Skip to Main Content
Contribute Try STAT+ Today

When he first saw the email, just after finishing rounds in the surgical intensive care unit, Dr. Joseph Schlesinger began to panic. A government staffer was writing to inform him that he’d made “a fatal error.”

Standing at the nurses’ pod, surrounded by patients in various states of post-surgical woe, he wondered what he should do next. Life-sustaining machines whirred around him, residents discussed prognoses — but he heard nothing. “It was like I was in a vacuum,” said Schlesinger, an assistant professor of anesthesiology and critical care medicine at Vanderbilt University and a staff physician at the Veterans Affairs medical center in Nashville. None of the cases in front of him needed his immediate attention, and so he left to address the crisis in his inbox.

His “fatal error” was not medical but typographical. A week before, he had submitted an 84-page grant application to the Department of Veterans Affairs. It had taken him months to write, late nights of work before early mornings in the operating room. Now, it turned out, all of that toil had been wasted — because of a problem with fonts.

advertisement

“Unfortunately, this application used a non-approved font typeface and point size in the Research Plan, which is a fatal error,” the VA staffer wrote on Sept. 19, “and therefore has been withdrawn from review.”

But Schlesinger is certain he never used the prohibited lettering, and he’s delved into the arcana of graphic design to prove it. What he found was evidence of a technological glitch — and he is incensed that something so piddling should stall his own research and that of others.

advertisement

VA officials retort that rules are rules, and that their 130 pages of guidelines, intended to promote fairness, include a warning about this computer-generated gaffe. The agency’s Rehabilitation Research and Development Service even has ruler-wielding enforcers, who’ve unmasked and thrown out eight poorly formatted submissions of the 346 they’ve received this year.

The issue has arisen before. In 2015, in response to a paleontologist’s post-rejection rage, the United Kingdom’s Natural Environment Research Council released a statement saying it tossed out “just” 4 percent of submissions based on font. Last year, after colleges lost millions in federal grants over typeface mistakes, members of Congress stepped in, and Education Secretary Betsy DeVos forbade withholding funds over such trivialities.

Grants are the lifeblood of science. Researchers are as much grant-getters as they are experiment-designers, the way truckers can’t drive without thinking about fuel. To those experienced in requesting cash from both the VA and the National Institutes of Health, rejection on the basis of margin or serif, if not the most pressing problem in biomedical research, is unsurprising and frustrating.

“Happens all the time,” said Jake Seliger, principal at Seliger + Associates Grant Writing Services, of applications thrown out for reasons of formatting. “This is part of the reason we have a business, because part of grant-writing is following every instruction, no matter how absurd-seeming it may be.”

“We have had our share of these glitches here at VA Boston,” Terence Keane, associate chief of staff for research and development at VA Boston Healthcare System, wrote in an email. “No one is ever happy when they make a silly error.”

At first, Schlesinger thought he had in fact made an error, somehow both silly and fatal at once. He rushed to his office from the ICU to check. He opened his word processor, clicked down his Research Plan line by line. It was all 11-point Helvetica. Like an ecologist sampling for forest diversity, he plunked his cursor down at different points in every paragraph, highlighting here and there. Still, all he found was 11-point Helvetica. He even printed it out, and used a ruler to measure the number of characters per horizontal inch and the spacing of his lines. By the VA’s rules, he said, everything should have been fine.

In the hundreds of millions of dollars allocated to VA research every year, the $200,000 he was asking for seem inconsequential — but that amount is what would have allowed him to study collaborative songwriting as a way of improving veterans’ PTSD symptoms. He had already submitted the proposal once, after five months of preparation, and received encouraging comments from reviewers. This time, he’d taken that feedback into account and submitted it again. “When it gets rejected at the administrative level, that’s where it ends,” he said. “They don’t even send it out to the reviewers. The reviewers may not even know I sent it in for resubmission.”

Now, he raced up three flights of stairs to see Russ Beebe, who had tools more powerful than a ruler and the naked eye to figure out what had gone wrong. Beebe had spent decades working in graphic design — at newspapers, ad agencies, print shops — and he helps Vanderbilt researchers format the images in their grant applications.

Graphic designers, before sending out a document, will often do what’s known as a preflight analysis — a kind of computerized quality control that searches out issues in the thickness of hairlines and the bleeding of images — and that is what Beebe did post-hoc for Schlesinger. The problem popped up immediately. The government website requires submissions in PDF format, and in the conversion from Microsoft Word, some bits had been automatically modulated from Helvetica to Arial, and downsized by a few decimal points. That was why, as VA staffers had noted, certain sections had exceeded “the 15 characters per horizontal inch and 6 lines per vertical inch allowed.”

How, you might wonder, did the agency pick up on such a minute change? Patricia Dorn, the VA director of Rehabilitation R&D, who withdrew Schlesinger’s application, told STAT that each intake officer has a keen eye for any suspicious-looking margins or uncanny fonts.

“They convert it back into Word, and … they see the font type and the font size, and then they print the page, which is what we asked people to do,” she said. “They literally get a ruler out and do character counts and they do line counts.”

“We have these rules and guidelines and very specific parameters that I feel I have to uphold for everybody in the system. I’m not saying that because I’m some crazy rule freak. It’s a competitive process.”

Patricia Dorn, the VA director of Rehabilitation R&D

Schlesinger said he did, in fact, print off a page from each section of his PDF, as the instructions suggest, but the differences in font were indistinguishable to his eye. Dorn insisted that any researcher who doesn’t have her staff’s acuity for marginal irregularities should ask her office for help.

“It’s tragic when an application doesn’t go to panel, I’m not going to say it isn’t, but I’m also going to say we have these rules and guidelines and very specific parameters that I feel I have to uphold for everybody in the system,” said Dorn. “I’m not saying that because I’m some crazy rule freak. It’s a competitive process.”

She emphasized that her typeface decisions are about fairness, so that no one researcher gets more space than any other. Because of the way those rules are enforced, sometimes an accidental smaller font size gets penalized even if a researcher’s application may not actually contain more material than it would have with the right lettering.

Whether you find it reassuring or dismaying that there are government officers measuring letters and lines with rulers depends on your point of view. For Schlesinger, the drama over his “fatal error” was deeply upsetting — though he plans to resubmit at the next deadline, with some preflight help from Beebe.

“When you have physicians leave the VA, it’s because of these types of issues thematically. The bureaucratic issues burn us out at the VA,” said Schlesinger. “I’ve decided to continue to work at the VA, but I did talk to my boss about cutting down on my time at the VA out of sheer frustration. Ultimately, it hurts the vets, and it’s not their fault, they deserve excellent health care. We just want to provide that care, we want to provide great science and discovery, and issues like this stall it.”

  • Discarding and application over a trivial deviation in typography was deemed more important, in the world of scientific research, than evaluating the concepts and ideas presented in the application. That is simply, completely and utterly wrong. Worse, there are commenters here who support this behavior, with no thought given whatever to whether the proposed research might have led to a clinically signficant breakthrough. Uh-huh! Nope. Printed in Univers instead of in Helvetica Neue. 12 vs. 11 point. In the shredder. If it’s so damned important, return it to the sender and ask for a resubmission. How is the greater good served by such mindless, nitpicking rigidity?

    • Seth, this is not about crushing great science over an inadvertent type font. This is about an action that was very likely intentional as a means to gain unfair advantage in peer review. All universities have people that check these things, but somehow this slipped through the cracks? Possible, but not likely.

      Except for one-shot requests for proposals, any rejected application for an NIH, NSF, VA, etc. grant may be revised (fixed) and resubmitted. So, not to worry, medical science will not be deprived of this proposed breakthrough study.

    • It seems to me that the requirement, fifteen characters per inch, demands the use of a monospaced font.

  • Would you believe that some years back, the NIH did the same thing to a blind researcher. It was either wrong font or wrong font size. The researcher was blind; I feel the need to repeat.

    • Deb, you are obviously not a researcher (principal investigator: PI) by profession. Except for fellowships, universities and institutes apply for grants on behalf of the PI who actually will head the study. All grant applicant organizations have professional bureaucracies that ensure 100% compliance with requirements. Usually, a dean-equivalent submits applications for the PIs – even for blind PIs. Some schools authorize some of their senior PIs to submit (never a good idea). The font lapse is either an inexcusable institutional failure or intentional – blind or sighted.

  • As an IT person with an interest in science, this issue strikes me as asinine. According to the article, the original submission, created in Word, met the requirements. The process of converting it to PDF and then back into Word caused the issue, not the person submitting the request; how is that fair? There are so many ways this issue could be solved (mostly very simple) that do not require multiple file conversions, wasting paper/toner/electricity to physically print out electronic files, and then manually inspecting them with rulers. This incompetent waste of time and resources is a 1995 solution to a very simple problem.

    • Dude, totally agree! Perhaps a cloud-based form with an automated word count “doo-hicky” written in LOLCODE. Then any web reader can set their own browser font for max readibility (fyi – my preferred browser font is 38 point FF Matinee Gothic)

  • Submitted to NIH last April, and my project is in the current cycle. It’s so sad to stress good people (submitters AND reviewers) out. A solution is to provide a template – a fillable form with character count limits and allocate file space for uploading tables and images. This is a process used by many venture funds and foundations. Cloud based software can go a long way to help.

    • Your proposal sounds easy, but that is what Grants.gov thought it was achieving during Obama’s 1st term. When that system fell short of expectations, HHS created ASSIST to coordinate all the evolving forms NIH tosses at applicants. It remains to be seen if ASSIST fixes anything. The problem seems to be legacy systems in the 430 departments, agencies, and sub-agencies in the federal government. It is a virtual tower of babble.

      PERSPECTIVE: Prior to the very late 90’s, all NIH applications were submitted on paper PHS-398 forms with 3 -12 copies depending on announcement. Around submission deadline time, UPS and FEDEX trucks cued up for blocks around the Center for Scientific Review to offload tons (no exaggeration) of 398s. The hallways of all institute offices were barely passable for the floor-to-ceiling file cabinets filled with applications past, present, and pending.
      Then came Adobe Acrobat, and NIH very quickly went digital. Around the turn of the century, the pdf scan blitz started, and file-drawer by file-drawer, all those 398s and associated pink sheets (reviews) went onto servers. The age of digital applications was well in swing by 2002 or so but the new one-application-fits-all 424 form could not support NIH grant applications so they just tacked on the PHS398 form and then another and another as HHS developed “improvements.”

      Ideally, HHS should create a new application form. However, if one looks at the pace of the 2008 Congressionally-mandated common DOD/VA electronic health record system; 10 years later and billions spent – it remains vaporware. With president after president more focused on recessions and wars – creating new IT systems is likely to remain something IT offices putz with using unspent agency budget money.

  • Not understanding or following the rules is a perfectly good reason to reject because it indicates a lack of diligence on the part of the author. I hope they reject for spelling errors too. There’s no excuse for being lazy about these things.

  • This story sounds like an outrageous abuse of bureaucratic nit-picking. IT IS NOT. Font rules in Grants.gov ensure an even playing field in peer review technical competition. Thus, all applicants must make their case on forms that limit space for information. Limited verbosity makes peer review uniform and a bit easier.

    A an unauthorized font can enable scrunching letters with the same point size into a smaller space, thereby enabling applicants to include more details and arguments than competitors who play by the rules. Thus, what seems like a nitpick is merely a way to prevent cheating – whether inadvertent or intentional.

    This rule achieves several aims. First, it ensures equal readabilty to human reviewer eyes, which could vary simply due to font choice (Using character count limits alone does not acheive that end). Second, it enables the use of computer screening, so the government is not paying people to be font police. Note: when an application is rejected for font use, normally a human does eyeball the application before initiating a rejection notice.

    • Thomas, you might want to check your own reply, before getting on your high-horse. What is the beginning of the second paragraph supposed to be?

    • “First, it ensures equal readabilty to human reviewer eyes, which could vary simply due to font choice. (Using character count limits alone does not acheive that end).”

      The neat thing about computers is that the reader can choose the font that’s most readable to them. Changing the font face or the font size does not affect the character count. You would still have X number of characters to make your case, regardless of whether you’re using 25pt Helvetica or 10pt Open Sans.

      “Second, it enables the use of computer screening, so the government is not paying people to be font police. Note: when an application is rejected for font use, normally a human does eyeball the application before initiating a rejection notice.”

      A couple of things:

      1. The article states that *people* were the reason the person’s submission was rejected. It flat-out says that their people “have a good eye” for this kind of thing. They are, in fact, literally paying people to be font police.

      2. Computers don’t care what font size or face you use. It’s all just UTF or ASCII values to them. Font and size are literally just window dressing. Enforcing a font face and/or size in a digital context does nothing for screening except enforcing a rule that’s archaic to begin with.

      3. Using PDF format (and/or having to convert between PDF and Word) is arguably *the* worst format to ask for if you want computer enforcement. Not only does the conversion try to be “helpful” (as illustrated in the article), but both PDF and Word doc formats are proprietary and quite fragile in how they do things. Something as simple as a difference in software version can cause issues like this and worse, and third party programs are even more finicky because things are so opaque.

      The best way for all parties involved would be to have submissions done in one of the many plain text formats, enforced by a character count. If they really want to keep the archaic measurements, then they can set the reader to whatever font and size they want and measure accordingly.

  • The obvious solution here is to have a character limit, rather than squabbling over the physical presentation of the characters. This would still be fair to all participants, and avoid people wasting countless hours on such minutiae.

  • It is ridiculous that the government is paying employees to look for for font size differences in grant proposals that can’t be seen with the naked eye. Total waste of VA funds that it needs to put elsewhere. If they are worried about fairness and making sure that every submission gets the same amount of “space” to make their case for a grant, just enforce a word count limit.

    • Actually, these rules came into effect with stricter page limits and the computer (where you can change fonts and spacing so easily). I agree with Thomas, this sounds like nit picking but is to ensure fairness and to help with peer review. The reviewers have to able to read the grant. Why would a “word count” be any more fair? The applicant should be mad at his institution, which didn’t check the grant before it went out. Or after all his hard work, why didn’t he check the final copy prior to submission?

    • Monica, a character count would be more fair, because that’s ultimately what the combination of characters per inch, lines per inch, margins, font, font sizes, and total number of pages is limiting. With those rules, you can probably distill it all down to a single X number of characters and all of the allowed fonts will probably fall into a very small range of each other. The problem with the rule is that it’s based on the idea that things like font size and face can’t be changed on the reviewer’s end (which may have been the case during the time that they could be written on the computer, but had to be submitted on paper). That doesn’t have to be the case anymore. Submit in any one of the plain text based formats, and reviewers can use whatever font, size, and even color combination that works *for them* for readability, and you avoid stupid problems like this altogether.

Comments are closed.