle.

    Author at Zettelkasten.de • http://christiantietze.de/

  • @ctietze Thankyou for your input! That is interesting, I had not considered a Macbook Air at all, because I had gathered from other students around the labs at uni that they were frustrated with it not managing to handle multivariate models in R and not producing the charts they needed... but I never actually spoke to anyone specifically, it was just a general feeling I got and I could have exaggerated or made it up

  • edited August 26

    @Steph said:
    @ctietze Thankyou for your input! That is interesting, I had not considered a Macbook Air at all, because I had gathered from other students around the labs at uni that they were frustrated with it not managing to handle multivariate models in R and not producing the charts they needed... but I never actually spoke to anyone specifically, it was just a general feeling I got and I could have exaggerated or made it up

    I'd say that anything on the market will handle what you've mentioned: manipulating text and connecting to the internet.

    I've been using a 2011 MacbookAir. As an English teacher, the most processing intensive thing I do is make screencasts to use as tutorials. So basic video and audio editing works just fine.

    I haven't come across anything that it can't do. I use Tinderbox, Devonthink, Scrivener, Mellel, Bookends. They show no sign of wanting an upgrade.

    That said, newer equipment will have cooler stuff like touch ID's touch bars and whatever else. They look nicer too. Then again, newer Airs don't have USB or 14 inch screens. I think asthetics and extras are all you really need to weigh in on....unless you're doing the computations you mentioned...so then maybe do go and speak to those people :)

    Good luck!

  • @Bart thanks for your input too! :) yeah yeah I absolutely definitely will be doing data analysis in R, (not very complicated) multivariate statistics, just running models and generating charts, it is not that complicated but I think what takes the processing power is running and re-running chunks of code many times over with minor changes

  • As for going back and asking coursemates, yeah it is weird, but one girl who I definitely heard ranting about "don't get a mac, it doesn't work, can't do it" etc - she is not answering yet.

    One guy says he has an Air and it was fine for everything yeah... then he added at the end that he still did his data analysis on the stats lab computers cos they were better (!) So that was a kind of non-answer.

    It is great having the lab computers, and basically I have already done 2 years of postgrad psych study with them as my only computers, just going into the department whenever I wanted to use them. But I'd really like to not be dependent on them and to be more mobile than that.

  • @Steph said:
    So that was a kind of non-answer.

    Could you ask a professor? Or maybe you could rent out a mac and try it? There's AV companies that rent them out for presentation... Otherwise, get a Pro. They're expensive but you'll keep it as long or longer than I've had my Air.

  • Can you give details on the computation you perform? How large the datasets are, etc.? I bet someone knows someone if you get specific, and I can ask around on Twitter and other chats.

    Author at Zettelkasten.de • http://christiantietze.de/

  • @ctietze awesome! Thankyou! I will tell you about the analysis from one course I had last semester. Its methods seemed very useful and attractive for making sense of all kinds of relationships in data. It will come in useful for further research at phd stage (and I hope I understand it a bit better by then - I passed it but with a very average mark). It was pretty typical for the kinds of thing I want to do.

    Linear mixed-effects modelling:
    It is a single framework for modelling longitudinal, cross-sectional experimental data with a variety of dependent variable types. It models variation at multiple levels of nested and crossed structure, and includes predictors to explain that variance.
    In other words, in contrast to simple linear regression which tries to explain a relationship with one straight line with a slope and an intersect, the story of our data might best be explained by many regression lines. Each of these will have a slope and an intersect, some of which will be fixed and some random depending on how they vary within the groups.

    For example, if you are exploring how "years of service" affects "salary", then you will have observations from many individuals. You will definitely need to know how long they have been working at their current job, and you'll probably want to know how old they are. But there are loads of other things that affect the story: company, department, occupation, industry, years of education, gender? And do you treat department as the same across companies: is THAT company's finance department comparable to THAT company's? Or shall we treat them separately? Also, for most variables there is a huge amount of individual variation, just by coincidence, and even the same person can be different from time to time. So we need to find a way to get that out the way so we can study the main effects.

    Then once you have built your model for all the sources of variation in your data, you run many different iterations to try and pin down the parameters. The number of parameters rises steeply as you bring in more variables, so you work with matrices of partial derivatives for all parameters. Then you test different but similar models, to see which has the better parameters and indicators of best fit. I guess this is what takes computing power.

    Uhh sorry, if you have read this far then you will be asleep by now... as for size of dataset not massive,I think a typical project could have eg. 500-1000 observations of eg 8 relevant variables, so up to 8000 long observations, sometimes fewer.

    Is that the sort of thing that helps you know what sort of computations they are? More detail, less? Anyway, thankyou so much for taking time to engage with my questions :D

    I guess the keywords are: linear mixed models, matrices for partial derivatives, lmer function in R

    @ctietze said:
    Can you give details on the computation you perform? How large the datasets are, etc.? I bet someone knows someone if you get specific, and I can ask around on Twitter and other chats.

Sign In or Register to comment.