Remix.run Logo
wat10000 2 days ago

I think this is an argument for the log scale. I'd argue that the things you say it communicates are not actually correct.

Adobe's size has been growing exponentially pretty much this whole time. The rate increased slightly in the mid-2010s. SumatraPDF started out that way too, but managed to level out after about a decade.

Relative size is what matters here. That increase from ~2.5MB to ~5MB in the mid-90s was pretty significant for the time. In terms of the impact on users, it's probably at least as important if not more so than going from 300MB to 600MB 25-30 years later.

qualeed 2 days ago | parent | next [-]

>Relative size is what matters here.

This is where our disconnect is. Relative change in size means nothing to me. I care about the absolute size of the final thing I'm installing.

Adobe big, getting bigger. Sumatra small, staying small.

conductr 2 days ago | parent | prev [-]

I disagree, am with qualeed on this one. I don’t think the size doubling means much at all except raising the question of why did it double? What was added that I care about? My instinct tells me nothing so it’s shouldn’t really be acceptable except this is par for the course these days. Nobody cares about bandwidth it’s just assumed to be fast and unlimited by nearly every publisher of software.

In the 90s that jump cost me in terms of modem time. I couldn’t download anything else for an extra 30-60 minutes that day (if I remember my speeds correctly). Today, extra 300mb costs me less than a minute and I can easily continue multitasking in the process.

wat10000 2 days ago | parent [-]

Imagine there had been a 50MB jump in 1998. That would be a major WTF moment. Now imagine a 50MB jump in 2025. We'd barely notice.

Saying Adobe's crazy growth spike started six years ago is just pointing to the knee in the exponential curve. It's had pretty much the same curve since version 1.0. And SumatraPDF had the same exponential growth for quite a while.

If absolute numbers are what matters and an extra 300MB is not important, then why not scale the Y axis to 1TB and squash everything to the bottom?