I did an experiment like this in the mid-90s as an undergraduate for a research methods lab assignment.

I showed true and made-up (NPR Wait Wait style) articles to elderly people (over 70, in a care home), printed in one of 3 formats - on newsprint in the font the NY Times uses, on glossy paper in a somewhat 'friendly' font (like comic sans, but less so), and finally in dot matrix, on green-and-white fanfold, all upper case.

There were remarkably few differences in the first two, for either true or false stories, but NEARLY EVERYONE (like, > 85%) rated the computer fanfold printed stories as being true.

One person's reaction has stuck with me to this day: "If it weren't true, the computer would have rejected it!"

The customer account statements for Bernard L Madoff Investment Securities were notoriously old fashioned computer printouts. [1][2] There was some coverage with a former client, I can't find now, where he even talked about the appearance of the statements encouraging trust.

I wonder if Bernie Madoff researched it, or if it just came with a con man's intuition.

[1] http://www.pbs.org/wgbh/pages/frontline/madoff/financial/

[2] http://dealbook.nytimes.com/2008/12/15/a-look-at-madoffs-tra...

> One person's reaction has stuck with me to this day: "If it weren't true, the computer would have rejected it!"

Must have been a Prolog programmer!

over 70 years of age or over 70 sample size?

Age. N was about 20.

I'm gonna go with age because of the qualifier "in a care home"

Maybe his thought process was that giving the age twice ("elderly" and then "70" -- do you need both?) would be redundant, so that the redundant number might not be an age.

Indeed. I also consider 70 to be pretty young for a retirement home. Statistically speaking, I assume the vast majority of 70 year olds (95+%) are living independently. So I thought either OP is being redundant and low-balling or giving a sample size. Didn't know.

Btw, through a quick web-search, it appears the average for new residents is mid-80s.

Blogspam for a 2012 article: http://mobile.nytimes.com/blogs/opinionator/2012/08/08/hear-...

Part 2: http://mobile.nytimes.com/blogs/opinionator/2012/08/09/hear-...

>Blogspam

This word really isn't appropriate. (Similar to my thoughts here - https://news.ycombinator.com/item?id=9572254 ).

The article here is interesting, approachable and good. Of course, your links add to it, it's the source mentioned in the first paragraph. But we can still read interesting secondary sources.

It doesn't add anything significant that's not in the original article. Worse, unlike most blogspam, it doesn't even directly link to the original article.

And, bafflingly, it doesn't even give examples of all the fonts.

All right. Though I thought the mention was pretty explicit, and a good summary of the research there. I also thought it's a super-quick read, and I learned something.

More than 20 years ago, at the company where I was working, I needed to order some sort of expensive thing. Printing a purchase order on the dot matrix printer connected to our accounting computer was a hassle, so I duplicated the form in whatever word processing program we were using on the Apple Mac. It even said Purchase Order at the top.

I faxed it to the vendor, and received a reply: "We will ship the item when we receive a copy of the actual purchase order."

I did nothing but re-print the order in Courier font, and faxed it to them. It was just the ticket. The boss said, "Now you know why we keep that old dot matrix printer on the accounting computer."

I liked the article, but I did not like the eye in the header blinking at me as I read it.

Same here, luckily I have a very handy bookmarklet (that I probably found here at some point) to delete any DOM element on the page:

    javascript:(function(){document.styleSheets[0].addRule(".highlighted_to_remove","background:red !important");var e=function(e){if(e.keyCode==27){i()}};document.addEventListener("keydown",e);var t=function(e){e.stopPropagation();this.classList.add("highlighted_to_remove");return false};var n=function(e){e.stopPropagation();this.classList.remove("highlighted_to_remove");return false};var r=function(e){this.parentNode.removeChild(this);i();e.preventDefault();e.stopPropagation();return false};var i=function(){var i=0;var s=document;while(s=document.body.getElementsByTagName("*").item(i++)){s.removeEventListener("mouseover",t);s.removeEventListener("mouseout",n);s.removeEventListener("click",r);s.classList.remove("highlighted_to_remove")}document.removeEventListener("keydown",e)};var s=0;var o=document;while(o=document.body.getElementsByTagName("*").item(s++)){o.addEventListener("mouseover",t);o.addEventListener("mouseout",n);o.addEventListener("click",r)}})()

Another bookmarklet that does the same thing, but a bit more polished:

    javascript:(function(){var e=document.body.style.cursor;document.body.style.cursor="crosshair";var t=document.createElement("div");var n="border:1px solid #3280FF;background-color:rgba(50,128,255,0.5);position:absolute;z-index:2147483647;display:none;";var r="pointer-events:none;";var i="transition:width 60ms,height 60ms,left 60ms,top 60ms;";n+=r+"-webkit-"+r+"-moz-"+r;n+=i+"-webkit-"+i+"-moz-"+i;t.setAttribute("style",n);document.body.appendChild(t);var s=null;var o=function(e){var n=e.target;if(n!==s&&n.parentNode){var r=n.getBoundingClientRect();var i=document.documentElement;var o=document.body;var u=i.clientTop||o.clientTop||0;var a=i.clientLeft||o.clientLeft||0;var f=window.pageYOffset||i.scrollTop||o.scrollTop;var l=window.pageXOffset||i.scrollLeft||o.scrollLeft;var c=l-a+r.left-1;var h=f-u+r.top-1;t.style.display="block";t.style.left=c+"px";t.style.top=h+"px";t.style.width=r.width+"px";t.style.height=r.height+"px";s=n}};var u=function(n){document.body.style.cursor=e;if(n.target.parentNode)n.target.parentNode.removeChild(n.target);if(t.parentNode)t.parentNode.removeChild(t);window.removeEventListener("click",u,false);window.removeEventListener("mouseover",o,false);n.stopPropagation();n.preventDefault()};window.addEventListener("mouseover",o,false);window.addEventListener("click",u,false)})();void(0)

Alternatively, right click the element and choose inspect element, then add "display: none" to its css rules.

Even easier, right click, inspect element and hit the delete key :)

If you disable Javascript you're left with a blank page with just the blinking eye in the middle.

It's called graceful degradation: figure out what parts of the user experience are so important they have to work even without JS enabled, and make sure they do. They clearly knew what their priorities were.

Right click->inspect element->delete node

Did this too. That eye is doing a disservice to that site - if wasn't a web developer who knows how to remove that, I would not have finished reading the article and left that page. Annoying things chase people away, guarantee that site gets less return visitors because of it.

IIRC there were studies that images of eyes force people to answer questions more honestly. I wonder which effect is stronger.

hahaha same here.. a bit annoying

“It’s absurd to think that we would be nudged by one typeface over another, into believing something to be true. Something disturbing about it, I’d go so far to say.”

It's really not disturbing.

It's obvious that the way something is presented can affect your judgement about things like the motives (or credibility or even identity) of the person saying it, and it's obvious that this will in turn affect whether you believe what you're being told. Things like tone of voice, choice of words, and typeface all have an effect.

Why does it surprise people that the manner in which you communicate can affect how likely people are to believe what you're saying?

Yeah, what's with these assumptive statements?

Now, Morris is the first to admit that this wasn’t exactly a true scientific experiment, but you know what? That’s okay! Because the findings are still fascinating.

Experiment validity and the interestingness of the results are orthogonal. Considering the topic, I'd hypothesize that the less valid the experiment is, the more the experimenter is motivated to make the results look interesting.

It's not surprising, but it is disturbing. Because this result holds independent of the content being communicated. A good presentation doesn't just affect your evaluation of opinions, ethical arguments, or he-said-she-said judgements. It affects your evaluation of proofs, engineering decisions, and concrete physical evidence. Decisions that are formal enough, in the mathematical sense of the word, that we should be relegating them to computers if at all possible. That's disturbing.

Typeface, clothes, voice tone, social status... the list goes on.

That's exactly what the article says further down.

Not really. It says the findings are fascinating/interesting, and I'm saying they are not.

Reminds me of the Higgs boson announcement, which was surprisingly delivered in comic sans:

http://www.theverge.com/2012/7/4/3136652/cern-scientists-com...

I suppose the context kept the font choice from detracting too much in that case!

That blinking eye took away whatever effect the article's font had on me.

I put up another window to block the eye so I could read in peace. I don't what they were thinking.

"Now, Morris is the first to admit that this wasn’t exactly a true scientific experiment, but you know what? That’s okay!"

No its not ok because it means that the findings, however fascinating, can't be accepted.

What if... this whole article is an experiment in itself in that it is completely false, but has used its fonts and the blinking eye to convince us otherwise.

When I see documents nicely type set with LaTeX, I have a prior expectation that it is more likely true.

Presumably we believe people wouldn't invest time and energy going to great lengths to present something that isn't true in a well-designed and complex way. Which is stupid really, because the effort someone will put in is almost always related to the reward they get out rather than anything to do with truth - if presenting something false has a high return people will go to great lengths to present it extremely well.

I was just thinking somewhat the same thing, making my son familiar with mono-spaced fonts, so when he found some source code, it'll be like home to him when he grows up...

guess it's going to work, to some extent

p/s: he's just 4yo now

That's really neat that you want your son to be familiar with monospaced fonts so source code displayed that way will look normal and comfortable. Screenplays too!

So I hope I won't be out of line with a friendly suggestion... :-)

Don't teach him that there is something special about monospaced fonts that lets you write code - that code can only be written in a monospaced font. I honestly think that the majority of programmers believe this, but it just isn't so.

I haven't written any of my code in a monospaced font in over ten years - and the funny thing is you'd never be able to tell that from reading my code. It reads the same and the formatting looks the same whether you view it in a proportional or monospaced font.

I've run into many programmers who are shocked when they find out that I write code in a proportional font. They ask, "How can you do that? How can it possibly work?"

Rather than go into the details here of how it can possibly work - it's really quite simple and becomes apparent once you try a proportional font - imagine that one day your son asks, "Daddy, why do we have to code in monospaced fonts? I like Trebuchet! [Or Comic Sans, or whatever.] Why can't I code in that?"

Do you answer, "Don't be silly, of course you can't do that. We always program in monospaced fonts. You couldn't possibly code in a proportional font, it just doesn't work."

Or, "Well, that's a good question. I've always used monospaced fonts, but I never really thought about it. Why don't you experiment and see how it works out, maybe you can teach me something new!"

Don't misunderstand, these are just examples, I don't mean to put words in your mouth or say what kind of conversation you and your son might have. But I think most programmers today would give the first answer, and that makes me a bit sad.

Whatever your own thoughts about fonts, it is definitely cool that you want to open the doors for your son to be interested in programming!

..and then, by the time he grows up, elastic tabstops become commonplace and people stop using monospaced fonts for code :)

Yeah I think this is true: most people would probably (initially) trust a paper using Computer Modern more than one using Times New Roman.

I think the takeaway from this is that people associate "odd" fonts with different things (but rarely professionalism), which makes them scary to use in a professional context. Similar to how most of us are "afraid" of wasps because we've had previous bad experiences with them, most people have experienced out of the normal fonts in the wild, but most often in a joke, or on a bad internet site. This previous experience, like the wasp, is taken with us subliminally to the next time we see the font.

Standard fonts exist because if you use them everywhere, users are forced to evaluate the content, and not the font.

Speaking of bad experiences though, I tend to avoid Betteridge's[0] too.

[0] http://en.wikipedia.org/wiki/Betteridge%27s_law_of_headlines

But this seems to be one of the few exceptions to that law. The headline was a question, and the answer was yes. Well, almost, sorta, kinda.

In today's issue: Does Betteridge's law of headlines have exceptions?

Did anyone check what font the NYT website used at the time? I'd expect the font most similar to that one to win the survey.

Man, I just get angry with people this sloppy. How can you write:

"Now, Morris is the first to admit that this wasn’t exactly a true scientific experiment, but you know what? That’s okay! Because the findings are still fascinating."

without dying of embarrassment? It is equivalent to:

"Hey, this next bit might not be true but that doesn't matter because it's a good story!"

Its just offensive to so blatantly mention the unreliability of the findings and then go on to dismiss the notion of uncertainty entirely in the remaining text. Why did he even put in that mention?!

Just because it's not a true scientific experiment, that does not necessarily mean that it's uncertain.

That said, I'd rather people say this, than that they don't mention any uncertainty at all.

Simply saying "this is uncertain information" does not absolve the use of "truth" and "prove" without qualifiers as the text does.

Except for that mention the text does not act like there are any uncertainty, hence the mention is a simple and dishonest cover-your-ass.

Comic Sans and/or XKCD-style charts[1][2] certainly are useful if you want to convey the back-of-the-envelope nature of your findings.

[1] https://www.chrisstucchio.com/blog/2014/why_xkcd_style_graph... [2] http://dan.iel.fm/xkcd/

Rather than any intrinsic property of the font, I'd ascribe it to the associations we have to it. Has previous things we read in this font been trustworthy?

Are there cross-cultural studies of this effect? If you are wrong, trustworthy fonts will have the same set of properties in cultures with similar alphabets. Pair of cultures which rate trustworthiness of fonts differently despite sharing the same language and alphabet could prove you.

I believe there is too much homogeneity among cultures using the same fonts (i.e. latin alphabet) that there isn't going to be such a case.

For example, hand-writing-like fonts are used in comics both in US style, european and south american stuff; newspapers and books have been printed with serif fonts since about forever; international brands look the same; MS Office/windows has provide everyone with the same fonts for decades.

Baskerville has gorgeous ampersands: http://www.marsie.com/wp-content/uploads/2009/05/marsie-ampe...

It is Comic Sans that can make us believe something is untrue.

I wrote a somewhat similar article about an year ago: http://www.adpushup.com/blog/how-typography-affects-readers/

Quoting from the article: An experiment conducted by Kevin Larson and Rosalind Picard looked at how typography affects mood and performance of the readers. It happened across two studies, where they took three kind of measurements: Relative Subject Duration (RSD), Likert Scale and through a cognitive task which was different for both studies.

There were twenty participants, half of them got the text with poor typography and half got good typography.

In the first study, the participants were given twenty minutes to read the passage, amid which (after 15 minutes) they were interrupted and asked to estimate how long they had been reading for. After the RSD data was recorded, they were asked to fill out the Likert preference questionnaire with 6 statements, where one of the statements was whether they found the text easy to read.

Next, they were given a cognitive task, which for the first study was a Candle problem.

The second study was very similar to the first one, except for the RSD data, they were interrupted 17 minutes into reading, and the cognitive performance task was changed to remote associates test.

Results - Participants with good typography underestimated their time by 3 minutes 18 seconds in the first study and by 5 minutes and 21 seconds in the second study. This shows that when a passage is well structured, the readers are more engrossed in the text. - With the Likert scale questionnaire in the first study, the results were not reliable. But in the second study, some scores were in favor of good typography. - For the Candle problem, out of the 10 participants who were shown good typography, 4 successfully completed the task, while 0 participants completed the task in the other group. - In the Remote associates task, the results were not statistically significant or reliable with the good typography group succeeding on 52% of the task at an average speed of 6395ms, and the poor typography group succeeding at 48% of the task at an average speed of 6715ms.

Even though the result was not reliable enough in the second study, the good typography group did perform better. This can be attributed to the fact that they were in a better mood before starting the task. Hence, they were better at problem solving.

It would be fascinating if they could release the anonymized raw data though. I'm curious about the vague statement "people are much more likely to believe something written in Baskerville than any of the other fonts." For example, I want to know the sample size and the significance level at which the conclusion is obtained.

"Quidquid latine dictum, altum videtur"

I'm heavily biased toward anything published as Latex PhD template instance.

This is new -- a blog where you can't read below the fold by any known method. Arrow keys? Nope. PageDn? Nope. Scroll wheel? Nope. Scroll bar? Disabled.

[Chrome 43, Win7]

There was a Letter of Note, some famous director talked about setting documents in Georgia because it made them look more professional.

I tend to assume any blog about design/typography with article text in a font larger than 15-20px is written by a clueless newbie.

In short: quidquid latine dictum sit, altum videtur. Serious-looking fonts impart more trust into content.

I am mostly certain that I am less trusting of this website because of its font choice.

Well played. And hilarious.