Hacker News Reader

i cared about doing clean installs/configuring when i was a teenager and had nothing else to do. now i just want my computer stay out of my way and make me money. os x, time machine, etc all let me do that

You do know you're not the typical laptop buyer, right? The majority of people buy a laptop, take it home, and want it to just work.

Do you buy a new refrigerator expecting to have to clean out the compressor the minute you take it home? I certainly hope not.

If I could make a 100x profit on my time doing it, you bet.

To put more forcefully: if I needed to maintain one of those room sized refrigerators, and could save $10,000 for a couple hours of messing with a compressor the decision is a no brainer.

I agree that mortals can't/won't do this -- doesn't mean the HN crowd is being intelligent when they say what the guy I replied to was saying :p

EDIT: Downvoters, please explain -- For people with high computing requirements who insist on the x220 / Macbook air form factor, this is an accurate picture of the ROI.

Look at the performance differences between the macbook air ssd (http://www.anandtech.com/show/3991/apples-2010-macbook-air-1...) and the vertex 3 (http://www.anandtech.com/show/4316/ocz-vertex-3-240gb-review...). If your job commonly has you IO bound you will see at least 100 hours of your life saved over the 2,000 hours of a year of 40 hour work weeks (a 5% savings -- not all your time will be spent IO bound). And that's not including the difference between a 2.7 Ghz and a 1.7 Ghz processor...

Someone with high computing requirements shouldn't be using an ultra light notebook; get a desktop or get an amazon ec2 instance for a few pennies.

That's just it: with a vertex 3 and a maxed out CPU the x220 doesn't perform like a thin and light -- I'm getting the same IO performance as a vertex 3 in a desktop, and although I could get a CPU boost with a desktop processor I'd have to give up on having all my work with me at all times / at conferences.

The EC2 option has always interested me, but short of EBS RAID its IO actually isn't better than a vertex 3 -- they're crazy fast; if you haven't tried them I highly recommend you do :)

Because you are missing the point. The discussion is on basic marketing which implies to the general populace, not HNers who are comfortable re-installing their OS. And you preface with "How odd..." either unknowingly or just plain condescendingly describing 95% of the computer buying public. Then you fib that Macs are not ready for use out-of-the-box. Then you put forth the classic defense of exaggerating the value of your work.

ps I did not downvote.

Thanks for the reply. It cleared up my misunderstanding -- I think HN readers don't do a good job thinking when it comes to Apple stories.

(for those curious here is the breakdown of why pbreit is wrong:

1. The guy I was replying to is by definition an HNer, and was talking about buying an air over a sony SB due to reinstall concerns. My "how odd" was in reply / directly to him.

2. Macs aren't ready to use out of the box. No computer is. You need to, at the very least, install all the apps you use and copy over your data. This takes considerable time. Did it come with Chrome? Did it come with Office? Did it come with (your editor of choice)? Did it come with Little Snitch? 1password? The caps-lock key switched to a control key? Inconsolata installed? No. And that's not a complete list by a long shot. Adding an hour long reinstall to this list isn't a big deal.

3. The x220 and Macbook air do have the same form factor from a functional perspective. They both are quite easy to keep with you at all times, hence both inhabiting the thin and light category used by most review sites. Yes the air is thinner, but it gives up being able to accept good hard drives to do so... making the x220 the smallest you can go and still get good performance.

4. I wasn't exaggerating the value of my work. I don't even think my numbers were above average for engineer pay...)

With #2 you're being disingenuous. All computers need some work, but PCs require much more work to uninstall crapware.

As for #3, only one of the laptops you listed fits comfortably in my girlfriend's purse. As for the Air's hard drive? She has zero complaints, and I doubt her (or many others) would notice tiny differences in sequential read times (which is, I think, a measure of app-opening performance).

Okay, last reply and then I leave you guys to the fate of irrational happiness paying more for worse hardware.

re 2: It takes an hour to reinstall windows. If you swap out the hard drive (so a vertex 3 in a MBP), this is something you have to do with an apple anyway. One hour is not "much more work". Don't kid yourself.

re 3: If the MBA owner does anything CPU intensive they will likely profit by upgrading to the x220. Look at the real world numbers (http://www.anandtech.com/show/3991/apples-2010-macbook-air-1...) to get a better picture here. x220 numbers are harder to show, as no reviewers saw fit to review with a good hard drive in it -- just compare with the MBP in that review and know that the x220 with a vertex 3 is between 2 and 10x the 2010 15" MBP on those benchmarks (http://www.notebookreview.com/default.asp?newsID=6056&re...).

But really, I think you're probably right -- it sounds like a mac would be better for you

Why would you be doing something CPU intensive on a MBA like computer? Surely you just ssh into the beefy workstation in the corner or in the data centre to do the heavy lifting? Seems like a non-argument to me.

That was my point: you wouldn't on a MBA like computer. You would on a high-performance small formfactor laptop like the x220. See my other replies for more information

No, you'd do it on a Mac Pro, or, if you want to try and save money, on a PC hidden in a closet somewhere. You'll need a PC hidden in a closet somewhere as soon as you need a cluster anyway. Same for compilation, Xcode even makes it easier for you to set up a distributed build.

Now, there are reasons to get a strong laptop. A good one is if you need some serious power on the go. Are you running that bioinformatics software while on an airplane?

No, you'd use EC2 servers to avoid doing your own maintenance, and you'd do so as a last resort because GUIs are useful and remote development is laggy. http://www.useit.com/papers/responsetime.html

You're here linking benchmark charts, because as a power user you believe power users represent a large subset of computer owners. We don't. We're in the minority.

My girlfriend, and everyone else in the humongous non-power-user subset, does not care about how a Vertex 3 can bury an Air's SSD. She doesn't know the Vertex exists! If she did, she'd probably ask why she should spend $300 on it if her Air already works amazingly well.

We're talking about different groups here. You're talking about the type of people who read computer hardware blogs like Anandtech. I'm talking about everyone else in the world.

> implying that power users are the only ones who should think about performance.

1.0 second is about the limit for the user's flow of thought to stay uninterrupted. This turns out to make a big difference in productivity. See http://www.useit.com/papers/responsetime.html

I'm talking about anyone who values their time. You're talking about secretaries looking at cat pictures. You're also doing that on a thread started when a HNer said he purchases laptops via the same method.

Please:

1. Read the whole thread to ensure your "points" weren't already made elsewhere

2. Consider not replying when you don't have any information not obviously inferable from your previous post.

I'm talking about anyone who values their time. You're talking about secretaries looking at cat pictures.

Calm down there, Sparky. Are you really suggesting someone doesn't value their time because they bought an Air over an X220 with a SSD? You understand that such an assertion is ridiculous, right?

Here's a "point" for you: given the economic concept of utility, are you shocked that someone would pass on an X220 + Vertex in favor of an Air and a month of groceries? How about the idea that the marginal benefit of a Vertex 3 over an Air may be tiny enough to some that they'll pass?

We haven't even discussed failure rates, which are anecdotally bad on older Vertexes; can you blame someone for choosing an Air, along with 3 years of service at any Apple Store?

Also, another "point" for you: I sincerely doubt you would've responded like that in person, so you should probably re-read HN's commenting rules and think twice next time.

1. I accept your point that those with limited capital sometimes need to sacrifice their potential -- if you can't afford college then your annual income will be lower (though perhaps not lifetime income).

2. Diminishing marginal utility is not usually considered to apply to time -- are you saying that your girlfriend would rather use a mac than add an additional ~year of work/play time to her lifespan?

3. Note that the concept of diminishing marginal utility is different than the concept of utility.

4. I'm like this in person.

That's the 2010 Macbook Air, which did indeed have a brutally slow Core 2 Duo. It would be more useful to look at the mid-2011 model, which has a processor three generations newer (due to some sort of tedious dispute between Intel and NVidia, there were never embedded NVidia chipsets for Nehalem, so Apple waited until Intel GPUs got good enough, in Sandy Bridge; they never made a Nehalem Air).

Wait, I thought Apple hardware just works and delivers exceptional user experience without concerning the user with "2010" vs "mid-2011" or specifics of which CPU it uses and why.

Who's claiming that? For a while, the Macbook Air was in fact a particularly bad choice if you wanted high processor performance (which is a big if; computers are now fast enough that for many users it hardly matters). This is no longer the case. Simple enough.

As we're talking about the X220 here, it might be worthwhile to mention that Thinkpads do not ship with crapware. Period.

Regarding #2: plug in your previous Mac's Time Machine drive and it takes two minutes to initiate a restore. All your apps and data are restored automatically for you while you sit back and have lunch. What's that process like on Windows?

Windows also has backup solutions. If you use one of those setup time is similarly decreased. Please google first next time.

I've used Windows for decades and have yet to find a solution that is satisfactory for reinstalling apps. Documents and other data, sure that's easy on Windows and I fully agree with you there.

Rather than suggesting I google something (what, "windows backup software"? and spend several days becoming familiar with the myriad solutions?) since you seem to be so informed on similarly performing backup software as Time Machine and this is a site meant to have civil discussions, perhaps you can recommend a Windows backup solution that can restore all your apps as easily as Time Machine can. Or, for that matter, how I can network two Windows machines together and copy all the apps from one machine to the other like I can on a Mac by copying the contents of the Applications folder.

1. Not all mac apps can be copied by dragging folders out of the applications folder.

2. There are many many different backup solutions for windows -- many of which are far better than Time Machine for certain use cases (say, managing backups for several thousand machines). You probably want to look at http://www.codinghorror.com/blog/2008/01/whats-your-backup-s... for a simple solution.

Have you ever met an ordinary computer user? You're describing things an HNer might do with a new computer. Most people just leave the crapware on and never install anything. They either don't know how to remove it or don't know that it's the thing slowing their computer down.

Yeah I downvoted you. You claimed it was odd that people would buy based on X when it's clearly you that are the oddball running extremely performance critical tests that demand specific hardware. It's pretty clear what your purchase decision should be and it's also pretty clear that this doesn't apply to over 99% of everyone else in the market for an ultraportable laptop.

This comment continues in that line again and paints your absurd outlier case of 4-10x improvement as "an accurate picture of the return on investment" while again committing the error of thinking these 2 machines have the same form factor.

Me personally, I wouldn't let the cost of one hour drive my buying decision.

But Aunt Tillie sure would.

Never met Aunt Tillie, but my guess would be what's 1 hour for you is days work for her, if she'd be able to complete it at all.

Bioinformatic workloads, huh? Sounds important.

I don't believe you.

If it was major number crunching, you'd batch that crap on the fastest servers you could afford.

If it's interactive, the computer is waiting on you more than doing work.

Just admit it: you're a technophile that likes to fiddle with all the knobs. Fine. Whatever floats your boat.

But don't play stupid about the OP's article.

HP, Dell, Lenovo, etc. have TERRIBLE marketing communications supported by TERRIBLE product family line ups.

The OP's point (hello!) is Apple's simplified product family allows them to achieve better economies of scale then their competitors.

Very simple.

First of all, the thing you are replying to isn't in reply to OP's article. It's in reply to a particularly stupid comment.

Secondly, the HN crowd doesn't seem to consider that when selling to large corporations the Dell, Lenovo, etc. "overly complex" approach makes sense -- that there might be a reason that apple isn't the #1 computer manufacturer, especially for institutional purchases.

Fourthly, before talking about "mostly waiting on user input" you should consider the research. ("1.0 second is about the limit for the user's flow of thought to stay uninterrupted") start with http://www.useit.com/papers/responsetime.html

Fifthly, you seem unfamiliar with how scientific computing works. Here's the process:

1. Hypothesis (usually but not always done by professor or adviser)

2. Find a dataset, or multiple datasets that should be able to falsify #1

3. Code up an initial test in R or Mathematica (or Matlab if you don't like FP) -- does the correlation even exist? How well do the off-the-shelf methods work?

4. If you think you can do better, write it up in C++/Java, testing as you go

5. If the process requires a lot of CPU time for near perfection (NN / RBM / GA / GP), run task on server for use in your publication (graphs / discrimination percentages)

Only #5 is really effective to put on the server... Working with Mathematica remotely is a PITA with non-university hardware (licences), and remote interfaces have annoyingly high latency IMHO.

Sixthly, I do like fiddling with knobs, but to use your own turn of phrase back at you... just admit it: you didn't think before writing what you wrote. You're not analyzing and trying to understand anything in your post, only stating an opinion backed up with no facts. Like a child wandering in to a university lecture and announcing that there are no negative numbers because your teacher told you so.

That you were not downvoted to oblivion before I saw this is merely symptomatic of HNs irrational love of apple products.

Can you even do a clean install on most Windows laptops that are purchased, unless you also purchase Windows separately? Pretty much all Windows laptops I've seen since around 2003 don't come with stock Windows, but come with a 'recovery disk' which reimages the disk back to factory default - meaning, with the original crapware.

You download the CD image from digitalriver, then put that on a USB stick

See http://forums.mydigitallife.info/threads/14709-Windows-7-Dig...

That's hilarious. Is this seriously the case now?

It's just a matter of principle. If I buy a new machine (any type of machine), it should be ready for use as soon as I take it home. If I have to configure it that's fine, but things shouldn't ship broken.

I'm a typical HN reader (and a programmer) and I don't think I've changed any settings on my Macbook Pro other than the mouse tracking speed.