On the Turing Completeness of MS PowerPoint [pdf] | Hacker News Reader

On the Turing Completeness of MS PowerPoint [pdf]

andrew.cmu.edu

160 pointsgwern16 comments

The video accompanying this paper (recommend watching for a good laugh): https://www.youtube.com/watch?v=uNjxe8ShM-8

Not that the paper lacks any chucklegoofs, but that video takes the cake.

Ig Nobel Prize in Computer Science

It is now clear that Scott McNealy's banishment of PowerPoint required the creation of much less-capable systems, and thus to the demise of Sun.

Next step, emulate powerpoint within powerpoint

imagine the number of clicks it will take :D

This is literally the twentieth (!!) time in less than two months that this talk/paper has been submitted to Hacker News. The entire concept of this kind of website (Hacker News / reddit / etc.) really needs some major innovation to deal with this variability of response to content combined with repetition of material over time due to the rotation of readers (which eventually leads people to move to new communities of people with a different seed base of "stuff already seen"; a problem I call "the cohort problem" and have given talks about at times).

(There is also something weird going on where Algolia claims this specific post appeared 19 hours ago while Hacker News itself claims it appeared only 3 hours ago. What's up with that?)

As mentioned in another comment, I will argue the video is a much better way to experience this, due to a combination of the humor of the presenter and being able to watch it actually function. (After seeing it on Hacker News when it was posted six or seven weeks ago, I showed this video to the first day of a class I am teaching this quarter on the nature of programming languages.)

https://news.ycombinator.com/item?id=14016364

> This is literally the twentieth (!!) time in less than two months that this talk/paper has been submitted to Hacker News.

It may be the twentieth time the talk has come up, but it's not for the paper; I check my submissions with hn.algolia.com usually.

> (There is also something weird going on where Algolia claims this specific post appeared 19 hours ago while Hacker News itself claims it appeared only 3 hours ago. What's up with that?)

Algolia is counting my original submission last night, not the boosted resubmission by dang or whichever HN mod decided to give it a second chance.

Personally, I didn't find the video better than the paper.

I maintain "talk/paper" as the talk video links to the paper; the opposite happens to not be true, which is why it is so much more interesting to tell people "there is also a video". FWIW, if your goal is just to link to different landing pages or sub-representations of the same thing, we could also submit the PowerPoint file itself. (I will also say that I guessed it was moderators that made the bump but purposely didn't make that explicit. That to me just underscores the need for a more fundamental fix to the way that content appears, as it makes clear that the twentieth time posted this was also just going to be do poorly again without manual intervention.)

I mean... for what it's worth, I'd never seen it before, I would never have seen it without the manual intervention, and I loved it and think it's perfect for HN. I don't think we really need a 'fundamental fix' for anything, seems to be Working As Intended (tm) ;)

> That to me just underscores the need for a more fundamental fix to the way that content appears, as it makes clear that the twentieth time posted this was also just going to be do poorly again without manual intervention.

I agree, there's far too much randomness to HN. The two-tier structure is part of the problem; even with mod boosts, it's frequently the case for a link to fall off /newest with no upvotes, while on another occasion, it'll get +100. I see this all the time when checking my submissions: a link will be submitted 5 times previously over many years, and only 1 will do well - obviously, the link quality hasn't changed, yet the ratings are contradictory.

You sound like a ribosome in a follicle complaining "More keratin?! Really?! Can't we make something else?"

That's so Microsoft. Historically, Microsoft systems tended to execute anything that came within range of an execution engine. This formed the basis of many, many exploits. Gradually, they tightened up the defaults for AutoRun, Excel spreadsheets, and Word documents. Then came PowerShell and a whole new generation of exploits.

I think that's not what this particular case is about, see the video for a demonstration of what the Turing machine looks like here. In particular, the Turing machine progresses one state at a time, advanced by user clicks (!). It uses AutoShapes and On Click Animations, the number of which depends on the number of states and the number of tape cells.

Writing an exploit in that would be a real challenge.

Shh.

Reply on news.ycombinator.com